Storage Performance Council (SPC) www.storageperformance.org spcadmin@storageperformance.org Presentations and Speakers An Introduction to the SPC Leah Schoeb (Sun Microsystems, Inc.) SPC-1C Preview Walter E. Baker (SPC Administrator & Auditor) SPC-3 Development Status Mel Boksenbaum (Hitachi Data Systems) SPC-2 Announcement Bruce McNutt (IBM Corporation & SPC-2 Chair) SPC-2 “Wave One” Results Walter E. Baker (SPC Administrator & Auditor) Q&A Sandy Wilson (Fujitsu & SPC Steering Committee Chair) CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 2 An Introduction to the SPC Leah Schoeb Sun Microsystems, Inc. CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 3 Council Mission The SPC is a non-profit corporation founded in 1998 to define, standardize, and promote the first industry standard storage benchmarks to disseminate objective, verifiable performance data to developers, product managers, and customers of computer systems CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 4 SPC Objectives Drive performance improvement in the computer systems industry for storage Insure that customers can accurately compare products in a multi-vendor world Establish a level “playing Storage Industry field” for all manufacturers Performance Experts Results Dissemination Publicize benchmark results Service and Support Ensure accuracy and authenticity of published results CMG 2005 Storage Performance Council (SPC) Test Tools Benchmark Kits Performance Analysis Tools www.StoragePerformance.org New Benchmark Development SPC Benchmark Maintenance Results Validation Peer Review Audit Page: 5 SPC Organizational Structure Membership Full Members Board of Directors - Elects Steering Committee - Approves Standards - Reviews Compliance Associate Members Benchmark Subcommittees Steering Committee-Management -Direction - Administration CMG 2005 Storage Performance Council (SPC) Administrator Standing Committees www.StoragePerformance.org Academic Members No Cost Page: 6 SPC Membership CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 7 SPC Benchmarks Application Environment SPC-1 SPC-x SPC-2 Database LDQ CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Applications are clustered based on characteristics of load presented to the I/O Subsystem SPC-x benchmark will be defined for each cluster, starting with SPC-1 Page: 8 SPC-1C Preview Walter E. Baker Gradient Systems, Inc. SPC Administrator and Auditor CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 9 SPC-1 and storage components There is a distinct need to provide product performance comparison for storage configuration components such as disk drives, HBAs/controllers, small storage subsystems (single enclosure), etc. SPC-1 provides that type of product performance comparison, but is typically used for large, complex storage configurations CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 10 SPC-1C (Component) SPC-1C will be the first industry-standard performance benchmark to provide product performance comparisons for individual storage components such as disks, HBAs/controllers, small storage subsystems (single enclosure), storage software, etc. SPC-1C will be identical to SPC-1 in workload and reporting requirements. SPC-1C will have reduced audit and filing fees. CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 11 SPC-1C Producer Audience Storage component vendors comprise the primary producer audience for SPC-1C. SPC-1C will provide performance evaluation and comparison for storage components such as: • Disk drives • HBAs/controllers • Small Storage Subsystems (single enclosure) • Processors used in the above components • Storage software such as Logical Volume Managers CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 12 SPC-1C Consumer Audiences Storage solution providers (vendors, resellers, etc.), as well as end users, are the primary consumer audiences for SPC-1C. SPC-1C will allow storage solution providers to evaluate the storage components that comprise the larger storage solution offered by the provider. • Disks used in an array • HBAs/Controllers • Small Storage Subsystems • Logical Volume Managers CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 13 SPC-1C Configuration Components SPC-1C configurations will consist of the following two components: • One HBA/Controller (single/dual/quad…). • One of the following disk drive configurations: – One, two, or four disks in a “standalone” configuration. » An external enclosure may be used but only to provide power and/or connectivity for the disks. – A small storage subsystem (any form factor) with a maximum of sixteen disks. CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 14 SPC-1C Availability Test An optional SPC-1C Availability Test will replace the required SPC-1 Persistence Test. Applicable in SPC-1C configurations that provide uninterrupted operation and “hot” rebuild/recovery in the case of a single physical drive failure. CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 15 SPC-1C Availability Test The SPC-1C