Hitachi Virtual Storage Platform

Transcription

1 Hitachi Virtual Storage Platform Performance Guide FASTFIND LINKS Document Organization Product Version Getting Help Contents MK-90RD

2 Hitachi Ltd. All rights reserved. No part of this publication may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopying and recording, or stored in a database or retrieval system for any purpose without the express written permission of Hitachi, Ltd. (hereinafter referred to as "Hitachi"), and Hitachi Data Systems Corporation (hereinafter referred to as "Hitachi Data Systems"). Hitachi and Hitachi Data Systems reserve the right to make changes to this document at any time without notice and assume no responsibility for its use. This document contains the most current information available at the time of publication. When new or revised information becomes available, this entire document will be updated and distributed to all registered users. All of the features described in this document may not be currently available. Refer to the most recent product announcement or contact your local Hitachi Data Systems sales office for information about feature and product availability. Notice: Hitachi Data Systems products and services can be ordered only under the terms and conditions of Hitachi Data Systems' applicable agreements. The use of Hitachi Data Systems products is governed by the terms of your agreements with Hitachi Data Systems. Hitachi is a registered trademark of Hitachi, Ltd., in the United States and other countries. Hitachi Data Systems is a registered trademark and service mark of Hitachi in the United States and other countries. ShadowImage and TrueCopy are registered trademarks of Hitachi Data Systems. AIX, FICON, FlashCopy, IBM, MVS/ESA, MVS/XA, OS/390, S/390, VM/ESA, VSE/ESA, z/os, zseries, z/vm, and zvse are registered trademarks or trademarks of International Business Machines Corporation. All other trademarks, service marks, and company names are properties of their respective owners. Microsoft product screen shots reprinted with permission from Microsoft Corporation. ii

3 Contents Preface xi Intended audience xii Product version xii Document revision level xii Changes in this revision xii Referenced documents xiii Document organization xiii Document conventions xiv Convention for storage capacity values xv Accessing product documentation xvi Getting help xvi Comments xvi 1 Performance overview Hitachi Performance Monitor overview Server Priority Manager overview Performance of high-priority hosts Upper-limit control Threshold control Cache Residency Manager overview Prestaging data in cache Priority mode (read data only) Bind mode (read and write data) Virtual Partition Manager overview Interoperability of Performance Monitor and other products Cautions and restrictions for monitoring Cautions and restrictions for usage statistics Using Server Priority Manager Contents iii

4 3 Monitoring WWNs Viewing the WWNs that are being monitored Adding new WWNs to monitor Removing WWNs to monitor Adding WWNs to ports Editing the WWN nickname Connecting WWNs to ports Deleting unused WWNs from monitoring targets Monitoring CUs Displaying CUs to monitor Adding and removing CUs to monitor Confirming the status of CUs to monitor Monitoring operation Performing monitoring operations Starting monitoring Stopping monitoring Setting statistical storage ranges About statistical storage ranges Viewing statistics Setting the storing period of statistics Working with graphs Basic operation Objects that can be displayed in graphs Usage rates of MPs Usage rate of a data recovery and reconstruction processor Usage rate of cache memory Write pending statistics Access paths usage statistics Throughput of storage system Size of data transferred Response times Cache hit rates Back-end performance Hard disk drive usage statistics Hard disk drive access rates ShadowImage usage statistics Detailed information of resources on top 20 usage rates iv Contents

5 8 Changing display of graphs Graph operation Changing displayed items Changing a display period Adding a new graph Deleting graph panel Server Priority Manager operations Overview of Server Priority Manager operations If one-to-one connections link HBAs and ports If many-to-many connections link HBAs and ports Port tab operations Analyzing traffic statistics Setting priority for ports on the storage system Setting upper-limit values to traffic at non-prioritized ports Setting a threshold WWN tab operations Monitoring all traffic between HBAs and ports Excluding traffic between a host bus adapter and a port from the monitoring target Analyzing traffic statistics Setting priority for host bus adapters Setting upper-limit values for non-prioritized WWNs Setting a threshold Changing the SPM name of a host bus adapter Registering a replacement host bus adapter Grouping host bus adapters Containing multiple HBAs in an SPM group Deleting an HBA from an SPM group Switching priority of an SPM group Setting an upper-limit value to HBAs in an SPM group Renaming an SPM group Deleting an SPM group Creating virtual cache partitions Cache Logical Partition definition Purpose of Cache Logical Partitions Corporate use example Best practices for cache partition planning Minimum software requirements for cache partitions Default CLPR names Hardware best practices Cache Logical Partition workflow Calculating cache capacity Cache capacity without specialized applications Contents v

