With managing application performance as a top priority for its Oracle E-Business Suite ERP deployment, the economy has forced this business to eliminate budget increases across the board.  So as data accumulates and transaction volumes increase, this Enterprise Architect is challenged to maintain performance levels.  Initial tests and ROI calculations have confirmed that Oracle Exadata will provide the scalable performance necessary to accommodate the business’s needs.  However, the budget will only allow for one Oracle Exadata appliance deployed.

This team creates a copy of their production application for test, development, and staging as well as another copy for disaster recovery purposes.  They currently leverage storage-based replication to make the copies–which in some cases are “virtual copies”–meaning they are only pointer-based copies that require a small fraction of the total amount of storage that their current production database requires.
The expected benefits of the Oracle Exadata platform include improved performance, significant columnar compression for some key tables in their application, and simplified management.  However, when the same database is copied to a non-Exadata platform for test and development, the organization would lose these benefits.  The amount of effort required to change their currently efficient non-production copy processes starts to mitigate the benefits that Exadata promises to bring.

With new developments in solid-state drive storage technology, this organization is looking at leveraging Oracle ASM, advanced partitioning, and an ILM strategy to address their performance and cost constraints rather than moving to Oracle Exadata.  This management team is looking at their current storage maintenance renewal schedule and will conduct another cost-benefit analysis then.  In the meantime, they will continue to look longingly at that Database Machine (akin to a puppy in the pet store window) and find more cost-effective alternative approaches to find improved application performance and cost reduction opportunities.

But back to Oscar W….when you can pay literally just a few cents per GB for a SATA drive off of Amazon or from your local big box store, how come you’ll have to fork over a handful of dollars per GB (all in) when something-pretty-similar is packaged up as mid-range modular storage….or maybe get a PO raised at around $15 per GB (all in) for high-end enterprise disk systems…..or even something north of that by 3 times (or more) for an SSD appliance? I’m guessing the quote gave it away, eh? It’s all about the value. What features and functions are there? What performance and protection? What’s the service and support? What does it do for you, as opposed to what it is.

And that’s the essence of storage tiering–putting the right data on the optimum storage device according to its needs at any given time. You wouldn’t expect 1000 square feet of living space to cost the same in rural Africa, suburban Perth, central London, or the International Space Station. And you wouldn’t expect the same quality or quantity of food for–say–$10 in Dairy Queen, a Vegas buffet, and a 3 Star Michelin restaurant. Value isn’t defined by one statistic alone, whether it’s housing, food…or data storage. So ’dollar per gig’ is certainly not everything. Value–usefulness–is. Or at least it should be–but we have decades of indoctrination to the contrary. Do you think we can really get away from thinking just about the raw cost of storage and moving to tiers of business value? What an obvious–but still often ignored–idea! We should hope we can get around to accepting the notion: after all, as dear old Oscar Wilde himself also wrote ”An idea that is not dangerous is unworthy of being called an idea at all.”

You can read Mark’s other blog entries at The Business of Storage.

There’s been a slew of announcements over recent weeks and I’m trying to raise my own viewpoint to look at such things in light of how much they can contribute to this growing need for [financial] revolution rather than just [product] evolution. I don’t expect silver bullets, but at least it’d be good to see revolutionary zeal! And–before I get the angry e-mails from product managers anxious to prove the worth of particular programs and enhancements–I am not saying that all the evolutionary stuff isn’t useful and wonderful and isn’t going to give users value, keep them in a job, and their employers profitable for a while … it’s just that it isn’t enough for the long haul.

Most of the big systems vendors and largest storage companies have had announcements recently; many are showing an acknowledgment that change is a-coming. Even though they naturally disagree on the exact method, there is agreement that the “big IT picture” will look different very soon. Please don’t read anything into the running order here, it’s random. HP is all about driving towards a Converged Infrastructure, with virtual pools of resources that get flexibly applied to IT issues. NetApp, in essence, also similarly sees a single large pool of storage, although obviously in its version the broader compute abilities are via partnerships rather than an in-house approach. HDS has been talking up the cloud and convergence and is significantly moving into a broader IT space by packaging its own servers. EMC‘s contribution stresses the cloud and co-operative packages, such as VCE, while also shouting loud about the growing/coming importance of solid state. Dell is looking to bring its streamlined simplicity to the enterprise as well as to smaller businesses. Meanwhile, IBM quietly continues to do–and to be invested in–just about everything! And from [what still feels like] left field, Oracle continues to build out one of the more complete compute stacks there is.

What’s common across all of these is that they are holding up ways to change the paradigm of all the separate silos in IT. Perhaps the one common theme is that the future will contain less piece-parts and end-user engineering than it has until now. Call it integration, stacks, convergence: you can even whisper the “cloud” word! Call it what you will, but IT generally, and storage specifically, will be more packaged. And that does have the potential to provide major financial change. Of course, these vendors all have masses of good, solid, evolutionary stuff as well, but it’s the tectonic infrastructure shifts that we’ll be looking back upon in a decade or two and not the details of an upgrade here or a sales campaign there.

Sometimes, what looks like just a technology “tweak” or marginal play, can also represent the tip of the iceberg for a really big change. I already mentioned solid state and it is looking to be that. Even though we’ve been talking about it in terms of upgraded excitement and expectations for a few years now, the signs are that the prophecies will come true. Solid state–largely flash today–will serve an ever-higher percentage of active IO; it’s happening already and the pace will accelerate as the economics get more attractive. While the rush of individual announcements of late (including, for example, a PCIe device from Micron, small mega-fast products from Texas Memory, and enhanced endurance from AnoBit, plus the sale of Pliant to SanDisk and this week new drives from OCZ and software from FlashSoft to drive IO acceleration in server) look evolutionary, they are simply manifestations of a larger shift.

More on all this in the coming weeks …

Read more of Mark’s blog entries at The Business of Storage.

You can read Mark’s other blog entries at The Business of Storage.

Background

ESG conducted an in-depth survey of senior IT professionals concerning their organizations’ IT spending plans and priorities for 2011. Survey participants represented midmarket (100 to 999 employees) and enterprise-class (1,000 employees or more) organizations in North America and Western Europe. The top ten responses are shown in Figure 1. Priorities were fairly evenly distributed with every listed priority getting at least one response. The top two priorities for organizations were increasing the use of server virtualization and managing data growth.[1]

In another recent study, ESG found that server virtualization is becoming ubiquitous in IT organizations of all sizes. Of 1,602 survey respondents, nearly three-quarters (74%) said their organization currently uses server virtualization. An additional 19% of organizations are in the evaluation or planning phase, leaving just 7% of the midmarket and enterprise organizations surveyed by ESG not currently using server virtualization and having no plans to do so.[2]

IT professionals who have experience with server virtualization know that like server virtualization, storage virtualization enables consolidation, which reduces complexity and cost. Storage and server virtualization are also similar in their ability to provide increased availability, fault tolerance, and mobility. Given the similarities, it’s not surprising that, in an earlier study, ESG learned that an increasing number of organizations are deploying server and
storage virtualization together. Fifty-seven percent had either already deployed storage virtualization in conjunction with server virtualization or planned on doing so within the next 24 months at the time of that survey.[3]

One of the most significant findings involved the substantial economic benefit of deploying an intelligent storage virtualization solution in conjunction with server virtualization. As an example, 79% of early adopters with a large SAN believe they have reduced their annual storage hardware spending to some degree, with a reported mean annual savings of 19.7%. Storage hardware savings were the most significant due to the ability to reclaim and re-use existing storage (the least expensive storage is the storage you already have). Consolidating storage software that previously ran on multiple servers or storage systems onto a centrally managed infrastructure also helped reduce storage software spending and storage administration costs.

IBM Storwize V7000

The IBM Storwize V7000 is a midrange disk system that has been designed to be easy to use and manage, enabling rapid deployment with minimal resources. Storwize V7000 is virtual storage that offers efficiency and flexibility through built-in SSD optimization and “thin provisioning” technologies while enabling users to virtualize and re-use existing disk systems, as shown in Figure 2.

Storwize V7000 advanced functionality also enables non-disruptive migration of data from existing storage, simplifying implementation and minimizing disruption to users.

IBM Storwize V7000 can be deployed in a data center to provide:

• Dramatic performance improvement with IBM System Storage Easy Tier functionality, providing automatic storage tiering between hard disk drives and solid state drives (SSD)
• Optimized capacity utilization and availability
• Significantly improved storage management ease-of-use
• A centralized platform for the management of SAN-attached storage capacity
• Heterogeneous interoperability between a wide variety of hosts, operating systems, and storage systems
• Online data migration services
• Remote replication and disaster avoidance services
• Space efficient provisioning and copy services

Storwize V7000 may be used to provide all these functions to enhance existing installed storage or as part of a new storage deployment.