Availability Test will report: • If processing continues in the case of a single physical drive failure. • The level of performance when operating in “degraded mode” after the failure. • The time required to recover from the failure. • The level of performance during the recovery. CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 16 SPC-3 Development Status Mel Boksenbaum Hitachi Data Systems CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 17 Description of effort A protocol independent benchmark that measures storage performance as seen by applications SPC-1 and SPC-2 characterize lower level components in the storage hierarchy • Storage from the bottom up SPC-3 will take the end-user view of the virtualized storage environment. • Storage from the top down CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 18 Environments and workloads File Services Expansion/Business Use Traversal/Backup Database OLTP Decision Support Workflow CMG 2005 Storage Performance Council (SPC) Refinement/Digital Film Editing Traversal/Playback www.StoragePerformance.org Page: 19 Critical Questions Draft Workload Definition available www.storageperformance.org Is a measurement at the application level useful? What factors make it interesting? Scale? Volume, users, bandwidth Constraints? Price, latency, protocol Topology? NAS, SAN, DAS Your input is invited and eagerly awaited! spc3-input@storageperformance.org CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 20 SPC-2 Announcement Bruce McNutt IBM Corporation Chair, SPC-2 Technical Subcommittee CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 21 Announcing SPC-2 (Dec 6, 2005) New test of storage system performance for environments that require large scale data movement. Built using a series of application-oriented sequential workloads. Audited results will be posted today on the SPC web site for 6 tested storage products. Available now to SPC members/associate members. CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 22 SPC-1: Predominately random access Represents environments such as OLTP, database, and mail server: CMG 2005 Storage Performance Council (SPC) Read Write All (pct) (pct) (pct) Random 29 32 61 Sequential 11 28 39 All 40 60 100 www.StoragePerformance.org Page: 23 SPC-2: Large Block, Sequential I/O Sequential tests: valuable to many customers Customers with dedicated servers. • Data warehouse & Data mining • Scientific Processing • Media Customers with a mix of applications including sequential. • Banking & Financial, Insurance, Manufacturing, Medical, Government, many others… • To such customers, the sequential component of performance is often the most visible. CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 24 SPC-2: A collection of sequential tests Why a collection? • Types of sequential processing can be very different. • It is easy to include a variety of tests. • Since the hardware imposes a ceiling on all data rate measurements, a variety of measurements will tend to reinforce each other. CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 25 Sequential processing tested in SPC-2 Large file processing: simple sequential processing of one or more large files. Large database queries: scans or joins of large relational tables. Video on demand: individualized playback from a digital film library. CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 26 SPC-2 measurements (Example) Large File Processing - Data Rate Large Database Query - Data Rate 5,000 5,000 4,500 Data Rate, MB/sec Data Rate, MB/sec 4,000 3,500 3,000 2,500 2,000 1,500 1,000 500 0 Write 256KiB Read/Write Read 256KiB 256KiB 1 Stream 12 Streams Write 1024KiB 24 Streams Read/Write 1024KiB 48 Streams 4,500 4,000 3,500 3,000 2,500 2,000 1,500 1,000 500 0 64KiB w/ 1 IO/Stream Read 1024KiB 1 Stream 96 Streams 64KiB w/ 4 IOs/Stream 12 Streams 1024KiB w/ 1 IO/Stream 24 Streams 48 Streams 1024KiB w/ 4 IOs/Stream 96 Streams 6,000 4,000 2,000 MI 4400 Streams 0 0:00 0:05 0:10 0:15 0:20 0:25 0:30 0:35 0:40 0:45 0:50 0:55 1:00 1:05 1:10 1:15 1:20 1:25 1:30 1:35 1:40 1:45 1:50 1:55 2:00 2:05 2:10 2:15 2:20 Data Rate, MB/sec Data Rate -- Video On Demand Time within Test Run, min CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 27 Watch This Space Additional results are expected to accumulate rapidly Watch for them on the SPC Web site: • http://www.StoragePerformance.org Aiming for release to non-SPC members in 1H2006 CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 28 SPC-2 “Wave One” Results Walter E. Baker Gradient Systems, Inc. SPC Administrator and Auditor CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 29 SPC-2 “Wave One” Results SPC-2 “Wave One” results submitted by four member companies: • Fujitsu Limited • Hewlett-Packard Company • IBM Corporation • Sun Microsystems, Inc. CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 30 SPC-2 “Wave One” Results Six storage products: • Fujitsu Storage Systems ETERNUS 6000 Model 900 • HP StorageWorks 8000 Enterprise Virtual Array • IBM TotalStorage SAN