6 Formula to size VOL capacity of internal storage Formula to size VOL capacity of external storage Formula to size VOL capacity of Dynamic Provisioning or Dynamic Tiering Cache capacity with Dynamic Provisioning or Dynamic Tiering Cache capacity with Cache Residency Manager Cache capacity with Extended Remote Copy (XRC) for Mainframe Cache capacity with Universal Volume Manager Adjusting the cache capacity of a CLPR Creating a CLPR Migrating resources to and from a CLPR Deleting a CLPR Troubleshooting Virtual Partition Manager Estimating cache size About cache size Calculating cache size for open systems Calculating cache size for mainframe systems Cache Residency Manager cache areas Cache Residency Manager system specifications Managing resident cache Cache Residency Manager rules, restrictions, and guidelines Launching Cache Residency Viewing Cache Residency information Placing specific data into Cache Residency Manager cache Placing LDEVs into Cache Residency Manager cache Releasing specific data from Cache Residency Manager cache Releasing LDEVs from Cache Residency Manager cache Changing mode after Cache Residency is registered in cache Troubleshooting Troubleshooting resources Calling Hitachi Data Systems Support Center A Export Tool A-1 About the Export Tool A-2 Installing the Export Tool A-2 System requirements A-3 Installing the Export Tool on a Windows system A-3 Installing the Export Tool on a UNIX system A-4 Using the Export Tool A-4 Preparing a command file A-4 vi Contents

7 Preparing a batch file A-8 Running the Export Tool A-9 File formats A-10 Processing time A-10 Termination code A-11 Log files A-12 Error handling A-12 Export Tool command reference A-13 Export Tool command syntax A-13 Conventions A-13 Syntax descriptions A-14 Writing a script in the command file A-14 Viewing the online Help for subcommands A-14 Subcommand list A-15 svpip A-15 retry A-16 login A-16 show A-17 group A-18 Short-range A-34 long-range A-37 outpath A-39 option A-40 apply A-41 set A-41 help A-42 Java A-43 Exported files A-45 Monitoring data exported by the Export Tool A-45 Resource usage and write-pending rate statistics A-47 Parity groups, external volume groups, or V-VOL groups statistics A-51 Volumes in parity/external volume groups or V-VOL groups statistics.... A-52 Volumes in parity groups, external volume groups, or V-VOL groups (at volumes controlled by a particular CU) A-54 Port statistics A-56 Host bus adapters connected to ports statistics A-57 Volumes (LU) statistics A-57 All host bus adapters connected to ports A-58 MP blades A-59 Remote copy operations by TC/TCz (whole volumes) A-60 Remote copy operations by TC and TCz (for each volume (LU)) A-61 Remote copy by TC and TCz (volumes controlled by a particular CU).... A-62 Remote copy by UR and URz (whole volumes) A-64 Remote copy by UR and URz (at journals) A-65 Remote copy by UR and URz (for each volume (LU)) A-66 Remote copy by UR and URz (at volumes controlled by a particular CU).. A-67 Causes of Invalid Monitoring Data A-68 Contents vii

8 Troubleshooting the Export Tool A-70 Messages issued by the Export tool A-72 B Performance Monitor GUI reference B-1 Performance Monitor main window B-2 Edit Monitoring Switch wizard B-5 Edit Monitoring Switch window B-5 Confirm window B-6 Monitor Performance window B-7 Edit CU Monitor Mode wizard B-17 Edit CU Monitor Mode window B-17 Confirm window B-19 View CU Matrix window B-21 Select by Parity Groups window B-23 Parity Group Properties window B-25 Edit WWN wizard B-26 Edit WWN window B-26 Confirm window B-26 Edit WWN Monitor Mode wizard B-27 Edit WWN Monitor Mode window B-27 Confirm window B-31 Delete Unused WWNs window B-32 Add New Monitored WWNs wizard B-33 Add New Monitored WWNs window B-33 Confirm window B-35 Add to Ports wizard B-37 Add to Ports window B-37 Confirm window B-39 Monitor window B-40 MP Properties window B-42 Edit Time Range window B-44 Edit Performance Objects window B-46 Add Graph window B-55 Wizard buttons B-64 Navigation buttons B-64 C Server Priority Manager GUI reference C-1 Server Priority Manager window C-2 Port tab of the Server Priority Manager main window C-3 WWN tab of the Server Priority Manager main window C-7 D Virtual Partition Manager GUI reference D-1 Partition Definition tab (Storage System selected) D-2 Partition Definition tab, Cache Logical Partition window (all CLPRs) D-3 viii Contents