The goal of ESG Lab’s testing of the IBM Storwize V7000 was to validate enterprise-class features and functionality including ease of use, non-disruptive virtualization, automated tiering, data mobility, and copy services. The performance benefits of IBM System Storage Easy Tier were also explored.

ESG Lab Validation

ESG Lab performed hands-on evaluation and testing of the IBM Storwize V7000 at an IBM facility located in Tucson, Arizona. A SAN-attached IBM 3850-X5 server was used during testing as shown in Figure 3 and documented in the Appendix. An IBM DS4800 storage array was used to validate the ability to virtualize and manage external storage.

Ease of Use

ESG Lab testing began with a high level examination of the Storwize V7000 user interface shown in Figure 4.

The Storwize V7000 graphical user interface is a browser-based, easy to navigate intuitive GUI. The home screen shown in this screen shot provides an excellent graphical flow chart of the system components and steps required to provision storage to host systems. The getting started view also provides e-learning and information center links that can be leveraged to provide detailed information on each component in the storage provisioning process.

ESG Lab Testing

ESG Lab leveraged the getting started view to confirm physical system components and resources as well as a gateway to configuring more advanced system features. The flow diagram and component detail information was repeatedly referenced during the configuration of many storage components including managed disks, pool, and volume creation to host provisioning and external device imports.

The Storwize V7000 GUI provides an intuitive view of the system components organized on the left side of each pane in column format, allowing the administrator to scroll from component to component with ease. Hovering over an individual component will provide drilldown options of the administration tasks that can be performed for that component. Figure 5 shows the administration options available for the Storwize V7000 volumes.

Figure 5 highlights the drilldown options for volume operations, but it should also be noted that the middle of the screen displays system status information in this view. The system status pane provides a high level detailed view of system information moving from left to right. Figure 5 displays a physical view of system capacity via the bar graph to the left and a component view in the middle of the pane. From the system status screen, users can hover over each object and display detailed system information. The blue section of the bar chart will display used capacity while the black section will show total system physical capacity. A click on the link below the physical display will present a table of system statistics to the right or users can click on a disk shelf to display detailed information on the drives in each enclosure.

The Storwize V7000 disk system presents storage as volumes. Volumes are virtual containers, abstracted from physical disk drives by the concepts of managed disks and pools presented to host servers. The volume by pool view allows the administrator to see detailed information for each Storwize V7000 volume, including its pool association. Figure 6 shows the volume details and administration options available for volumes in pool WEB-SP.

ESG Lab Testing

ESG Lab validated the ease of executing typical storage administration tasks with the Storwize V7000 by creating a thin provisioned volume copy in an existing storage pool. Figure 6 shows the volume copy creation steps as well as the percentage completed. Volume copies can be assigned to hosts and used for a number of operations (e.g., off-host backups, test/dev, or data migration).

Efficient Data Management

With the Storwize V7000, the storage administrator has the ability to deploy enterprise-class management features to improve storage efficiencies. Thin provisioning and snapshot technology are included in the Storwize V7000 code.

Thin provisioning employs an abstraction layer between the physical disk drive and the storage presented to the host system. With the Storwize V7000, physical disks are configured in RAID sets as managed disks, or MDisks. The MDisks are then assigned to storage pools. Virtualized volumes are created from pool storage and presented to hosts. The volumes can be created larger than the physical storage as shown in Figure 7. Because the volume is abstracted from the physical storage layer, capacity can be dynamically managed transparently to the host operating system as requirements change.

Figure 7 shows thin provisioned storage as compared to traditionally provisioned storage. The left side of the figure shows the one to one relationship of a traditional volume with a 5 TB volume presented to a host server consuming 5 TB of physical space. The right side of the figure shows a 5 TB virtual container being presented to a host server using only 100 GB of physical space.

ESG Lab Testing

ESG Lab tested thin provisioning on the Storwize V7000 by over-subscribing the storage array. Thin provisioned volumes were created with more capacity than the physical array as shown in Figure 8. The Storwize V7000 had a physical capacity of 38.8 TB and a used capacity 34.0 TB. ESG Lab created thin provisioned volumes to present a virtual capacity of 41.5 TB.

The bar graph in Figure 8 shows the used capacity in blue, the physical capacity in black, and the virtual capacity in green. At this point in ESG Lab testing, the Storwize V7000 had presented more capacity to host servers than it physically contained, yet the used capacity was only 34.0 TB.

FlashCopy is a copy service provided with the Storwize V7000 disk system. There are three different FlashCopy options available in the Storwize V7000—Snapshot, Clone, and Backup—as shown in Figure 9. Implementing FlashCopy enables the instantaneous copy of data from a source volume to a target volume.

The FlashCopy Snapshot is a thin provisioned copy with the ability to automatically expand as necessary. The FlashCopy Clone option is a copy with the same properties as the source volume. FlashCopy Backup is a copy with the same properties as the source and the ability to create incremental copies.

ESG Lab Testing

ESG Lab validated snapshot functionality on the Storwize V7000 by creating a FlashCopy Clone of an existing volume that was already presented to a Windows host with an 80 GB database file residing on it. Figure 10 show the first steps for creating a FlashCopy copy volume from the FlashCopy Actions menu.

The new Clone option was selected and mappings were created from source volume (MWL-JET1_01) to target volume (MWL-JET1) as shown in Figure 11.

When the Clone copy was fully synchronized on the Storwize V7000, the 80 GB database file was deleted on the Windows host. After the file deletion, the source and target mapping were reversed on the Storwize V7000. The Windows host system was rebooted and the 80 GB database file was immediately available on the host while the volume synchronization process ran in the background as shown in Figure 12.

Powerful Enterprise-class Features

The Storwize V7000 disk system is a feature-rich storage solution that provides a powerful set of management tools enabling great flexibility while helping to make the most challenging storage administration tasks easy to tackle. As shown previously in Figure 5, both internal and external storage can be virtualized and managed via the Storwize V7000 graphical user interface. The GUI provides an easy to use management console and includes configuration wizards for many administration tasks. The ability to combine multiple storage concepts like external storage device imports and thin provisioning add to the power of the Storwize V7000, creating flexible capacity management.

ESG Lab Testing

ESG Lab tested and confirmed the ability to virtualize external storage devices by importing, presenting, and managing a DS4800 volume and its associated data via the Storwize V7000 system. As shown in Figure 13, a DS4800 volume was imported to the Storwize V7000, combined with a thin provisioned internal volume, and presented to a host server with multiple virtual Windows machines.

The volume import procedure was conducted and configured using the Storwize V7000 Import Wizard. The wizard made the import procedure easy and intuitive by providing descriptions and configuration options during the short two-step setup process.

The Import Wizard is just one of the administration wizards that can be launched from the Storwize V7000 GUI. As shown in Figure 14, the Import Wizard was used to import a volume and its data from and external DS4800 storage array.

Figure 15 shows the pool selection screen—the second and final step of the volume import process.

The original volume data was maintained during the import process and once imported, presented back to the original Windows machine via the Storwize V7000.

Upon completion of the import process, a thin provisioned volume (Copy 1) was added to the original imported volume as shown in Figure 16.

Figure 16 shows the traditional provisioned Copy 0 volume and the thin provisioned Copy 1 volume. It should be noted that each volume contained the same 8.8 GB of used data. The traditional volume used 400 GB of physical space to store 8.8 GB while the thin volume only consumed the 8.8 GB of physical space to store the same data.

Predictably Scalable Performance

IBM uses the latest hardware and field-proven storage software from the SVC and DS8000 storage product lines to deliver the exceptional performance of the IBM Storwize V7000. The hardware platform is built for speed and scalability using a pair of highly available controllers powered by the latest Intel chipsets. Each controller is equipped with 8 GB of high speed cache memory. Up to eight Fibre Channel interfaces (8 Gbps) and four iSCSI interfaces (1 Gbps) are supported across a pair of controllers for host connectivity. A mix of SAS, Nearline SAS, and SSD drives is supported to meet a wide variety of price/performance requirements.

The performance of the hardware platform is amplified by field-proven IBM storage software including the SVC storage virtualization engine, the DS8000 RAID stack, and Easy Tier, illustrated in Figure 18, which automatically moves frequently-used data to high performance SSD drives and infrequently-used data to cost-effective Nearline SAS drives. Thousands of customers across the globe rely on the field-proven performance of the SVC engine at the heart of the Storwize V7000.