Volume Controller 3.1 • IBM TotalStorage DS8300 • Sun StorEdge 6130 Array • Sun StorEdge 3510 FC Array CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 31 SPC-2 Reported Data SPC-2 Reported Data Priced Storage Configuration Name: SPC-2 MBPS™ SPC-2 Price-Performance Total Price ASU Capacity Data Protection Level The above SPC-2 MBPS™ value represents the aggregate data rate of all three SPC-2 workloads: Large File Processing (LFP), Large Database Query (LDQ), and Video On Demand (VOD) SPC-2 Large File Processing (LFP) Reported Data Number of Data Rate Data Rate Streams per Stream Price-Performance LFP Composite Write Only: 1024 KiB Transfer 256 KiB Transfer Read-Write: 1024 KiB Transfer 256 KiB Transfer Read Only: 1024 KiB Transfer 256 KiB Transfer The above SPC-2 Data Rate value for LFP Composite represents the aggregate performance of all three LFP Test Phases: (Write Only, Read-Write, and Read Only). SPC-2 Large Database Query (LDQ) Reported Data Number of Data Rate Data Rate Streams per Stream Price-Performance LDQ Composite 1024 KiB Transfer Size 4 I/Os Outstanding 1 I/O Outstanding 64 KiB Transfer Size 4 I/Os Outstanding 1 I/O Outstanding The above SPC-2 Data Rate value for LDQ Composite represents the aggregate performance of the two LDQ Test Phases: (1024 KiB and 64 KiB Transfer Sizes). SPC-2 Video On Demand (VOD) Reported Data Number of Data Rate Data Rate Streams per Stream CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Price-Performance Page: 32 SPC-2 Reported Data Overall Composite: • Large File Processing (LFP) • Large Database Query (LDQ) • Video on Demand Delivery (VOD) SPC-2 Reported Data Priced Storage Configuration Name: SPC-2 MBPS™ SPC-2 Price-Performance Total Price ASU Capacity Data Protection Level The above SPC-2 MBPS™ value represents the aggregate data rate of all three SPC-2 workloads: Large File Processing (LFP), Large Database Query (LDQ), and Video On Demand (VOD) CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 33 SPC-2 Reported Data SPC-2 Reported Data Priced Storage Configuration Name: SPC-2 MBPS™ SPC-2 Price-Performance Total Price ASU Capacity Data Protection Level The above SPC-2 MBPS™ value represents the aggregate data rate of all three SPC-2 workloads: Large File Processing (LFP), Large Database Query (LDQ), and Video On Demand (VOD) • SPC-2 MBPS™: – 263.01 / 343.72 / 1,025.62 / 2,129.47 / 3,217.53 / 3,517.75 • SPC-2 Price-Performance ($ per SPC-2 MBPS™): – $137.52 / $201.41 / $384.31 / $415.37 / $563.93 / $679.23 • Range of disk devices (spindles) configured: – 14 / 36 / 112 / 280 / 512 / 672 CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 34 SPC-2 Reported Data Large File Processing (LFP) SPC-2 Large File Processing (LFP) Reported Data Number of Data Rate Data Rate Streams per Stream Price-Performance LFP Composite Write Only: 1024 KiB Transfer 256 KiB Transfer Read-Write: 1024 KiB Transfer 256 KiB Transfer Read Only: 1024 KiB Transfer 256 KiB Transfer The above SPC-2 Data Rate value for LFP Composite represents the aggregate performance of all three LFP Test Phases: (Write Only, Read-Write, and Read Only). • SPC-2 LFP Composite Data Rate (MB/s): – 221.66 / 248.33 / 833.19 / 1,875.53 / 2,573.57 / 2,707.28 • SPC-2 Price-Performance ($ per MB/s): – $163.19 / $278.77 / $471.61 / $473.06 / $732.76 / $849.18 CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 35 SPC-2 Reported Data Large Database Query (LDQ) SPC-2 Large Database Query (LDQ) Reported Data Number of Data Rate Data Rate Streams per Stream Price-Performance LDQ Composite 1024 KiB Transfer Size 4 I/Os Outstanding 1 I/O Outstanding 64 KiB Transfer Size 4 I/Os Outstanding 1 I/O Outstanding The above SPC-2 Data Rate value for LDQ Composite represents the aggregate performance of the two LDQ Test Phases: (1024 KiB and 64 KiB Transfer Sizes). • SPC-2 LDQ Composite Data Rate (MB/s): – 323.60 / 389.60 / 1,118.66 / 2,273.09 / 3,697.33 / 4,389.76 • SPC-2 Price-Performance ($/MB/s): – $111.78 / $177.69 / $352.34 / $394.92 / $452.32 / $591.08 CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 36 SPC-2 Reported Data Video on Demand Delivery (VOD) SPC-2 Video On Demand (VOD) Reported Data Number of Data Rate Data Rate Streams per Stream Price-Performance • Number of Streams: – 310 / 500 / 1,500 / 2,848 / 4,300 / 4,400 • SPC-2 VOD Data Rate (MB/s): – 243.78 / 393.22 / 1,124.99 / 2,239.77 / 3,381.70 / 3,460.23 • SPC-2 Price-Performance ($/MBs): – $148.37 / $176.06 / $350.36 / $394.92 / $573.31 / $646.25 CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 37 SPC Q&A Sandy Wilson Fujitsu Computer Systems Corporation Chair, SPC Steering Committee CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 38 SPC Q&A Why are there not more SPC member companies releasing SPC-2 results for this announcement? What type of testing was done, prior this initial release of the benchmark specification? Why is the benchmark only available to SPC members and not to the general public? CMG 2005 Storage Performance Council (SPC) www.StoragePerformance.org Page: 39