9 Partition Definition tab, Cache Logical Partition window (one CLPR) D-4 Select CU dialog box D-7 E Cache Residency Manager GUI reference E-1 Cache Residency window E-2 Multi Set dialog box E-8 Multi Release dialog box E-10 Index Contents ix

10 x Contents

11 Preface This document describes and provides instructions for using Hitachi Performance Monitor, Hitachi Virtual Partition Manager, Hitachi Cache Residency Manager, and Hitachi Server Priority Manager software. Please read this document carefully to understand how to use these products, and maintain a copy for reference purposes. This preface includes the following information: Intended audience Product version Document revision level Changes in this revision Referenced documents Document organization Document conventions Convention for storage capacity values Accessing product documentation Getting help Comments Preface xi

12 Intended audience Product version This document is intended for system administrators and HDS representatives who are involved in installing, configuring, and operating the Hitachi Virtual Storage Platform storage system. Readers of this document should be familiar with the following: RAID storage systems and their basic functions. The Hitachi Virtual Storage Platform storage system and the Hitachi Virtual Storage Platform User and Reference Guide. The Storage Navigator software and the Hitachi Storage Navigator User Guide. This document revision applies to VSP microcode x or later. Document revision level Changes in this revision Revision Date Description MK-90RD October 2010 Initial release MK-90RD December 2010 Supersedes and replaces MK-90RD MK-90RD January 2011 Supersedes and replaces MK-90RD MK-90RD April 2011 Supersedes and replaces MK-90RD MK-90RD August 2011 Supersedes and replaces MK-90RD MK-90RD November 2011 Supersedes and replaces MK-90RD MK-90RD March 2012 Supersedes and replaces MK-90RD MK-90RD July 2012 Supersedes and replaces MK-90RD MK-90RD August 2012 Supersedes and replaces MK-90RD MK-90RD November 2012 Supersedes and replaces MK-90RD MK-90RD January 2013 Supersedes and replaces MK-90RD MK-90RD July 2013 Supersedes and replaces MK-90RD MK-90RD December 2013 Supersedes and replaces MK-90RD Added two conditions as probable causes of invalid monitoring data (Causes of Invalid Monitoring Data on page A-68). Added two storage system maintenance operations that can affect monitoring data (Cautions and restrictions for monitoring on page 2-2). Updated the conditions under which the upper limit setting is disabled (when activity falls below the threshold settings on all prioritized ports, not on any prioritized port) (step 5 in If one-to-one connections link HBAs and ports on page 9-2, caution in Setting a threshold on page 9-13, Table 9-2 Relationship between the thresholds of the prioritized port and the upper limits of the non-prioritized port on page 9-14). xii Preface

13 Updated the maximum cache size from 508 GB to 1,008 GB and the maximum cache residency size from 504 GB to 1,004 GB (step (6) in Creating a CLPR on page 10-9, Partition Definition tab, Cache Logical Partition window (one CLPR) on page D-4). Corrected the description of the Export Tool installation directory on UNIX systems (Installing the Export Tool on a UNIX system on page A- 4, Preparing a command file on page A-4, Preparing a batch file on page A-8). Corrected the reference to a termination code in UNIX Bourne shell scripts (from %? to $?) and UNIX C shell scripts (from %status to $status) (Termination code on page A-11). Referenced documents Virtual Storage Platform documentation: Hitachi Copy-on-Write Snapshot User Guide, MK-90RD7013 Provisioning Guide for Mainframe Systems, MK-90RD7021 Provisioning Guide for Open Systems, MK-90RD7022 Hitachi ShadowImage for Mainframe User Guide, MK-90RD7023 Hitachi ShadowImage User Guide, MK-90RD7024 Hitachi Storage Navigator User Guide, MK-90RD7027 Hitachi Storage Navigator Messages, MK-90RD7028 Hitachi TrueCopy for Mainframe User Guide, MK-90RD7029 Hitachi TrueCopy User Guide, MK-90RD7030 Hitachi Universal Replicator for Mainframe User Guide, MK-90RD7031 Hitachi Universal Replicator User Guide, MK-90RD7032 Hitachi Universal Volume Manager User Guide, MK-90RD7033 Hitachi Virtual Storage Platform User and Reference Guide, MK- 90RD7042 Document organization The following table provides an overview of the contents and organization of this document. Click the chapter title in the left column to go to that chapter. The first page of each chapter provides links to the sections in that chapter. Chapter Chapter 1, Performance overview Chapter 2, Interoperability of Performance Monitor and other products Chapter 3, Monitoring WWNs Description Provides an overview of performance monitoring and management of the Virtual Storage Platform storage system. Describes the interoperability considerations for Performance Monitor. Provides instructions for monitoring control units (CUs) using Hitachi Performance Monitor. Preface xiii