The performance scalability of the SVC engine has been demonstrated in published Storage Performance Council (SPC) results. SPC is a vendor-neutral standards body designed to be a source of comparative storage subsystem performance information. There are two application workloads supported by SPC: SPC-1 simulates an interactive database application and SPC-2 simulates a bandwidth-intensive application such as those that process large files, large database queries, or video on demand. When this report was published, IBM had the industry-leading SPC-1 result of 380,489 IOPS with a six-node SVC cluster in front of two IBM DS8700 disk arrays.[4] The SVC engine running inside a dual controller Storwize V7000 had excellent published results of 56,510 SPC-1 IOPS[5] and 3,123 SPC-2 MBPS.[6]

ESG Lab Mixed Workload Testing

ESG Lab has confirmed that the performance and scalability of the Storwize V7000 is well suited for a mix of applications running in a consolidated virtual server environment. The ESG Lab mixed workload performance benchmark was designed to measure the performance capabilities of a single Storwize V7000 storage system subjected to an IO-intensive mix of virtualized business applications running on a pair of IBM x3850 X5 servers in a virtual server environment powered by VMware vSphere. Taking a cue from the server-focused VMware VMmark benchmark, a cell concept was used during the design of this test. Each cell was composed of four applications (e-mail, database, web server, backup jobs), each running in its own virtual machine. IBM System x3850 X5 servers were used to exercise up to four cells and sixteen virtual applications in parallel. The Microsoft Jetstress utility was used to simulated Exchange 2010 traffic and the Oracle Orion utility was used to emulate OLTP and OLAP database traffic. The industry standard Iometer utility was used to emulate thousands of web server users and up to four backup jobs.

Storwize V7000 disk capacity was used for all storage capacity including VMware virtual disk files (VMDK), Windows Server 2008 R2 operating system images, application executables, and application data. The operating system images were installed on VMDK volumes. All of the application data volumes under test were configured as mapped raw LUNs (also known as raw device mapped, or RDM, volumes).

Mixed workload testing was performed with a single IBM Storwize V7000 equipped with 215 SAS drives. The Exchange VMs were configured with 64 SAS drives for the database and 25 drives for logs. Oracle was configured with 52 drives and the web server and backup workloads were configured with 32 drives each.

Volume ownership was balanced across the dual controllers within the Storwize V7000 and distributed evenly over the eight host interfaces. The volumes were spread evenly over two VMware host groups with a multipath policy of most recently used (MRU).

As shown in Figure 17, a single Storwize V7000 delivers the IO processing power and bandwidth needed to simultaneously support 54,208 simulated Exchange 2010 mailboxes and 5,015 Oracle Orion small database IOs per second and 849 MB/sec of throughput for large OLAP Oracle Orion operations and 5,015 simulated web server IOPs and 644 MB/sec of throughput for bandwidth-intensive backup jobs—all while delivering predictably fast response times. Note how the performance of the Storwize V7000 scaled well as a mix of real-world application workloads ran in parallel on up to 16 virtual machines running on a pair of powerful IBM x3850 X5 servers.

Easy Tier

The solid state drives (SSD) supported by the Storwize V7000 can be used to improve application performance while simultaneously reducing power and cooling requirements in the data center. Easy Tier increases the efficiency and simplicity of deploying SSD drives. Easy Tier, and its associated tool Storage Tier Advisor, help IT managers plan for, deploy, and manage the deployment of SSD drives in conjunction with traditional spinning hard drives.

As shown in Figure 18, Easy Tier uses sub-LUN data tiering technology that can be used to automatically move frequently-used data to high performance SSD drives and infrequently-used data to Nearline drives. Easy Tier can be configured to improve the performance for existing storage being virtualized by the IBM Storwize V7000 system and  can operate with any combination of internal or externally virtualized SSD or HDD devices.

ESG Lab evaluated the performance boost that can be achieved with SSD drives and Easy Tier. Twenty-four SSD drives were added to the pool of SAS drives used during the first round of mixed workload testing. An Easy Tier policy was configured to automatically move frequently-used application data to 24 high speed SSD drives.

After re-running the mixed workload on all of the available virtual machines for 24 hours, Easy Tier more than tripled the amount of mixed workloads the solution could handle. [7]

At the application level, Easy Tier delivered:

• 341% more web server IOs per second
• 277% more Oracle OLTP IOs per second
• 69% more Exchange 2010 users
• 33% faster Exchange database read response times
• 43% faster Oracle OLTP IO response times

ESG Lab Validation Highlights

• The Storwize V7000 was exceptionally easy to configure and manage. ESG Lab was able to provision virtualized, thin provisioned storage in less than 10 minutes using intuitive, automated wizards.
• ESG Lab performed a non-disruptive import of an external disk volume and presented the data intact back to the original host without requiring a reformat of the storage during import.
• IBM Storwize V7000 was able to thin provision the imported volume to better use capacity without destroying user data.
• An IBM Storwize V7000 and a pair of IBM x3850 X5 servers delivers the IO processing power and bandwidth to concurrently support up to:
  • 54,208 mailboxes using the Microsoft Exchange 2010 Jetstress  utility
  • and 5,015 database IOs per second for small OLTP IOs with the Oracle Orion utility
  • and 849 MB/sec of throughput for large OLAP Oracle Orion operations
  • and 5,015 simulated web server IOPs
  • and 644 MB/sec of throughput for simulated backup jobs
  • with predictably fast response times and scalability
• As the number of virtual machines sharing a single Storwize V7000 increased, performance scaled in a near linear fashion with predictably fast response times (5.2 to 11.3 milliseconds for e-mail database reads, 4.72 to 5.06 milliseconds for Oracle OLTP IO operations).
• Easy Tier and 24 SSD drives more than tripled the mixed IO capacity of the Storwize V7000 (3.21 times more) as it noticeably improved application-level performance:
  • 33% faster Exchange response times
  • 43% faster Oracle OLTP database response times

Issues to Consider

• ESG Lab tested a Storwize V7000 system with a total of 240 drives (216 SAS, 24 SSD), which was beyond the maximum number of supported internal drives when this report was published. In this context, internal drives are the drives located within drive expansion trays connected to, and managed by, Storwize V7000 controllers. This limitation does not apply to the drives within a disk array that is externally virtualized with the Storwize V7000. IBM has publically stated that support for 240 internal drives is planned for general availability in early 2011.
• The field-proven architecture of the SVC currently support SVC code running on up to four pairs of controllers managed as a single system. The IBM Storwize V7000, which leverages the same SVC code base, currently supports only a single pair of controllers. ESG Lab believes that support for more than a single pair of controllers would be a welcome future enhancement for organizations that would like to scale the benefits of a singly managed pool of V7000 storage.
• ESG Lab is confident that most IT organizations will exceed the Easy Tier and SSD benefits documented in this report. This is due to the fact that real-world applications tend to have concentrated hot spots that are great candidates for Easy Tier and SSD. In contrast, the synthetic benchmarks used during ESG Lab testing are more uniformly random than real-world applications. As a result, ESG Lab believes that less SSD capacity should provide similar—if not better— performance benefits in real-world production environments.
• The test results/data presented in this document are based on industry-standard benchmarks deployed in a controlled environment. Due to the many variables in each production data center environment, it is still important to perform capacity planning and testing in your own environment to validate a storage system configuration.

The Bigger Truth

One of the largest challenges facing IT today is managing unabated data growth in dynamic, virtualized environments. Innovation and automation are needed to reign in the cost and complexity associated with unrelenting growth and ever-changing business requirements. Virtualization technologies are needed to simplify, consolidate, and automate routine IT functions.

Innovative virtualization technologies are being deployed by a growing number of IT organizations. A recent ESG survey of 1,602 IT professionals indicates that server virtualization is leading the charge. Seventy-four percent of respondents report that they are actively using server virtualization.[8] Server virtualization, like storage virtualization, is powerful game-changing technology that can be used to consolidate and simplify complex IT infrastructure. While the cost savings of consolidation drove the first wave of server virtualization adoption, the second wave is being driven by the mobility, flexibility, and enhanced fault tolerance that can be achieved with a virtualized IT infrastructure.

Like the early adopters of server virtualization, early adopters of storage virtualization report that management complexity has been simplified and costs have been reduced. Given the synergistic benefits, it is not surprising that ESG research indicates that a growing number of forward-looking organizations are deploying storage and server virtualization together. Fifty-seven percent have deployed both already or plan on doing so.[9]

ESG has confirmed that the IBM Storwize V7000 is a modular storage system with built-in efficiency and storage virtualization. Presenting virtual disk capacity on a just in time basis, space-efficient virtual disks eliminate the cost of unused and unallocated storage as they simplify the task of assigning storage capacity to applications.  Leveraging the field-proven storage virtualization capabilities of the IBM SVC product line, the Storwize V7000 provides a centralized platform for valuable data services including online data migration, copy services, mirroring, and remote replication.

ESG Lab was particularly impressed with the exceptional ease of use and performance of the IBM Storwize V7000. The management GUI, which was derived from the IBM XIV storage product line, is wizard-driven and extremely easy to use. Industry standard SPC results and ESG Lab mixed workload testing in a VMware environment confirmed the performance prowess of the platform. Performance testing with Easy Tier and SSD drives reduced application-level response times as it more than tripled the amount of work that the virtualized infrastructure could handle.

