Manuals / Intel / Computer Equipment / Power Supply

Intel 7xx Servers, 170 Servers, AS/400 RISC Server manual 500, Parallel Virtual Tapes

Models: 7xx Servers 170 Servers AS/400 RISC Server

1 368

Download 368 pages 6.76 Kb

Page 257

15.13 Parallel Virtual Tapes

NOTE: Virtual tape is reading and writing to the same DASD so the maximum throughput with our concurrent and parallel measurements is different than our tape drive tests where we were reading from DASD and writing to tape.

								Parallel Virtual Tape for Large File
										570 16 way 128GB Memory
3500						800 DASD units for Virtual Tape Drives


3000


2500


2000
2000																					1 Virtual Tape Device
																					1 Virtual Tape Device
GB/HR																					2 Virtual Tape Devices



1500																					3 Virtual Tape Devices



1000																					4 Virtual Tape Devices


																					5 Virtual Tape Devices

500

0

SaveRestore

Operation

	Parallel Virtual Tape for User Mix
		570 16 way 128GB Memory
	800 DASD units for Virtual Tape Drives
	300
	250
	200	1 Virtual Tape Device
		1 Virtual Tape Device
GB/HR	150	2 Virtual Tape Devices
	100	3 Virtual Tape Devices
	100	4 Virtual Tape Devices
	50	5 Virtual Tape Devices
	50
	0	Restore
	Save	Restore
		Operation
IBM i 6.1 Performance Capabilities Reference - January/April/October 2008

© Copyright IBM Corp. 2008

Chapter 15. Save/Restore Performance

257

Image 257

Intel 7xx Servers, 170 Servers, AS/400 RISC Server manual 500, Parallel Virtual Tapes