14 Chapter Description Chapter 4, Monitoring CUs Chapter 5, Monitoring operation Chapter 6, Setting statistical storage ranges Document conventions Provides instructions for monitoring WWNs using Hitachi Performance Monitor. Provides instructions for monitoring operations using Hitachi Performance Monitor. Provides instructions for setting statistical storage ranges using Hitachi Performance Monitor. Chapter 7, Working with graphs Provides instructions for working with graphs of performance data. Chapter 8, Changing display of graphs Chapter 9, Server Priority Manager operations Chapter 10, Creating virtual cache partitions Chapter 11, Estimating cache size Chapter 12, Managing resident cache Chapter 13, Troubleshooting Appendix A, Export Tool Appendix B, Performance Monitor GUI reference Appendix C, Server Priority Manager GUI reference Appendix D, Virtual Partition Manager GUI reference Appendix E, Cache Residency Manager GUI reference Provides instructions for changing display of graphs of performance data. Provides instructions for operating the Server Priority Manager software. Provides instructions for creating virtual cache partitions using Hitachi Virtual Partition Manager. Provides instructions for estimating cache size using Hitachi Cache Residency Manager. Provides instructions for performing Cache Residency Manager operations. Provides troubleshooting information for Performance Monitor, Virtual Partition Manager and Cache Residency Manager. Provides instructions for using the Export Tool. Describes the Hitachi Storage Navigator windows and dialog boxes for Performance Monitor. Describes the Hitachi Storage Navigator windows and dialog boxes for Server Priority Manager. Describes the Hitachi Storage Navigator windows and dialog boxes for Virtual Partition Manager. Describes the Hitachi Storage Navigator windows and dialog boxes for Cache Residency Manager. This document uses the following typographic conventions: Convention Bold Italic screen/code Description Indicates text on a window or dialog box, including window and dialog box names, menus, menu options, buttons, fields, and labels. Example: Click OK. Indicates a variable, which is a placeholder for actual text provided by the user or system. Example: copy source-file target-file Note: Angled brackets (< >) are also used to indicate variables. Indicates text that is displayed on screen or entered by the user. Example: # pairdisplay -g oradb xiv Preface

15 Convention Description < > angled brackets Indicates a variable, which is a placeholder for actual text provided by the user or system. Example: # pairdisplay -g <group> Note: Italic font is also used to indicate variables. [ ] square brackets Indicates optional values. Example: [ a b ] indicates that you can choose a, b, or nothing. { } braces Indicates required or expected values. Example: { a b } indicates that you must choose either a or b. vertical bar Indicates that you have a choice between two or more options or arguments. Examples: [ a b ] indicates that you can choose a, b, or nothing. { a b } indicates that you must choose either a or b. This document uses the following icons to draw attention to information: Icon Meaning Description Tip Provides helpful information, guidelines, or suggestions for performing tasks more effectively. Note Calls attention to important and/or additional information. Caution Warns the user of adverse conditions and/or consequences (for example, disruptive operations). WARNING Warns the user of severe conditions and/or consequences (for example, destructive operations). Convention for storage capacity values Physical storage capacity values (for example, disk drive capacity) are calculated based on the following values: Physical capacity unit Value 1 KB 1,000 bytes 1 MB 1,000 KB or 1,000 2 bytes 1 GB 1,000 MB or 1,000 3 bytes 1 TB 1,000 GB or 1,000 4 bytes 1 PB 1,000 TB or 1,000 5 bytes 1 EB 1,000 PB or 1,000 6 bytes Logical storage capacity values (for example, logical device capacity) are calculated based on the following values: Preface xv

16 Logical capacity unit Value Accessing product documentation Getting help Comments 1 KB 1,024 bytes 1 MB 1,024 KB or 1,024 2 bytes 1 GB 1,024 MB or 1,024 3 bytes 1 TB 1,024 GB or 1,024 4 bytes 1 PB 1,024 TB or 1,024 5 bytes 1 EB 1,024 PB or 1,024 6 bytes 1 block 512 bytes The Hitachi Virtual Storage Platform user documentation is available on the Hitachi Data Systems Portal: Check this site for the most current documentation, including important updates that may have been made after the release of the product. The Hitachi Data Systems customer support staff is available 24 hours a day, seven days a week. If you need technical support, log on to the Hitachi Data Systems Portal for contact information: Please send us your comments on this document: doc.comments@hds.com Include the document title and number, including the revision level (for example, -07), and refer to specific sections and paragraphs whenever possible. All comments become the property of Hitachi Data Systems. Thank you! xvi Preface