IBM Storwize V7000 is an easy to use, rock-solid midrange storage platform with sophisticated enterprise-class functionality and built-in storage efficiency. ESG Lab believes that IT managers within mid- to large-sized organizations shopping for their next storage infrastructure upgrade—especially those with legacy storage they’d like to continue to utilize—should seriously consider the IBM Storwize V7000.

Appendix

[2] Source: ESG Research Report, The Evolution of Server Virtualization, November 2010.

[3] Source: ESG Research Report, The Impact of Server Virtualization on Storage, December 2007, N=332.

[4] http://www.storageperformance.org/benchmark_results_files/SPC-1/IBM/A00087_IBM_DS8700_SVC-5.1-6node/a00087_IBM_DS8700_SVC5.1-6node_executive-summary-r1.pdf

[5] http://www.storageperformance.org/benchmark_results_files/SPC-1/IBM/A00097_IBM_Storwize-V7000/a00097_IBM_Storwize-V7000_2-node_SPC1_executive-summary.pdf

[6] http://www.storageperformance.org/benchmark_results_files/SPC-2/IBM_SPC-2/B00052_IBM_Storwize-V7000/b00052_IBM_Storwize-V7000_SPC2_executive-summary.pdf

[7] For more detail on the ESG Lab mixed workload testing of the IBM Storwize V7000, see the full ESG Lab Validation Report, IBM Storwize V7000: Real-world Mixed Workload Performance in VMware Environments, January 2011.

[8] Source: ESG Research Report, The Evolution of Server Virtualization, November 2010.

[9] Source: ESG Research Report, The Impact of Server Virtualization on Storage, December 2007, N=332.

“Scale-out” storage systems overcome the physical boundaries of racks and modules to create virtualized systems that can scale front-end processing power as well as back-end capacity by adding new processor or capacity nodes while still functioning as a single system. They often don’t need the levels of individual physical disk management, data layout, and performance tuning required by traditional monolithic and modular systems.

Consequently, scale-out platforms can provide a path to both increased performance and operational cost reduction—benefits that are increasingly putting these systems on the radars of enterprise IT buyers. Scale-out solutions can typically expand into the multi-petabyte range under a single system image, providing an ideal platform for storage consolidation. They help IT reduce management costs and data center equipment footprint, which reduces floor space and power and cooling costs. Finally, increased storage consolidation onto a shared resource means utilization rates are higher, so users get more “bang” for their storage buck.

Designed as a companion publication to ESG’s recent research report on scale-out storage market trends, ESG has created a scale-out storage market forecast—defined as ESG’s estimate of the total, worldwide annual capacity and revenue associated with scale-out storage in the commercial and government sectors—through the year 2015.

ESG Research Report Scale-out Storage Market Forecast 2010-2015 Executive Summary

The Challenges

A worldwide wave of server and storage consolidation is reducing the cost and complexity of delivering IT services to the business.   Consolidation is clearly a priority as a growing number of organizations embrace server virtualization technology.  In a recent survey, ESG asked IT decision makers to list their top priorities over the next 12-18 months.[1] As shown in Figure 1, increased use of server virtualization, data growth management, and data center consolidation were all top priorities.

However, despite the broad success of server virtualization, nagging issues and challenges exist. As a result, a low percentage of the potential workloads that can be virtualized have been migrated to virtual machines, and the consolidation ratios of virtual machines per physical server remains relatively low. A recent ESG survey explored the storage challenges associated with the next wave of server virtualization.[2] Given the rapid growth in the number of virtual machines being deployed, it’s no surprise that scalability, performance, and the overall volume of storage capacity have been identified as key challenges.

Consolidation and server virtualization are changing the way that IT infrastructure is managed.   Managing IT infrastructure from a centralized virtual server console is simplifying the process of deploying new applications.  Storage system management tools need to be integrated with the virtual server management interface and higher level application management frameworks to increase the value of a centrally managed IT infrastructure.

The Solution

The Engenio 2600 is a modular data storage system with balanced mixed workload performance and a rich set of application aware management tools.   The Engenio 2600 uses the latest 6 Gbps Serial Attached SCSI (SAS) interface for the back-end connection to disk drives and a rich set of front-end server connectivity options.  The iSCSI host attach option leverages the affordability and ubiquity of industry standard Ethernet technology.  The high performance 8 Gbps Fibre Channel (FC) host attach option provides connectivity for applications with high performance and availability requirements.  SAS, which has traditionally been used for an affordable connection to entry level disk arrays (often referred to as just a bunch of disks, or JBOD), is also supported for high speed, low cost host connectivity.

The Engenio 2600 builds on the field-proven reliability of LSI Corporation, based on the more than 300,000 storage systems that it has deployed world-wide.  Supporting up to 4 GB/sec of throughput and 40,000 IOPS, the key capabilities of the Engenio 2600 include:

• Four native 6 Gbps SAS host interfaces.
• Up to eight additional 1 Gbps iSCSI host interfaces.
• Up to eight additional 8 Gbps FC host interfaces.
• Up to four additional 6 Gbps SAS interfaces.
• Up to 192 high-speed SAS, cost-effective nearline SAS, self encrypting, or solid state drives.
• 3.5 and 2.5 inch drive enclosures.
• Up to 4 GB of cache.
• Advanced recovery capabilities, including snapshots and volume copies.
• Advanced availability capabilities, including dual controllers and remote replication.

A  growing set of application aware management plug-ins provide tight integration with management tools from Microsoft, VMware, Oracle and others.  Plug-ins simplify the management of Engenio 2600 storage with built-in provisioning, monitoring, event management, and advanced data recovery.  A growing set of management frameworks are supported, including VMware vSphere, Microsoft Systems Center Operations Manager (SCOM), and Oracle Enterprise Manager.

The Results

This report documents the performance and application management capabilities of Engenio 2600.  Performance testing with a mix of real-world applications in a VMware vSphere-enabled virtual server environment and mix of management framework plug-ins explores how:

• A single Engenio 2600 with 96 10K RPM SAS drives attached to a pair of powerful multi-core servers running a mix of real-world application workloads in 16 virtual machines supports up to:
  • 20,458 mailboxes using the Microsoft Exchange 2010 Jetstress  utility
  • and 3,724 database IOs per second for small OLTP IOs with the Oracle Orion utility
  • and 856 MB/sec of throughput for large OLAP Oracle Orion operations
  • and 3,490 simulated web server IOPs
  • and 1,260 MB/sec of throughput for simulated backup jobs
  • with predictably fast response times and scalability.
• Management tools were examined with a goal of confirming that provisioning, monitoring, and protecting application data residing on an Engenio 2600 storage system can be simplified with application aware capabilities including:
  • LSI Storage vCenter vSphere Plug-in.
  • LSI Storage Oracle Enterprise Manager Plug-in.
  • LSI Site Recovery Adapter for VMware Site Recovery Manager (SRM).
  • LSI integration with Microsoft Disbursed Cluster Storage Failover (DCSF).
  • LSI Management Pack for Microsoft Systems Center Operations Manager (SCOM).

The predictably fast, mixed workload performance scalability of the virtualized environment tested by ESG Lab is summarized in Figure 3. The results will be explored in detail later in this report, but for now it should be noted that the performance of the Engenio 2600 scaled well as a mix of real-world application workloads run in parallel on up to 16 virtual machines.

The balance of this report explores how the tests were accomplished, what the results mean, and why they matter to your business.

ESG Lab Validation

The real-world performance capabilities of the Engenio 2600 storage system were assessed by ESG Lab at an LSI facility. The methodology presented in this report was designed to assess the mixed workload performance and manageability of an Engenio 2600 in virtual server and consolidated application environments.

Mixed Workload Storage Performance Testing

Conventional server benchmarks were designed to measure the performance of a single application running on a single operating system inside a single physical computer. SPEC CPU2000 and CPU2006 are well known examples of this type of server benchmarking tool. Much like traditional server benchmarks, conventional storage system benchmarks were designed to measure the performance of a single storage system running a single application workload.  The SPC-1 benchmark, developed and managed by the Storage Performance Council, is a great example. SPC-1 was designed to assess the performance capabilities of a single storage system as it services an online interactive database application.

Traditional benchmarks running a single application workload can’t help IT managers understand what happens when a mix of applications are deployed together in a virtual server environment. To overcome these limitations, VMware created a mixed workload benchmark called VMmark.  VMmark uses a tile-based scheme for measuring application performance and provides a consistent methodology that captures both the overall scalability and individual application performance of a virtual server solution.  VMmark measures performance as a mix of application workloads are run in parallel within virtual machines deployed on the same physical server.

The novel VMmark tile concept is simple, yet elegant. A tile is defined as a mix of industry standard benchmarks that emulate common business applications (e.g., e-mail, database, web server). The number of tiles running on a single machine is increased until the server runs out of performance. A score is derived so that IT managers can compare servers with a focus on their performance capabilities when running virtualized applications.