Contents

January/April/October Toolcat Mkttools GET AS4ADV36 Package Table of Contents 107 Performance Observations and Tips117 System Application Server Instance 119157 10.2 DB2 for i5/OS access with Odbc 155158 11.1196 Supported Backup Device Rates 243199 200IPL Performance 19.6 User Pool Faulting Guidelines 310 19.7 17.7 File Level Backup Performance 292284 285334 Switchable IASP’s 327 Geographic Mirroring 329335 What the Estimator is Not 335Disclaimer Notice Special NoticesIpds Purpose of this Document Introduction Page Overview Interactive Indicators and MetricsISeries and AS/400 Risc Server Model Performance Behavior 3 V5R3 Disclaimer and Remaining Sections4 V5R2 and V5R1 Server Model Behavior V4R5 V5R2Page 100 Server Model Differences Custom Server Model Performance Highlights of Model 7xx Servers Model 7xx and 9/98 ModelPerformance Highlights of Model 170 Servers Additional Server Considerations Performance Highlights of Custom Server ModelsExample Interactive UtilizationServer Dynamic Tuning SDT Page Affects of Server Dynamic Tuning Managing Interactive Capacity Interactive/Server characteristics in the real worldIs Interactive really Interactive? Page Migration from Traditional Models Number of interactive transactions per hour Interval end time hour and minuteUpgrade Considerations for Interactive Capacity 13.2 V5R1 DSD Performance Behavior Interactive Processing Domino-Complementary ProcessingV5R1 Non-Domino Processing Running Linux on a Dedicated Server Logical Partitioning on a Dedicated ServerPage Effect of CPU Speed on Batch Batch PerformanceEffect of Dasd Type on Batch Batch Workload DescriptionExpert Cache Tuning Parameters for BatchJob Priority Dynamic Priority SchedulingPage I5/OS V6R1 SQE Query Coverage DB2 for i5/OS PerformanceNew for i5/OS V6R1 Query Attribute POWER5 Processor POWER6 Processor Records Inserted/Selected DB2 i5/OS V5R4 HighlightsI5/OS V5R4 SQE Query Coverage Common table expressions Deletes Derived tables I5/OS V5R3 HighlightsI5/OS V5R3 SQE Query Coverage I5/OS V5R3 SQE Performance Enhancements Like/Substring predicates DB2 Multisystem LOB columnsPartitioned Table Support SQE Optimizer V5R2 Highlights Introduction of the SQL Query EngineMaterialized Query Table Support Fast Delete Support SQE Database Statistics SQE Query EngineStatistic Usage Example Encoded Vector Indices EVIs IndexingSQE for V5R2 Summary DB2 for i5/OS Memory Sharing Considerations DB2 Symmetric Multiprocessing feature IntroductionDecision Support Queries SMP Performance SummaryRemote Journal Function Journaling and Commitment ControlJournaling Smapp System-Managed Access Path Protection SmappCommitment Control DB2 Multisystem for i5/OSCopyright IBM Corp Triggers Referential IntegrityDescription Variable Length FieldsPerformance Expectations Comparison to Normal Inserts Reuse Deleted Record SpaceDescription of Function Performance References for DB2 Copyright IBM Corp Communications Performance CCIN3 System i Ethernet SolutionsSoftware Communication Performance Test EnvironmentHardware Large data transfer FTP Communication and Storage observationsDasd Performance Streaming Performance TCP/IP non-secure performanceTCP/IP Secure Performance RR & Acrr Performance1167 565.4 530.0 479.6 462.1 202.2 SSL Performance249.7 53.4 48.0 31.3 27.4 478.4 55.7 53.3 36.9 31.9VPN Performance SSL Relative PerformancePerformance Observations and Tips VPN Relative Performance27.5 10.9 15.4 18.7 88.8Page APPC, ICF, CPI-C, and Anynet Page Speed = 10Mbps = 100Mbps HPR and Enterprise extender considerationsTest Type Hprip Link Hprip Link Speed AnyNet For network information select Networking Additional InformationWeb Server and WebSphere Performance Http Server powered by Apache Relative Performance Metrics Http Server powered by Apache for i5/OS V5R4 Relative Capacity for CGI Http Server powered by Apache for i5/OS Page Page Page Page PHP Zend Core for System ConfigurationZend Core for Database and Workload DescriptionOS / DB Database AccessDatabase Persistent versus Non-Persistent Connections Zend Platform Database Isolation LevelsPHP System Sizing What’s new in V6R1? WebSphere Application ServerThi ShoUsing direct map native Jdbc Tuning changes in V6R1Using ‘threadUsed=false custom property toolbox Jdbc Use of unix sockets toolbox JdbcPage Topology of the Trade Application Page WebSphere Application Server Trade Capacity Results Trade3 Measurement ResultsTrade 6 Primitives Trade Scalability ResultsDescription of Trade primitives in Figure WebSphere Trade 3 Primitives Accelerator for System Trade Jdbc Response Time 1 User Time seconds Page WebSphere