17 1 Performance overview This chapter provides an overview of the Storage Navigator software products that enable you to monitor and manage the performance of the Hitachi Virtual Storage Platform storage system. Hitachi Performance Monitor overview Server Priority Manager overview Cache Residency Manager overview Virtual Partition Manager overview Performance overview 1 1

18 Hitachi Performance Monitor overview Hitachi Performance Monitor enables you to monitor your Virtual Storage Platform storage system and collect detailed usage and performance statistics. You can view the storage system data on graphs to identify changes in usage rates, workloads, and traffic, analyze trends in disk I/O, and detect peak I/O times. If there is a decrease in storage system performance (for example, delayed host response times), Performance Monitor can help you detect the cause of the problem and resolve it. Performance Monitor provides data about storage system resources such as drives, volumes, and microprocessors as well as statistics about front-end (host I/O) and back-end (disk I/O) workloads. Using the Performance Monitor data you can configure Server Priority Manager, Cache Residency Manager, and Virtual Partition Manager operations to manage and fine-tune the performance of your storage system. Note: To correctly display the performance statistics of a parity group, all volumes belonging to the parity group must be specified as monitoring targets. To correctly display the performance statistics of a LUSE volume, all volumes making up the LUSE volume must be specified as monitoring targets. The volumes to be monitored by Performance Monitor are specified by control unit (CU). If the range of used CUs does not match the range of CUs monitored by Performance Monitor, performance statistics may not be collected for some volumes. Server Priority Manager overview Server Priority Manager allows you to designate prioritized ports (for example, for production servers) and non-prioritized ports (for example, for development servers) and set upper limits and thresholds for the I/O activity of these ports to prevent low-priority activities from negatively impacting high-prority activities. Server Priority Manager operations can be performed only for ports connected to open-systems hosts. Performance of high-priority hosts In a storage area network (SAN) environment, the storage system is usually connected with many host servers. Some types of host servers often require higher performance than others. For example, production servers such as database and application servers that are used to perform daily tasks of business organizations usually require high performance. If production servers experience decreased performance, productivity in business activities can be negatively impacted. To prevent this from happening, the system administrator needs to maintain the performance of production servers at a relatively high level. Computer systems in business organizations often include development servers, which are used for developing, testing, and debugging business applications, as well as production servers. If development servers 1 2 Performance overview

19 Upper-limit control Threshold control experience decreased performance, development activities can be negatively impacted, but a drop in development server performance does not have as much negative impact to the entire organization as a drop in production server performance. In this case, you can use Server Priority Manager to give higher priority to I/O activity from production servers than I/O activity from development servers to manage and control the impact of development activities. Using Server Priority Manager you can limit the number of I/O requests from servers to the storage system as well as the amount of data that can be transferred between the servers and the storage system to maintain production server performance at the required levels. This practice of limiting the performance of low-priority host servers is called upper-limit control. While upper-limit control can help production servers to perform at higher levels during periods of heavy use, it may not be useful when production servers are not busy. For example, if the I/O activity from production servers is high between 09:00 and 15:00 hours and decreases significantly after 15:00, upper-limit control for development servers may not be required after 15:00. To address this situation Server Priority Manager provides threshold control, which automatically disables upper-limit control when I/O traffic between production servers and the storage system decreases to a user-specified level. This user-specified level at which upper-limit control is disabled is called the threshold. You can specify the threshold as an I/O rate (number of I/Os per second) and a data transfer rate (amount of data transferred per second). For example, if you set a threshold of 500 I/Os per second to the storage system, the upper-limit controls for development servers are disabled when the I/O rate of the production servers drops below 500 I/Os per second. If the I/O rate of the production servers goes up and exceeds 500 I/Os per second, upper-limit control is restored to the development servers. If you also set a threshold of 20 MB per second to the storage system, the data transfer rate limit for the development servers is not reached when the amount of data transferred between the storage system and the production servers is less than 20 MB per second. Cache Residency Manager overview Cache Residency Manager enables you to store frequently accessed data in the storage system's cache memory so that it is immediately available to hosts. Using Cache Residency Manager you can increase the data access speed for specific data by enabling read and write I/Os to be performed at the higher front-end access speeds. You can use Cache Residency Manager for both open-systems and mainframe data. Performance overview 1 3