While VMmark is well suited for understanding the performance of a mix of application running on a single server, it was not designed to assess what happens when a mix of applications is run on multiple servers  sharing a single storage system. VMmark tends to stress server internals more than it does the storage system. The methodology developed by ESG Lab and presented in this report was designed to stress the storage system more than the servers. Taking a cue from the VMmark methodology, a tile-based concept was used. Each tile is composed of a mixture of four application workloads. Two physical servers, each configured with eight virtual machines, were used to measure performance as the number of active tiles was increased from one to four.

VMmark testing is performed with a single server, often attached to multiple storage systems.  When server vendors publish VMmark results, they make sure there is plenty of storage available so they can record the highest VMmark score. This provides IT managers with a fair comparison of the performance capabilities of competitive server technologies.

As shown in Figure 4, ESG Lab storage-focused benchmarking uses a different approach. Instead of testing with a single server and more than enough storage, multiple servers are attached to a single storage system. Rather than running application-level benchmarks which stress the CPU and memory of the server, lower level industry standard benchmarks are used with a goal of measuring the maximum mixed workload capabilities of a single storage system.

Test Bed

VMware vSphere version 4.1 was installed on a pair of servers, each with a pair of quad-core processors and a pair of dual-port host adapters. An Engenio 2600 storage system with 96 10K RPM SAS drives was connected to the servers through a pair of 8 Gbps FC switches, as shown in Figure 5.

Workloads

Industry standard benchmarks were used to emulate the IO activity of four common business application workloads:

• E-Mail: The Microsoft Jetstress 2010 utility was used to generate e-mail traffic. Similar to the Microsoft LoadGen utility used in the VMmark benchmark, Jetstress simulates the activity of typical Microsoft Exchange users as they send and read e-mails, make appointments, and manage to-do lists. The Jetstress utility is, however, a more lightweight utility than LoadGen. Using the underlying Jet Engine database, Jetstress was designed to focus on storage performance.
• Database: The Orion utility from Oracle was used to generate database traffic. Much like Jetstress, Orion is a lightweight tool that is ideally suited for measuring storage performance. Orion was designed to help administrators understand the performance capabilities of a storage system, either to uncover performance issues or to size a new database installation without having to create and run an Oracle database. Orion is typically used to measure two types of database activity: response-time sensitive online transaction processing (OLTP) and bandwidth sensitive online analytic processing (OLAP).
• Web Server: The industry standard Iometer utility was used to generate web server traffic. The IO definition was composed of random reads of various block sizes. The web server Iometer profile used for this test was originally distributed by Intel, the author of Iometer. Iometer has since become an open source project.[3] Iometer tests were performed on Windows physical drives running over VMware raw mapped devices.
• Backup: The Iometer utility was used to generate a single stream of large block sequential read traffic.  Operations that tend to generate this type of traffic include backup operations, scan and index operations, long running database queries, bulk data uploads, and copies. One 256 KB sequential read workload was included in each tile to add a throughput intensive component to the predominantly random IO profile of interactive e-mail, database, and web server applications. As most experienced database and storage administrators have learned, a throughput-intensive burst in IO traffic can drag down performance for interactive applications, causing performance problems for end-users. Adding a few streams of throughput-intensive read traffic was used to determine whether interactive performance would remain predictably responsive as the amount of mixed IO utilization increased.

Each of the four workloads ran in parallel, with the Jetstress e-mail test taking the longest to complete (approximately three hours). Configuration details and the settings for each of the workload generators are documented in the appendix.

Results

In a way, storage system benchmark testing is like an analysis of the performance of a car. Specifications, including horsepower and acceleration from 0 to 60, are a good first pass indicator of a car’s performance. But while specifications provide a good starting point, there are a variety of other factors that should be taken into consideration including the condition of the road, the skill of the driver, and gas mileage ratings. Much like buying a car, a test drive with real-world application traffic is the best way to determine how a storage system will perform.

Characterization

Performance analysis began with an examination of the low level aggregate throughput capabilities of the test bed.  This testing was performed using the Iometer utility running within the eight virtual machines that were used later during mixed workload testing.  The eight virtual machines accessed Engenio 2600 storage through eight 8 Gbps FC interfaces.

Iometer access definitions, which measured the maximum throughput from disk, were used for this first pass analysis of the underlying capabilities of the Engenio 2600.[4] Similar to a dynamometer horsepower rating for a car, maximum throughput was used to quantify the power of a turbo-charged Engenio 2600 storage engine. As shown in Figure 6, ESG Lab recorded a maximum throughput of 4.2 GB/sec.

What the Numbers Mean

• Much like the horsepower rating of a car, the throughput rating of a storage system is a good indicator of the power of a storage system’s engine.
• Storage throughput is a measure of the bandwidth available to the system. Throughput can be measured on a stream or aggregate basis. A stream is represented by one application or user communicating through one IO interface to one device. Aggregate throughput is a measure of how much data the storage system can move on a whole for all applications and users.
• ESG Lab recorded a peak aggregate throughput of 4.2 GB/sec in a VMware vSphere environment.
• When comparing the performance capabilities of two servers in a virtual server environment, the server with more cache tends to perform better. ESG Lab is confident that a similar pattern holds true for storage systems. A storage system with more cache—and better caching algorithms—should perform better in a virtual server environment.
• ESG Lab characterization testing indicates that the Engenio 2600 has more than enough cache and front-end bandwidth to meet the needs of virtualized applications.
• ESG Lab is convinced that the caching algorithms of the Engenio 2600 provide a significant performance boost during virtualized mixed application testing.

Virtual Machine Utilization

Mixed application testing began with a quick analysis of server CPU and RAM utilization to make sure that there were no bottlenecks between virtualized application workloads and the Engenio 2600. As expected, utilization was manageably low on the physical servers during the busiest mixed workload test (3.2% CPU utilization and 32% memory utilization were observed at the vSphere client).

Mixed Real-world IOPS Scalability

IOs per second, or IOPS, is a measure of the number of operations a storage system can perform in parallel. When a system is able to move a lot of IOPS—from disk and from cache— it will tend to be able to service more applications and users in parallel. Much like the horsepower rating for a car engine, the IOPS rating for a storage controller can be used as an indicator of the power of a storage system engine.

While IOPS out of a cache is typically a big number and can provide an indication of the speed of the front end of a storage controller, IOPS from disk is a more useful metric when determining the real-world performance of a storage system servicing a mix of business applications. For example, e-mail and interactive database applications tend to be random in nature and therefore benefit from good IOPS from disk. With that said, a mix of real-world applications tends to generate random and sequential IO traffic patterns that may be serviced from disk or from cache.

ESG Lab measured IOPS performance as reported by the Engenio 2600 as the number of virtual machines running mixed real-world application workloads increased from four through sixteen. With a mix of random and sequential IOs over 92 disk drives, the goal was not to record a big IOPS number; the goal with this exercise was an assessment of the scalability of the Engenio 2600 as an increasing number of applications are consolidated onto a single virtualized platform. The IOPS scalability during the peak period of mixed workload activity is shown in Figure 7.

What the Numbers Mean

• IOPS varied throughout the mixed workload test with peaks occurring during the Orion small IOPs phase and toward the end as the Jetstress utility performed a database consistency check.
• A peak of 14,709 IOPS was recorded during the four tile run.
• IOPS scaled well as mixed real-world application traffic increased from four through sixteen virtual servers.

Handling Throughput Spikes with Ease

As noticed during IOPS monitoring, peaks of throughput activity could be correlated to the periodic behavior of real-world applications. Two bursts of aggregate throughput were observed: the first during the Oracle large MBPS test which simulates a throughput-intensive OLAP application and the second during the Jetstress database consistency check. A VMware vSphere view of mixed workload performance on one of the servers is shown in Figure 8.

What the Numbers Mean

• An aggregate throughput level of 2.1 GB/sec was recorded as mixed, real-world applications were run on 16 virtual machines sharing a single Engenio 2600 storage system (1.1 GB/sec for one of the two physical servers is shown in Figure 8).
• As throughput intensified during the Oracle Orion OLAP test phase, bandwidth utilization for other mixed workloads operating in parallel remained steady.

Mixed Application Performance Scalability

Having looked at the IOPS and throughput ratings of the turbo-charged Engenio 2600 engine, here’s where the rubber meets the road as we examine performance at the application level. The output from each of the industry standard benchmark utilities was analyzed to determine the performance scalability and responsiveness of real-world applications running in a consolidated virtual environment.

Microsoft Exchange

The IO and performance efficiency of Microsoft Exchange have improved significantly over the years. Architectural improvements in Exchange 2010, including a new store schema, larger page sizes (8 KB to 32 KB), improved read/write coalescing, improved pre-read support, and increased cache effectiveness, have reduced the number of IOs per user up to 70% compared to Exchange 2007.[5] ESG Lab typically uses a value of 0.5 IOPS per mailbox to emulate a heavy Exchange user environment when testing with Jetstress 2007. A value of 0.12 IOPS per mailbox was used during Jetstress 2010 testing to reflect the 70% reduction in IOPS compared to Exchange 2007.