Application Server V51 Express Moderate, and one complex. See .4, for IBM WebFacingWebFaced applications, one simple, one More in line with the other two machines Version 5.0 of WebfacingMany 5250 applications have been implemented With best performance techniques, such asDisplay File Record I/O Processing Tuning the Record Definition CacheCache Size Effect Cache Management Definition Cache Content ViewerDefinitions would be retained in the cache with few cache Misses for not commonly used recordsCache Viewer Button operations Cache Management Record Definition Loader Button OperationCompression Record Definition Loader Button operations OperationWebfacing Compression Enabling Compression in IBM Http Server powered by ApacheAdditional Resources PartnerWorld for Developers Webfacing website Hats Customization WebSphere Host Access Transformation Services HatsWhat’s new with V5R4 and Hats Hats Customization CPW/User IBM Systems Workload Estimator for Hats System Application Server InstancePage WebSphere Commerce WebSphere PortalWebSphere Commerce Payments Performance Tips and Techniques Connect for iSeriesPerformance Observations/Tips Connect for iSeries Connector TypesPage What’s new in V6R1 Java PerformanceIntroduction IBM Technology for Java 32-bit and 64-bit Garbage Collection Native CodeJIT Compiler Classic VM 64-bitPage Garbage Collection Bytecode Verification Functional Considerations Determining Which JVM to UsePerformance Considerations General Guidelines Capacity PlanningJava Performance Tips and Techniques Classic VM-specific Tips I5/OS Specific Java Tips and TechniquesJava Language Performance Tips Consider the special property os400.jit.mmi.thresholdMinimize synchronized methods Java i5/OS Database Access Tips Instead, the above procedure should be written asPool Database Connections Resources IBM Software Solutions System i Cryptographic SolutionsIBM Hardware Solutions Cryptography PerformanceCSP API Sets Cryptography Performance Test EnvironmentMeasurement Results Software Cryptographic API PerformanceBytes/Second Hardware Cryptographic API PerformanceEncryption Threads Cipher Encrypt Performance4764 Transactions/second Bytes/secondCCA CSP Cryptography Observations, Tips and RecommendationsSupported number of 4764 Cryptographic Coprocessors ISeries NetServer File Serving Performance ISeries NetServer File Serving PerformanceConclusion/Explanations Response TimePage DB2 for i5/OS Jdbc and Odbc Performance 10.1 DB2 for i5/OS access with JdbcJdbc Performance Tuning Tips References for Jdbc 10.2 DB2 for i5/OS access with Odbc Odbc Performance Settings Page Microsoft Odbc webpage References for OdbcDomino on Mail and Calendaring Users MCU Domino Workload DescriptionsDomino Domino Web Access formerly known as iNotes Web AccessDomino Domain Monitor Additional memory was added for this test Domino Web Access client improvements with DominoResponse Time and Megahertz relationship Domino Edition Collaboration Edition and Domino Edition offeringsCollaboration Edition Performance Tips / Techniques Page Page Domino Subsystem Tuning Domino Web AccessMain Storage Options Performance Monitoring StatisticsMinimize Thousands V5R3 Main Storage Options Response Times Sizing Domino on SystemPerformance Data Collection Tools Lpar and Partial Processor ConsiderationsSystem i NotesBench Audits and Benchmarks Performance Improvements for WebSphere MQ V5.3 CSD6 WebSphere MQ for iSeriesQueue Manager Start Following an Abnormal End Checkpoint Following a Journal Receiver Roll-overConclusions, Recommendations and Tips Test Description and ResultsOther Sources of Information Linux on iSeries Performance Basic Requirements -- Where Linux RunsSummary Key IdeasLinux on iSeries Architecture Linux on iSeries Technical OverviewLinux on iSeries Run-time Support Basic Configuration and Performance Questions Computational Performance -- C-based code General Performance Information and ResultsWeb Serving Performance Number Processors PartitionComputational Performance -- Java Network Operations TCP/IP Function Megabit Ethernet LAN Virtual LANGcc and High Optimization gcc compiler option -O3 Gcc Compiler, VersionVirtual Disk Value of Virtual LAN and Virtual Disk13.7 DB2 UDB for Linux on iSeries Top Tips for Linux on iSeries Performance Linux on iSeries and IBM eServer Workload EstimatorPage Page Page Internal Native Attachment Dasd PerformanceDevices & Controllers Direct Attach Native Hardware CharacteristicsIntensive Workload Performance Comparison 14.1.2 iV5R2 Direct Attach DasdIntensive Workload 14.1.3.2 571B IOP vs Iopless 10 15K 35GB Dasd 14.1.3 