20 When Cache Residency Manager is used, total storage system cache capacity must be increased to avoid data access performance degradation for non-cache-resident data. The maximum allowable Cache Residency Manager cache area is configured when the cache is installed, so you must plan carefully for Cache Residency Manager operations and work with your Hitachi Data Systems representative to calculate the required amount of cache memory for your configuration and requirements. Cache Residency Manager provides the following functions: Prestaging data in cache Priority cache mode Bind cache mode Once data has been placed in cache, the cache mode cannot be changed without cache extension. If you need to change the cache mode without cache extension, you must release the data from cache, and then place the data back in cache with the desired mode. Prestaging data in cache Using Cache Residency Manager you can place specific data into userdefined Cache Residency Manager cache areas, also called cache extents, before it is accessed by the host. This is called prestaging data in cache. When prestaging is used, the host locates the prestaged data in the Cache Residency Manager cache during the first access, thereby improving data access performance. Prestaging can be used for both priority mode and bind mode operations. Prestaging occurs under any of the following circumstances: When prestaging is performed using Cache Residency Manager. When the storage system is powered on. When cache maintenance is performed. Figure 1-1 Cache Residency Manager cache area 1 4 Performance overview

21 Note: If the Cache Residency Manager cache area is accessed for I/O before the prestaging operation is complete, the data may not be available in cache at the first I/O access. To prevent slow response times for host I/Os, the storage system may interrupt the prestaging operation when the cache load is heavy. Do not use the prestaging function if you specify the Cache Residency Manager setting on a volume during the quick formatting operation. To use the prestaging function after the quick formatting operation completes, first release the Cache Residency Manager setting and then specify the setting again with the prestaging setting enabled. For information about quick formatting, see the Provisioning Guide for Open Systems or Provisioning Guide for Mainframe Systems. When external volumes are configured in the storage system, you need to disconnect the external storage system before powering off the storage system. If you power off the storage system without performing the disconnect external storage system operation and then turn on the power supply again, the prestaging process is aborted. If the prestaging process is aborted, you need to perform the prestaging operation again. The prestaging process is aborted if a volume is created, deleted, or restored during the prestaging operation. If the prestaging process is aborted, you need to perform the prestaging operation again after the create, delete, or restore volume operation is complete. Priority mode (read data only) In priority mode the Cache Residency Manager extents are used to hold read data for specific extents on volumes. Write data is write duplexed in cache other than Cache Residency Manager cache, and the data is destaged to the drives when disk utilization is low. The required total cache capacity for priority mode (normal mode) is: standard cache + Cache Residency Manager cache + additional cache The next table specifies the standard cache capacity requirements for priority mode operations. Meeting these requirements is important for preventing performance degradation. For more information about calculating cache size for priority mode, see Estimating cache size on page Table 1-1 Cache capacity requirements for CRM priority mode Settings of priority mode Specified number of cache areas is 8,192 or less and the specified capacity is 128 GB or less Specified number of cache areas exceeds 8,192 or the specified capacity exceeds 128 GB 1 GB = 1,073,741,824 bytes 16 GB 32 GB Standard cache capacity Performance overview 1 5

22 Bind mode (read and write data) In bind mode the Cache Residency Manager extents are used to hold read and write data for specific extents on volumes. Data written to the Cache Residency Manager bind area is not destaged to the drives. To ensure data integrity, write data is duplexed in the Cache Residency Manager cache area, which consumes a significant amount of the Cache Residency Manager cache. Bind mode provides the following advantages over priority mode: The accessibility of read data is the same as Cache Residency Manager priority mode. Write operations do not have to wait for available cache segments. There is no back-end contention caused by destaging data. The required total cache capacity for bind mode is: standard cache + Cache Residency Manager cache Cache Residency Manager bind data that has write attributes is normally not destaged. However, the data is destaged to disk in the following cases: During cache blockage that is caused by certain maintenance operations (for example, cache upgrades) or by cache failure. When the storage system is powered off. When the volume is deleted from Cache Residency Manager bind mode. The next table specifies the cache requirements for bind mode operations. Meeting these requirements is important for preventing performance degradation. For more information about calculating cache size for bind mode, see Estimating cache size on page Table 1-2 Bind mode cache requirements System Type RAID Level or Volume Type Capacity Specifications Cache Residency Cache Requirement Open systems RAID 5 (3390) or RAID 6 RAID 1, or external volumes Slot capacity: 264 KB Cache segment capacity: 16.5 KB Cache segments needed per slot: 48 (slot capacity / cache segment capacity) Slot capacity: 264 KB Cache segment capacity: 16.5 KB Cache segments needed per slot: 32 (slot capacity / cache segment capacity) 3 times the space required for user data: 1 slot = KB = 792 KB = 48 cache segments 2 times the space required for user data: 1 slot = KB = 528 KB = 32 cache segments 1 6 Performance overview