The Microsoft Jetstress 2010 utility was used to see how many simulated e-mail users could be supported by the Engenio 2600 during mixed workload testing. The number of IOPS and response time for each database and log volume was recorded at the end of each Jetstress run. A response time goal of 20 milliseconds or less for database reads is required to pass the test. These values are defined by Microsoft as a limit beyond which end-users will feel that their e-mail system is acting slowly.[6] The results are shown in Figure 9 and itemized in Table 1.

What the Numbers Mean

• The single tile mixed application test supported 5,067 Exchange users with an average DB disk response time of 5.2 milliseconds.
• Performance scaled to 20,458 users while the Engenio 2600 was busy servicing other applications concurrently.
• As the number of simulated e-mail users was increased, the Engenio 2600 provided excellent response times that are well within Microsoft’s guidelines. Note that response times for database reads are below the Microsoft recommended maximum of 20 milliseconds, which is shown as a dotted line in Figure 9.
• The IO efficiency improvements in Exchange 2010 reduce the cost of delivering e-mail support in mixed virtual server environments. In this case, ESG Lab supported up to 20,458 mailboxes on four virtualized Exchange 2010 servers in a mixed workload environment—more than twice the expected number of supported mailboxes within an Exchange 2007 environment.

Oracle

The Oracle Orion utility was used to measure small transfer (8 KB) response time and large transfer (1 MB) throughput. The small transfer results are used to predict the performance and scalability of response time-sensitive interactive database applications (e.g., OLTP). The large transfer results are used to predict the performance of throughput-intensive online analytical processing (OLAP) and decision support systems (DSS).

ESG used the following guidelines from an Oracle OpenWorld presentation to interpret the results:

Target 5-10 millisecond for response time critical IO. Start by assuming 30 IOPS per disk for OLTP and 20 MB/sec per disk in DSS. This is way below the theoretical value, but allows for media repair etc.[7]

For new or non-existing applications, use business rules or data model transaction profiles flow to understand what a transaction is and then extrapolate for transactions per second or hour. Optionally, you can use the numbers we have seen in our consulting gigs. Note that these are just guideline values. Use the following as basic guidelines for OLTP:

Low transaction system – 1,000 IOPS or 200 MB/sec

Medium transaction system – 5,000 IOPS or 600 MB/sec

High-end transaction system – 10,000 IOPS or 1 GB/sec (rarely achievable)[8]

The results for the four tile Orion test are summarized in Table 2. A sample Orion report is shown in the Appendix.

What the Numbers Mean

• The four tile test achieved a grand total of 3,724 small IOPS and 856 large MBPS while the system was simultaneously running a mix of real-world application workloads.
• Using Oracle’s back of the envelope sizing guidelines, this level of IO activity falls between the performance guidelines for a “low transaction system” and a “medium transaction system.”
• The total number of small IOPS processed during the busy four tile test yielded a rate of 53.5 small IOPS per drive, which exceeds the conservative Oracle planning guideline of 30 IOPS per drive.
• Orion reported an average latency of 5.15 milliseconds for the small IOPs workload. Given the Oracle guidance of 5 to 10 milliseconds, ESG Lab believes that these are excellent results—especially given the mix of IO-intensive workloads being serviced by the Engenio 2600 in parallel.

Web Server and Backup Reader

Performance results as reported by the Iometer utility for the web server and backup workloads during the one, two, three, and four tile tests are listed in Table 3.

What the Numbers Mean

• Performance scaled in a nearly linear fashion as the number of virtual machines running in parallel was scaled from four to sixteen.
• Given the cache friendly, read-only nature of web server IO traffic, ESG Lab believes that these results indicate that the Engenio 2600 has the horsepower required to service tens of thousands of simultaneous page requests.
• Each of the four backup streams sustained at least 300 MB/sec of throughput for the entire duration of the mixed workload test.  A stream of this magnitude could service the data needs of a number of simultaneous backup jobs, a very aggressive scan and index job, or a throughput-intensive database table scan.

Much like the electrical system in your home, figuring out how many appliances you can run in parallel before blowing a fuse is not a function of the number of wires behind the walls. What matters more is the design of the circuits used to distribute the right amount of power to appliances.  ESG Lab testing indicates that the Engenio 2600 engine delivers the right amount of power to virtualized applications when needed.

Application Aware Storage and Data Management

The Engenio 2600 supports a growing number of application aware plug-ins. The plug-ins, collectively referred to as enFusion, are available as a free download for a number of popular application-level management, high availability, and recovery frameworks, including the VMware vSphere Client, Microsoft System Center Operation Manager (SCOM), and Oracle Enterprise Manager.    ESG Lab confirmed that each of the plug-ins can be used to monitor, provision, and troubleshoot the storage system from an application perspective.

As shown in Figure 11, ESG Lab confirmed that application aware plug-ins can be used to monitor, provision, and troubleshoot the storage system from a virtual server management console.

In this example, Engenio 2600 storage capacity is being monitored from the VMware vSphere client.  Note how storage system specific information including the number of hot spares can be monitored without having to switch to the Engenio 2600 management console.

Application aware storage can not only be used to passively monitor Engenio 2600 storage systems, it can also be used to actively change storage system settings. In the screen shot shown in Figure 12, a new logical drive is being configured with recommended settings from the VMware management console.  Once again it should be noted that the administrator doesn’t need to switch to the storage management console to provision a new volume.

Application aware data and storage management can also be used to manage the advanced data protection and recovery capabilities of the Engenio 2600. Point in time snapshots can be scheduled and activated at the application level for popular Microsoft applications including Exchange, SQL Server and SharePoint. Leveraging the Microsoft VSS protocol, snapshots can be managed at the application level for nearly instant data protection and quick and easy recovery of a corrupt application data.

Similarly, application aware data management can be used to automate the recovery of application services after a disaster.  For example, the Engenio 2600 Site Recovery Adapter (SRA) for VMware Site Recovery Manager (SRM) can be used to automate the recovery of applications running in a virtual machine after a site failure.  The SRA adapter leverages the remote replication capabilities of the Engenio 2600 and APIs specified by VMware to create an intuitive management interface that’s used to configure, test, and automate the recovery of virtual machines at a remote site.

ESG Lab tested an Oracle application with VMware Site Recovery Manager and previous generation Engenio disk arrays.[9] The configuration of a complex recovery plan involving multiple virtual machines was wizard-driven and easy.  Fifteen minutes after starting a failover with a single mouse click, the entire application environment was up and running at a remote recovery site.  Thirty minutes later, the same wizard-driven process was used to fail back to the primary data center. From an end-user perspective, the recovered environment felt exactly the same—regardless of the data center delivering the services.  The network addresses, logins, and operating system preferences were the same. No application data was lost. And last, but not least, for the Oracle 11g order entry application tested by ESG Lab, there was no noticeable difference in application performance—even as data was replicated to a recovery site 100 Km away.

ESG Lab performed a failover test to confirm that virtual machines protected by a VMware recovery plan (and the applications running in those virtual machines) can be successfully restarted at the secondary data center.  The screen shot shown in Figure 13 shows the progress of the first recovery moments after the Run button was clicked.  ESG Lab noted that progress was very easy to follow.  Successfully completed steps are depicted in green, the currently executing step is blue, and any failed steps are shown in red.

ESG Lab Validation Highlights

• 4.2 GB/sec of aggregate throughput was sustained during characterization testing in a VMware-enabled virtual server environment.
• A single Engenio 2600 attached to a pair of  servers running a mix of real-world application workloads in 16 virtual machines supports up to:
  • 20,458 mailboxes using the Microsoft Exchange 2010 Jetstress  utility
  • and 3,724 database IOs per second for small OLTP IOs with the Oracle Orion utility
  • and 856 MB/sec of throughput for large OLAP Oracle Orion operations
  • and 3,490 simulated web server IOPs
  • and 1,260 MB/sec of throughput for simulated backup jobs
  • with predictably fast response times and scalability.
• Management tools were examined with a goal of confirming that provisioning, monitoring, and protecting application data residing on an Engenio 2600 storage system can be simplified with application aware capabilities including:
  • Storage vCenter vSphere Plug-in.
  • Storage Oracle Enterprise Manager Plug-in.
  • Site Recovery Adapter for VMware Site Recovery Manager (SRM).
  • Integration with Microsoft Disbursed Cluster Storage Failover (DCSF).
  • Management Pack for Microsoft Systems Center Operations Manager (SCOM).

Issues to Consider

• Generally accepted best practices and predominantly default VMware and Engenio storage settings were used during the design of this test. As expected after any benchmark test of this magnitude, deep analysis of the results indicates that tuning would probably yield slighter higher absolute results. Given that the goal of this test was not to generate a big number, ESG Lab is confident that the results presented in this report meet the objective of estimating performance scalability and responsiveness as a growing number of virtual machines share a consolidated pool of Engenio 2600 storage.
• The test results/data presented in this document are based on industry-standard benchmarks deployed together in a controlled environment. Due to the many variables in each production data center environment, it is still important to perform capacity planning and testing in your own environment to validate a storage system configuration.