571B14.1.3.1 571B RAID5 vs RAID6 10 15K 35GB Dasd Dasd 14.1.4 571B, 5709, 573D, 5703, 2780 IOA Comparison Chart14.1.4.4 14.1.5.1 14.1.6.2 Comparing 571E/574F and 571F/575B IOP and IOPLess1000 2000 3000 4000 Ops/Sec 14.1.7.4 14.1.8.1 Performance Limits on the 571F/575BPage 14.1.9.2 14.1.9.1 Direct Attach 571E/574F and 571F/575B Observations 100 200 300 400 500 600 Ops/Sec New in iV5R4M5 14.2.1 9406-MMA CEC vs 9406-570 CEC DasdRAID Hot Spare 14.2.3 12X Loop Testing Non Encrypted ASP vs Encrypted ASP New in iV6R1M0 Encrypted ASPNon Encrypted ASP vs Encrypted ASP 100 350 600 850 1100 1350 1600 1850 14.3.2 57B8/57B7 IOAPOWER6 520 57B8/57B7 6 RAID5 Dasd in CEC 14.3.3 572A IOA Page SAN Storage Area Network External 14.5 iV6R1M0 -- Vios and IVM Considerations Generic Concepts General Vios ConsiderationsGeneric Configuration Concepts Page Page Chdev -dev hdisk03 -attr pv=yes will assign a Pvid to hdisk3 Page Vios and JS12 Express and JS22 Express Considerations 001 10000 20000 30000 40000 50000 60000 Vios CPU Vios Average Dasd Response Time Seconds BladeCenter S and JS12 Express Page Page System Workload Response Time Seconds IBM i operating system 5.4 Virtual Scsi Performance Page Introduction Virtual Scsi Performance Examples Virtual Scsi Bandwidth-Multiple Network Storage Spaces Native vs. Virtual PerformanceNwsd Read Scaling Virtual Scsi Bandwidth-Disk Scaling Sizing Sizing when using Dedicated ProcessorsCPU milliseconds to process virtual Scsi I/O transaction Sizing when using Micro-Partitioning Disk 1GB Sequential Read Sizing memoryAIX Virtual IO Client Performance Guide Supported Backup Device Rates Save/Restore PerformanceUse Optimum Block Size Useoptblk Save Command Parameters that Affect PerformanceData Compression Dtacpr Data Compaction CompactDatabase File related Workloads WorkloadsUser Mix User Mix 3GB contains 12,300 objectsComparing Performance Data Ultra High Performing Backup Devices Lower Performing Backup DevicesMedium & High Performing Backup Devices Use of Multiple Backup Devices Parallel and Concurrent Library Measurements Large File Concurrent Save and Restore RatesLarge File Parallel Save and Restore RatesUser Mix Concurrent User Mix Concurrent RunsNumber of Processors Affect Performance 450 440 430 420100 Dasd and Backup Devices Sharing a TowerVirtual Tape 200Data To Save Hours Parallel Virtual Tapes 500Concurrent Virtual Tapes 1500 1000Large File Virtual Tape Scaling Save Write to Virtual tape Save and Restore Scaling using a Virtual Tape DriveLarge File Save User Mix Saves PCI-X ™ In a 5094/5294 tower use slot C08 or C09 High-End Tape Placement on System700 250 SLR60 15.19 5XX Tape Device RatesMeasurements in GB/HR all 8 Dasd in the system ASP Source File 1GB 15.21 5XX DVD RAM and Optical Library High Software CompressionASP 15.23 9406-MMA DVD RAM1GB 15.24 9406-MMA 576B IOPLess IOAIV5R4M5 July What’s New and Tips on Performance What’s New IV6R1M0 MarchIV5R4 January AugustIPL Test Description IPL PerformanceIPL Performance Considerations Active Database Software Configuration16.3 9406-MMA System Hardware Information 16.4 9406-MMA IPL Performance Measurements Normal 1 iV5R4M5 Normal IPL Power-On Cold Start16.5 9406-MMA IPL Performance Measurements Abnormal 1 iV5R4M5 Abnormal IPL Partition MSDMSD Affects on IPL Performance Measurements 16.7.2 5XX Large system Hardware Configuration 16.7.1 5XX Small system Hardware Configuration16.7 5XX System Hardware Information Database16.8 5XX IPL Performance Measurements Normal 1 Normal IPL Power-On Cold Start16.9 5XX IPL Performance Measurements Abnormal Measurement units are in hours, minutes and secondsIPL Tips 16.10 5XX IOP vs IOPLess effects on IPL Performance NormalIntegrated xSeries Adapter IXA Integrated BladeCenter and System x PerformanceV5R4 iSCSI Host Bus Adapter iSCSI HBA 17.2.1 IXS/IXA Disk I/O Operations Effects of Windows and Linux loads on the host systemIntegrated xSeries Servers IXS ISCSI Disk I/O Operations Extended Write OperationsCrtsbsd SBSDQGPL/QFPHIS POOLS1 10000 ISCSI virtual I/O private memory pool17.2.5 IXS/IXA IOP Resource Virtual Ethernet ConnectionsISCSI attached servers IXS and IXA attached serversMachine Pool MBytes Base Pool Qfphis Private Pool Total Disk I/O CPU CostCPW per 1k Disk Operations Further notes about IXS/IXA Disk OperationsISCSI Target, IXA Capacity Comparison MB per Second Disk I/O ThroughputVE Capacity Comparisons Virtual Ethernet CPU Cost and CapacitiesPP TCP Stream i5/OS to Windows VE CPW CostWindows CPU Cost File Level Backup PerformanceFlbu SAV / RST Rates Additional Sources of InformationMicrosoft Hardware Compatibility Test URL