23 System Type RAID Level or Volume Type Capacity Specifications Cache Residency Cache Requirement Mainframe (for example, , ) RAID 5 mainframe or RAID 6 RAID 1 mainframe, or external volumes Slot capacity: 66 KB Cache segment capacity: 16.5 KB Cache segments needed per slot: 12 (slot capacity / cache segment capacity) Note: Even though a mainframe track is 56 KB, because cache is divided into 16.5 KB segments, it requires 4 segments. Slot capacity: 66 KB Cache segment capacity: 16.5 KB Cache segments needed per slot: 8 (slot capacity / cache segment capacity) 3 times the space required for user data: 1 slot = 3 66 KB = 198 KB = 12 cache segments 2 times the space required for user data: 1 slot = 2 66 KB = 132 KB = 8 cache segments Virtual Partition Manager overview The Virtual Storage Platform can connect to multiple hosts and can be shared by multiple users, which can result in conflicts among users. For example, if a host issues many I/O requests, the I/O performance of other hosts may decrease. Virtual Partition Manager allows you to create multiple virtual cache memories, each allocated to different hosts, to prevent contention for cache memory. Performance overview 1 7

24 1 8 Performance overview

25 2 Interoperability of Performance Monitor and other products This chapter describes the interoperability of Performance Monitor and other products. Cautions and restrictions for monitoring Cautions and restrictions for usage statistics Using Server Priority Manager Interoperability of Performance Monitor and other products 2 1

26 Cautions and restrictions for monitoring Performance monitoring has the following cautions and restrictions: Storage system maintenance If the storage system is undergoing the following maintenance operations during monitoring, the monitoring data might contain extremely large values. Adding, replacing, or removing cache memory. Adding, replacing, or removing data drives. Changing the storage system configuration. Replacing the microprogram. Formatting or quick formatting logical devices Adding on, replacing, or removing MP blades Storage system power-off If the storage system is powered off during monitoring, monitoring stops while the storage system is powered off. When the storage system is powered up again, monitoring continues. However, Performance Monitor cannot display information about the period while the storage system is powered off. Therefore, the monitoring data immediately after powering on again might contain extremely large values. Microprogram replacement After the microprogram is replaced, monitoring data is not stored until the service engineer releases the SVP from Modify mode. While the SVP is in Modify mode, inaccurate data is displayed. Changing the SVP time setting If the SVP time setting is changed while the monitoring switch is enabled, the following monitoring errors can occur: Invalid monitoring data appears. No monitoring data is collected. To change the SVP time setting, first disable the monitoring switch, change the SVP time setting, and then re-enable the monitoring switch. After that, obtain the monitoring data. For details about the monitoring switch, see Starting monitoring on page 5-2. WWN monitoring You must configure some settings before the traffic between host bus adapters and storage system ports can be monitored. For details, see Adding new WWNs to monitor on page 3-2, Adding WWNs to ports on page 3-3, and Connecting WWNs to ports on page 3-4. Cautions and restrictions for usage statistics Usage statistics for the last three months (93 days) are displayed in long-range monitoring, and usage statistics for up to the last 15 days are displayed in short-range monitoring. Usage statistics outside of these ranges are deleted from the storage system. 2 2 Interoperability of Performance Monitor and other products

27 In the short range, monitoring results are retained for the last 8 hours to 15 days depending on the specified gathering interval. If the retention period has passed since a monitoring result was obtained, the previous result has been deleted from the storage system and cannot be displayed. When the monitoring switch is set to disabled, no monitoring data is collected. This applies to both long-range and short-range data. For short-range monitoring, if the host I/O workload is high, the storage system gives higher priority to I/O processing than to monitoring. If this occurs, some monitoring data might be missing. If monitoring data is missing frequently, use the Edit Time Range option to lengthen the collection interval. For details, see Starting monitoring on page 5-2. The monitoring data (short-range and long-range) may have a margin of error. If the SVP is overloaded, the system might require more time than the gathering interval allows to update the display of monitoring data. If this occurs, some portion of monitoring data is not displayed. For example, suppose that the gathering interval is 1 minute. In this case, if the display in the Performance Management window is updated at 9:00 and the next update occurs at 9:02, the window (including the graph) does not display the monitoring result for the period of 9:00 to 9:01. This situation can occur when the following maintenance operations are performed: Adding, replacing, or removing cache memory. Adding, replacing, or removing data drives. Changing the storage system configuration. Replacing the microprogram. Pool-VOLs of Thin Image, Copy-on-Write Snapshot, Dynamic Provisioning, and Dynamic Provisioning for Mainframe are not monitored. Note: When you run the CCI horctakeover or pairresync-swaps command for a UR pair or the BCM YKRESYNC REVERSE command for a URz pair, the primary and secondary volumes are swapped. You can collect the beforeswapped information immediately after you run any of the commands. Incorrect monitoring data will be generated for a short time but will be corrected automatically when the monitoring data gets updated. The incorrect data will also be generated when the volume used for a secondary volume is used as a primary volume after a UR or URz pair is deleted. Using Server Priority Manager Starting Server Priority Manager. Ensure that the Time Range in the Monitor Performance window is not set to Use Real Time. You cannot start Server Priority Manager in real-time mode. Interoperability of Performance Monitor and other products 2 3