The Bigger Truth

Server virtualization is being deployed by a growing number of organizations to lower costs, improve resource utilization, provide non-disruptive upgrades, and increase availability. Each benefit is fundamentally enabled by de-coupling servers, applications, and data from specific physical assets. Storage virtualization takes those very same benefits and extends them from servers to the underlying storage domain—bringing IT organizations one step closer to the ideal of a completely virtualized IT infrastructure.

While the benefits of a completely virtualized infrastructure are obvious to most IT managers, performance and manageability are real concerns. Server, storage, and application administrators are looking for answers to a number of questions:

• Can we meet performance service level agreements for a mix of business-critical applications?
• Does the storage system have the horsepower to serve mixed real-world applications?
• Can the storage system scale to accommodate future growth and consolidation?
• Can storage management be simplified with tools that we are familiar with?

The Engenio 2600, with next generation 6 Gbps SAS back-end technology and a flexible mix of SAS, iSCSI, and FC host connectivity options, is ideally suited for consolidation and virtualization in medium-sized businesses, mid-range environments, and remote sites.

ESG Lab confirmed that the performance and scalability of the Engenio 2600 is well suited for a mix of applications running in a consolidated virtual server environment in small to medium-sized businesses. A single Engenio 2600 simultaneously supported 20,458 simulated Exchange 2010 mailboxes and 3,724 Oracle Orion small database IOs per second and 856 MB/sec of throughput for large OLAP Oracle Orion operations and 3490 simulated web server IOPs and 1,260 MB/sec of throughput for bandwidth-intensive backup jobs—all while delivering predictably fast response times.

ESG Lab confirmed that the growing family of freely available Engenio 2600 application aware plug-ins can be used to provision and manage data and storage from an application perspective.   The VMware vCenter plug-in makes it easy to monitor and provision storage from a VMware virtual server administration console.  The site recovery adapter for VMware site recovery manager makes is easy to define, automate, and test the remote recovery of applications running in virtual machines.  Management framework plug-ins for Oracle Enterprise Manager and Microsoft System Center Operation Manager (SCOM) make it easy to monitor and manage storage from an infrastructure level.  These capabilities, along with a growing family of application-specific snapshot and cluster failover tools, can be used to simplify storage management using a single pane of glass that administrators are familiar with.

ESG Lab is pleased to report that the Engenio 2600, with a growing suite of application aware management interfaces, delivers balanced and predictable performance that is well suited for a mix of real-world business applications running in a VMware-enabled virtual server infrastructure.

Appendix

Virtual Machine and Drive Configuration

Engenio 2600 disk capacity was used for all storage capacity including VMware virtual disk files (VMDK), Windows Server 2008 R2 operating system images, application executables, and application data. The operating system images were installed on VMDK volumes. All of the application data volumes under test were configured as mapped raw LUNs (also known as raw device mapped, or RDM, volumes).

Application data and log volumes were configured as four drive RAID-1 volumes. Guest operating system volumes were configured using four-drive RAID-5 volumes. Volume ownership was balanced across the dual controllers and distributed evenly over the eight host interfaces. The volumes were spread evenly over two VMware host groups with a multipath policy of most recently used (MRU). The drive configuration is summarized in Table 5.

This is an example of the output created by the Jetstress utility. It shows the performance for one of four Jetstress tests running in parallel. Specifically, this report was created by the Jetstress utility running on a virtual machine within the fourth tile of the four tile test.

Microsoft Exchange Jetstress 2010

Performance Test Result Report

Test Summary

Database Sizing and Throughput

Jetstress System Parameters

Database Configuration

Transactional I/O Performance

Background Database Maintenance I/O Performance

Log Replication I/O Performance

Total I/O Performance

Host System Performance

11/19/2010 10:55:54 AM — Jetstress testing begins …
11/19/2010 10:55:54 AM — Prepare testing begins …
11/19/2010 10:55:56 AM — Attaching databases …
11/19/2010 10:55:56 AM — Prepare testing ends.
11/19/2010 10:55:56 AM — Dispatching transactions begins …
11/19/2010 10:55:56 AM — Database cache settings: (minimum: 32.0 MB, maximum: 256.0 MB)
11/19/2010 10:55:56 AM — Database flush thresholds: (start: 2.5 MB, stop: 5.1 MB)
11/19/2010 10:55:57 AM — Database read latency thresholds: (average: 20 msec/read, maximum: 100 msec/read).
11/19/2010 10:55:57 AM — Log write latency thresholds: (average: 10 msec/write, maximum: 100 msec/write).
11/19/2010 10:56:00 AM — Operation mix: Sessions 7, Inserts 40%, Deletes 20%, Replaces 5%, Reads 35%, Lazy Commits 70%.
11/19/2010 10:56:00 AM — Performance logging begins (interval: 15000 ms).
11/19/2010 10:56:00 AM — Attaining prerequisites:
11/19/2010 10:56:51 AM — \MSExchange Database(JetstressWin)\Database Cache Size, Last: 241778700.0 (lower bound: 241591900.0, upper bound: none)
11/19/2010 12:56:51 PM — Performance logging ends.
11/19/2010 12:56:51 PM — JetInterop batch transaction stats: 78235.
11/19/2010 12:56:51 PM — Dispatching transactions ends.
11/19/2010 12:56:51 PM — Shutting down databases …
11/19/2010 12:56:53 PM — Instance2308.1 (complete)
11/19/2010 12:56:53 PM — C:\JetStress-Results\Tile-4\Performance_2010_11_19_10_55_57.blg has 483 samples.
11/19/2010 12:56:53 PM — Creating test report …
11/19/2010 12:56:54 PM — Instance2308.1 has 12.0 for I/O Database Reads Average Latency.
11/19/2010 12:56:54 PM — Instance2308.1 has 1.2 for I/O Log Writes Average Latency.
11/19/2010 12:56:54 PM — Instance2308.1 has 1.2 for I/O Log Reads Average Latency.
11/19/2010 12:56:54 PM — Test has 0 Maximum Database Page Fault Stalls/sec.
11/19/2010 12:56:54 PM — Test has 0 Database Page Fault Stalls/sec samples higher than 0.
11/19/2010 12:56:54 PM — C:\JetStress-Results\Tile-4\Performance_2010_11_19_10_55_57.xml has 479 samples queried.

This is an example of the output created by the Oracle Orion utility for the database workloads. This example shows the performance of the four database VMs which ran in parallel during the mixed workload four tile test.

ORION VERSION 10.2.0.1.0

Commandline:

-run advanced -testname vmware -num_disks 5 -size_small 8 -size_large 1024 -type rand -simulate raid0 -write 30 -duration 150 -matrix basic

This maps to this test:

Test: vmware

Small IO size: 8 KB

Large IO size: 1024 KB

IO Types: Small Random IOs, Large Random IOs

Simulated Array Type: RAID 0

Stripe Depth: 1024 KB

Write: 30%

Cache Size: Not Entered

Duration for each Data Point: 150 seconds

Small Columns:,      0

Large Columns:,      0,      1,      2,      3,      4,      5,      6,      7,      8,      9,     10

Total Data Points: 36

Name: \\.\E:          Size: 1070596096

1 FILEs found.

Maximum Large MBPS=218.90 @ Small=0 and Large=10

Maximum Small IOPS=946 @ Small=25 and Large=0

Minimum Small Latency=5.13 @ Small=1 and Large=0

This is an example of the output created by the Iometer utility for the backup reader workload. This example shows the performance of the four backup jobs which ran in parallel during the mixed workload four tile test.

This is an example of the output created by the Iometer utility for the web server workload. This example shows the performance of the four web server VMs which ran in parallel during the mixed workload four tile test.

[2] Source: ESG Research, 2010 Server Virtualization Survey, September 2010.

[3] http://www.sourceforge.net/projects/iometer

[4] The configuration and methodology that was used during characterization testing is described in the Appendix.

[5] http://download.microsoft.com/download/D/1/B/D1BE3AEC-A9CD-4459-99F1-B28867FAE20B/Exchange2010TCP_parte8.pdf

[6] http://technet.microsoft.com/en-us/library/bb738152(EXCHG.80).aspx

[7] Current trends in Database Performance, Andrew Holdsworth, Oracle OpenWorld, November 2007. http://www.oracle.com/technology/deploy/performance/pdf/PerfTrends_Holdsworth.pdf

[8] Back of the Envelope Database Storage Design, Nitin Vengurlekar, RAC/ASM Development, Oracle Open World, November 2007. http://www.oracle.com/technology/products/database/asm/pdf/back%20of%20the%20env%20by%20nitin%20oow%202007.pdf

[9] See: ESG Lab Validation Report, Automated, Real-World Disaster Recovery Solutions, April 2009.