See V5R3 Information Logical Partitioning LparV5R2 Additions General TipsV5R1 Additions ConsiderationsPerformance on a 12-way system Lpar Performance ConsiderationsPage Lpar Throughput Increase Lpar Measurements 5700 21 % Over a Standalone 12-way5920 26 % SAP Performance Information Miscellaneous Performance InformationTPC-C Commercial Performance VolanoMark SPECjbb2000Dynamic Priority Scheduling Delay Cost TerminologyPriority Mapping to Delay Cost Curves Qdynptyscd = ‘1’ on Qdynptyscd = ‘0’ Performance Testing ResultsMain Storage Sizing Guidelines Memory Tuning Using the Qpfradj System ValueAdditional Memory Tuning Techniques Large Memory Systems User Pool Faulting GuidelinesSetobjacc Set Object Access Batch Interactive19.7 AS/400 NetFinity Capacity Planning AS/400 NetFinity Software Inventory Performance Conclusions/Recommendations for NetFinity General Performance Tips and Techniques Adjusting Your Performance Tuning for ThreadsHistory Coming ChangeSolution ProblemOptimization Levels General Performance Guidelines -- Effects of CompilationIn-lining Theory -- and Practice Typical Storage Costs System Level ConsiderationsBrief Example Order1 Orderj Ordert OrderNShort but Important Tip about Data Base Which is more important?Final Thought About Memory and Competitiveness HMT Described Hardware Multi-threading HMTHMT and SMT Compared and Contrasted HMT Feature SMT FeatureSome key similarities and differences are Models With/Without HMTPOWER6 520 Memory Considerations Aligning Floating Point Data on Power6 Page Switchable IASP’s High Availability Switchable Resources ConsiderationsHigh Availability Performance Workload DescriptionSystem Configuration Workload ConfigurationHardware Configuration Cabling MapGeographic Mirroring Switchover TipsWorkload Description Hardware Configuration Cabling Map Large System ConfigurationGeographic Mirroring Measurements Switchable Towers using Geographic MirroringSynchronization on an idle system Active StateGeographic Mirroring Tips IBM Systems Workload Estimator What the Estimator is Not Estimator AccessMerging PM for System i data into the Estimator Page Commercial Processing Workload CPW Appendix A. CPW and CIW DescriptionsCPW Application Description Compute Intensive Workload CIW Page Page Starting Collection Services Performance Data Collection ServicesHow Collection Services works Batch Modeling Tool Bchmdl Page Page V6R1 Additions October CPW values for the IBM Power Systems IBM i operating systemV6R1 Additions April CPW values for the IBM Power 595 IBM i operating systemV6R1 Additions August V6R1 Additions January Table C.3.1. CPW values for Power System ModelsIBM System i using the POWER5 processor technology IBM System i using the POWER6 processor technologyV5R4 Additions July V5R4 Additions January/May/August 2006 and January/AprilModel Edition Accelerator IBM ~ i5 Servers V5R3 Additions May, July, August, October 2004, JulyValue ExpressCPU ISeries Model 8xx Servers V5R2 Additions February, May, July750 540V5R2 Additions Model 810 and 825 iSeries for Domino FebruaryModels share L2 cache between 2 processors Standard Models 8xx ServersTable C.9.2.1 Standard Models 8xx Servers 10 V5R1 AdditionsTable C.9.1.1 Model 8xx Servers Model 8xx Servers10.3 V5R1 Dedicated Server for Domino Model 2xx ServersTable C.10.2.1 Model 2xx Servers Table C.10.3 .1 Dedicated Servers for DominoActive Stand-by 840-2352 840-2353 840-2354 Capacity Upgrade on-demand ModelsCPW Values and Interactive Features for CUoD Models 16500 Table C.10.4.1.1 V5R1 Capacity Upgrade on-demand ModelsTable C.11.1 Model 8xx Servers All new Condor models 11 V4R5 Additions11.1 AS/400e Model 8xx Servers Table C.11.3.1 Dedicated Server for Domino Dedicated Server for DominoTable C.11.2.1 Model 2xx Servers 12 V4R4 Additions SB Models12.1 AS/400e Model 7xx Servers Table C.11.4.1 SB ModelsTable C.12.2.1 Current Model 170 Servers Model 170 Servers Current 170 ServersPrevious Model 170 Servers AS/400e Dedicated Server for Domino2407 2408 2409 120 Table C.12.2.2 Dedicated Server for Domino14 AS/400e Custom Servers 13 AS/400e Model Sxx Servers15 AS/400 Advanced Servers Table C.13.1 AS/400e ServersTable C.15.2 AS/400 Advanced Servers V3R7 16 AS/400e Custom Application Server Model SB1Table C.15.1 AS/400 Advanced Servers V4R1 and V4R2 Table C.16.1 AS/400e Custom Application Server Model SB1 17 AS/400 Models 4xx, 5xx and 6xx SystemsTable C.17.1 AS/400 Risc Systems Table C.17.2 AS/400e Systems18 AS/400 Cisc Model Capacities Table C.18.8 AS/400 Advanced Servers Cisc Table C.18.6 AS/400 Cisc Model 9406 SystemsTable C.18.7 AS/400 Advanced Systems Cisc