28 I/O rates and transfer rates. Server Priority Manager runs based on I/O rates and transfer rates measured by Performance Monitor. Performance Monitor measures I/O rates and transfer rates every second, and calculates the average I/O rate and the average transfer rate for every gathering interval (specified between 1 and 15 minutes) regularly. Suppose that 1 minute is specified as the gathering interval and the I/O rate at the port 1-A changes as illustrated in Graph 1. When you use Performance Monitor to display the I/O rate graph for 1A, the line in the graph indicates changes in the average I/O rate calculated every minute (refer to Graph 2). If you select the Detail check box in the Performance Monitor windows, the graph displays changes in the maximum, average, and minimum I/O rates in one minute. Server Priority Manager applies upper limits and thresholds to the average I/O rate or the average transfer rate calculated every gathering interval. For example, in the following figures in which the gathering interval is 1 minute, if you set an upper limit of 150 I/Os to the port 1A, the highest data point in the line CL1-A in Graph 2 and the line Ave.(1 min) in Graph 3 is somewhere around 150 I/Os. It is possible that the lines Max (1 min.) and Min (1 min.) in Graph 3 might exceed the upper limit. Figure 2-1 Graph 1: actual I/O rate (measured every second) Figure 2-2 Graph 2: I/O rate displayed in Performance Monitor (the Detail check box is not selected) 2 4 Interoperability of Performance Monitor and other products

29 Figure 2-3 Graph 3: I/O rate displayed in Performance Monitor (the Detail check box is selected) Interoperability of Performance Monitor and other products 2 5

30 Note: TrueCopy: Server Priority Manager monitors write I/O requests issued from initiator ports of your storage system. When the remote copy functions: (TrueCopy, TrueCopy for Mainframe, Universal Replicator, and Universal Replicator for Mainframe) are used in your environment, Server Priority Manager monitors write I/O requests issued from initiator ports of your storage system. If you give a priority attribute to the RCU target port, all I/Os received on the port will be controlled as the threshold control and its performance data will be added to the total number of I/Os (or the transfer rate) of all prioritized ports. I/Os on the port will not be limited. If you give a non priority attribute to the RCU target port, I/O requests from the initiator port will not be controlled as threshold control and I/ Os on the port will not be limited. On the other hand, I/O requests from a host will be controlled as the upper limit control and I/Os on the port will be limited. Statistics of Initiator/External ports: The initiator ports and external ports of your storage system are not controlled by Server Priority Manager. Although you can set Prioritize or Non-Prioritize to initiator ports and external ports by using Server Priority Manager, the initiator ports and the external ports become the prioritized ports that are not under threshold control, regardless of whether the setting of the ports are Prioritize or Non-Prioritize. If the port attributes are changed from Initiator/External into Target/RCU Target, the settings by Server Priority Manager take effect instantly and the ports are subject to threshold or upper limit control. The statistics of the Monitor Performance window are sum total of statistics on Target/RCU Target ports that are controlled by Server Priority Manager. The statistics does not include the statistics of Initiator/External ports. Because the statistics of Initiator/External ports and Target/RCU Target ports are based on different calculation methods, it is impossible to sum up the statistics of Initiator/External ports and Target/RCU Target ports. Settings of Server Priority Manager main window: The Server Priority Manager main window has two tabs: the Port tab and the WWN tab. The settings on only one tab at a time can be applied to the storage system. If you make settings on both tabs, the settings cannot be applied at the same time. When you select Apply, the settings on the last tab on which you made settings are applied, and the settings on the other tab are discarded. Note: Settings for Server Priority Manager from Command Control Interface: You cannot operate Server Priority Manager from CCI and Hitachi Storage Navigator simultaneously.if you change some settings for Server Priority Manager from CCI, you cannot change those settings from Hitachi Storage Navigator. If you do, some settings might not appear. Before you change features that use Server Priority Manager, delete all Server Priority Manager settings from the currently used features. 2 6 Interoperability of Performance Monitor and other products