If you weren’t in NY or online or somehow haven’t heard, then here’s the summary: this past Tuesday there was a huge storage announcement multi-media event, that was decked in Guinness Book of Records bunting (the aforementioned Mini-packing, and a guy called Bubba jumping VMAX boxes on his Harley), and charmed by a grade-schooler (reports differ on whether he was 4th or 5th grade but he showed how easy it could be to provision and add storage)….and – somewhere in there – 41 new products and features were being launched by EMC on a miserable [weather] day in NY. The standout product items (just so I’m serving some sort of worthwhile news service here…) were the arrival of the VNX Family of products, the delivery of FAST VP (more granular tiering), and the addition of an archiving capability to the Data Domain stack. I’d go so far as to say those are in order of market importance, but I’m a hardware-geek so maybe there’s something else I’m overlooking, but this is also where a lot of the immediate chatter has been focused, so I guess I’m right.  OK, so that’s not 41 products and it may be that something else floated your boat, but I think it’s fair to say that these were the headliners. In expected EMC understatement this was paraded as the greatest storage announcement ever….in importance, it was clearly close on the heels of sliced bread, electricity, and the car (not just the Mini of course!)

But here’s the thing – there’s been more talk in the last 24-48 hours about the circus brouhaha than most anything else. I admit to falling into that in the above sentences.  But – heck – how many times am I going to write ‘Bubba’ in this blog?! Now, as we all slowly clear our heads and come down from the highs induced by all the smoke, mirrors, audio-volume, and hyperbole, it appears that the conversations are turning to the products. And that’s fine….but it also misses two most important points, I believe:

1. There’s been some excellent product and marketing counterpoints that I’ve seen from NetApp and HDS, amongst others. As you might expect, they are not joining in a chorus of approval, and are questioning, for example, whether this (VNX) is ‘real’ unified storage, and ’thanking’ EMC for following their leads into tiering (different approaches from the two of them but that’s for another day). By the way, the nature and culture of companies as expressed through their communications is fascinating: HDS has been considered and almost restrained at times; NetApp is more pugnacious and chest-thumping. Of course, both of them probably have a whole bunch of fair points to make; EMC did a sterling job, but even so I’ll bet there are points they skirted around or missed. However, none of that level of detail is really the big point. The big point is that, especially around the VNX announcement  (which ranges from small to SMB to lower enterprise, and offers unified plus easy-to-use storage)  I do think that EMC has – if nothing else – moved the conversation. And that’s not an illusion - it’s just the weight of a seasoned market leader driving change.
2. The other big news is EMC’s combo-deal of renewed enthusiasm for the channel, and the associated drive towards the smaller end of the business-storage-user. Please sit down before reading the next sentence – EMC talked about being a ‘price leader’ (!) which is akin to Sarah Palin coming out as a socialist. Of course, whereas the Palin move would be a huge reversal of an established position, one can argue that EMC’s move is simply a pragmatic extension of EMC’s desire to sell huge amounts of storage to wherever the biggest market opportunity lies. Every report and statistic I’ve seen says that the ‘lower end’ of the market is at the ‘higher’ end of growth and overall capacity shipments. In other words, it’s a major business opportunity. As a young salesman (for Memorex – RIP) in the UK in the 80′s I remember accounts of EMC offices having posters reading “our fair share is 100%”  – maturity and pragmatism may have softened the company’s target % a little, but not its ambition. You can argue (and all the competitors will) as to whether VNXe is the best/optimum/most thoughtful/suitable product out there… but the point is that it’s out there. And EMC seems to knows its intended purchasers well. For instance, ESG’s latest Spending Intentions Survey found that companies with less than 25 servers are the least likely to be adding new IT staff in 2011 – and therefore they will need simple, easy-to-manage solutions (so, no surprise, VNXe is wizard managed); the research also found that these users are more likely than any other group to be focused on data backup and recovery in 2011 – and so EMC is offering packaged software suites accordingly. Not that EMC is necessarily the first with all these things – but it sure is loud. Smart marketing.

And of course, marketing brings us back to the NY event. The glitz of spectacle yes, but the substance of specifications, too.  Both will be forgotten over time, but the essence of the event will be remembered. Is the industry really changed for ever? – probably not. But are the things we’ll all be discussing (and buying, from whomever) changed for the foreseeable future – almost certainly. If this industry were a school, then the gossip, grapevine and guesswork at Storage High is all-aflutter. Bubba, VNX, and that darned Mini certainly cheered things up on a wet January Tuesday, and for some time to come. Love it or hate it, a colleague here at ESG summarized the impact succinctly: people are talking about it all…so it’s mission accomplished.

You can read Mark’s other blog entries at The Business of Storage.

Here’s the scoop: the business intelligence solution addresses both analytics with PowerPivot and SQL Server 2008 R2 and collaboration for decision support processes with SharePoint 2010. HP and Microsoft are addressing small and mid-sized business Big Data needs with the HP Business Data Warehouse, optimized for SQL Server 2008 Parallel Data Warehouse leveraging multi-parallel processing (MPP) architecture with simplified administration.

Timing on this makes sense with Oracle Exadata and EMC’s Greenplum Data Compute Appliance targeting enterprise Big Data–less so for the smaller organizations with Big Data challenges and much smaller wallets.  End-users will be able to take advantage of a high performance data management machine using development tools and resources they may already own with SQL Server at an affordable price point.

After recently attending EMC’s Analyst Day where it announced the VNXe, it will only be a matter of time before more Big Data solutions are available and affordable for the little guys.

Data efficiency is just as it sounds–making the data we need more efficient from an access, usability, and manageability perspective.  Why?  So we can drive more value out of it–which I would argue is the entire raison d’être for IT to begin with.

We layer on server virtualization technologies to make our commodity server hardware more valuable –more readily usable by the various data elements we posses.

In the storage world, deduplication of data has been one of the hot efficiency “enablers,” along with thin provisioning, snapshots, virtualization, multi-tenancy, and compression.  Some of them are new, some have been around forever, it seems.  All of them are important.

When it comes to making data more efficient, it’s important to consider “Why” and not just how.  For example, most deduplication solutions are designed for backup–not primary use data environments.  I’m all for making backup more efficient, but that only represents a small fraction of the value potential for IT.  I would argue that primary storage/data use cases represent an order of magnitude (or more) value to an organization potentially.

We have done a decent job, over the last five or so years, at making our infrastructure systems that store/manage data far more efficient.  We can thin provision (virtualize) physical storage assets such that we get the most usage out of them.  We virtualize data (from a presentation perspective) with the use of snapshots.  With multi-tenancy, we can optimize the utilization of our physical assets across multiple constituents.  All that is good, but new technologies are allowing us to take this much further.

Compression has been around for a long time, but this is one technology going through a renaissance period.  Primary data compression (aka Real-Time) is going to change the fundamental efficiency and overall value proposition users will derive–because the closer to the point of data “creation” that you create efficiencies, the more value you get.

Think of it this way:  If you start with 100GB of primary data, over time you will back it up X times, so you’ll end up with 100GB of primary, and 100GBX of backup–or secondary–data.  Backup deduplication players such as Data Domain spend their time on the 100X problem (which is a good problem to spend time on, don’t misunderstand my point).  There are probably a lot of other uses/duplicates of the originating data throughout the organization between creation and backup–like in test/development, data warehouses, etc.

Optimizing data as early as possible is the key. From that point on, all the downstream benefits are magnified.  Less to move, less to manage, less to back up, less to copy, less to replicate, less to store, less to break.  Less is the new more.

I’m no genius but it seems to me that the way one does this most effectively is to leverage all of the tools at hand.  First, compress the data as much as you can.  We’ve proven that you can squish 50% + of the primary footprint out of almost any kind of data–including databases.  Second, deduplicate it.  Whatever is left post compression that you want deduped should be. People don’t want everything deduped, but that’s ok.  Eliminate what you want, and start clean. That gives you the perfect baseline.

From there snap it, thin provision it, copy it (virtually), and do whatever else you would like to do with it, but at least you begin the journey with an optimized footprint–which can only make everything else you do with that data far more efficient.

The trick is to compress in real-time, without suffering the performance penalties that you remember from 20 years ago.  It can be done today.  We do some amazing things at wire speed that weren’t possible even a few years ago.  There will be a lot of money and R&D spent in this area, as it seems clear to me that it simply has to happen.

Moving the value needle of storage/data optimization from the “dead” end of the wire to the “live” end (where data is created) is inevitably the best way to drive value all the way through the entire lifecycle of that data.  Everywhere that data sits, every time that data is manipulated or used, there is value that can be optimized.

Since I am always right eventually (seriously, it’s uncanny), this means there is a lot MORE money to made in this area.  Data Domain was just the beginning.

You can read Steve’s other blog entries at The Bigger Truth.