Presentation is loading. Please wait.

Presentation is loading. Please wait.

sDS05 - Performance Optimization is Easy with IBM Easy Tier

Similar presentations

Presentation on theme: "sDS05 - Performance Optimization is Easy with IBM Easy Tier"— Presentation transcript:

1 sDS05 - Performance Optimization is Easy with IBM Easy Tier
Brian F. Sherman – Distinguished Engineer 19 May 2014 sDS05 - Performance Optimization is Easy with IBM Easy Tier

2 Accelerate with ATS Storage Webinars
The Free IBM Storage Technical Webinar Series Continues in ATS Technical Experts cover a variety of Storage topics Audience: Clients who are either currently have IBM Storage products or considering acquiring IBM Storage products. Business Partners and IBMers are also welcome. How to sign up? To automatically receive announcements of the Accelerate with ATS Storage webinar series, Clients, Business Partners or IBMers can send an to Information, schedules, and archives: Located in the Accelerate with ATS Blog: Upcoming webinars: May 28th 12:00 ET/9:00 PT Accelerate with ATS: OpenStack - Storage in the Open Cloud Ecosystem June 10th 12:00 ET/9:00 PT Accelerate with ATS: Linear Tape File System Enterprise Edition (LTFS EE) June 26th 12:00 ET/9:00 PT Accelerate with ATS: DS8000 Announcement Update

3 Easy Tier - Introduction
Easy Tier provides automatic and manual volume management capabilities Automated sub-volume/LUN drive tiering (DS8000/SVC/Storwize family) Automated drive performance and capacity utilization balancing to remove hot spots or populate new, empty ranks (DS8000 and SVC/Storwize R7.3+) Prior to R7.3, SVC/Storwize family required manual rebalance script for capacity changes Manual command based volume relocation and pinning to specific tiers SVC/Storwize family have always had these capabilities DS8000 provides volume pinning with R7.1 Easy Tier interfaces A command line interface and a GUI for setup and management Storage Tier Advisor Tool (STAT) for I/O analysis and projected benefits Support for all server platforms with no additional software or host configuration Easy Tier feature DS8700 / DS8800 / DS8870 No charge licensed advanced function feature Have to order the feature and activate the license DS8870 default is to include Easy Tier feature vs. DS8800 have to specifically select it SVC / Storwize family No charge and no license is required (included in base code for SVC/V7000/FlexSystem V7000) prior to R7.3 Optional licensed feature on V3700/V5000 and with R7.3 for V7000 Storage Pool Balancing available on all platforms without a license and is enabled by default

4 Easy Tier Tiering – Basic Operations
Monitors and collects real time performance of each extent to determine how busy (heat) each extent is SVC/Storwize extent size on can be 16MB-8GB with default of 1GB (R previous default of 256MB) DS8000 extent size is 1GB Only small I/O is considered when prioritizing extents to be moved to Flash/SSDs Read and write I/Os 64K and greater are considered sequential Batch and sequential workloads do not interfere with Easy Tier performance optimization Easy Tier can move large IO to different HDD tiers to optimize sequential throughput Performance data continually collected every 5 minutes Migration plan for optimal data placement is generated based on approximately 24 hours of gathered performance statistics Cost of the data movement is considered when performing and migrating data leveraging daily/weekly data Cost/benefit analysis to only move data for noticeable benefit Objective is to use resources to optimize client workload performance and not constantly just moving data Movement of extents is scheduled Limited number of extents are chosen SVC/Storwize up to 2TB/day DS8000 up to 10TB/day Migration continually occurs throughout the next 24 hours according to the established migration plan Warm demote keeps Flash/SSD from becoming overloaded and is independent of the migration plan DS8000 collects Raw I/O statistics Small I/O or Random Writes, reads, cumulative response time, cumulative KiB transferred Large I/O or Sequential Rank Configuration Rank parameters

5 Easy Tier - Process Virtualization Layer Easy Tier Code IOM DPA DMP DM
Domino DB2 Exchange Easy Tier - Process A “hot extent” exists on a HDD when an application makes frequent use of the same area or extent of a volume Virtualization Layer The I/O Monitor captures access patterns and generates usage statistics sending them to the Data Placement Advisor Volume Easy Tier Code The Data Placement Advisor identifies hot extents and outputs potential data migrations to the Data Migration Planner IOM DPA DMP DM The Data Migration Planner performs analysis to deliver recommended data migration plan to the Data Migrator based on the physical storage characteristics Data Migrator confirms and schedules data migration activity based on the data migration plan using built in function to seamlessly relocate the data to higher performing storage without any application interruption Example: Two-tier Storage Pool SAS/NL-SAS Flash/SSD

6 Easy Tier – Automated Data Relocation Improves Performance
Flash/SSD Array Logical Volume Cold Extents Migrate Down Hot Extents Migrate Up SAS Array 10K/15K HDD Cold Extents Migrate Down Hot Extents Migrate Up NL-SAS Array 7200 HDD Improve performance up to 3x with as little as 3-10% Flash/SSD

7 Easy Tier Terminology - Heat and Skew
Hot data Hot data is simply extents that have more I/O workload, relatively speaking, when compared to other extents within the pool and tier Cold data Cold data either has low (or no) I/O workload. Cold data would not benefit from a higher tier and thus is not promoted and is considered for demotion Warm data Warm data is the rest of the workload that is not considered hot or cold. Warm data could be promoted – but that would depend on the workload level and available resources Hot and Warm data will reside on Flash/SSD to maximize capacity Skew Highly skewed workload has a small number of hot extents Low skewed workload has a more even distribution of workload to extents Hot Cold Warm Skew

8 Continuous Easy Tier Evolution
this is a header test Continuous Easy Tier Evolution Easy Tier 1 (DS8700 R5.1 and SVC/Storwize prior to R7.3) Automated cross-tier performance management for SSD/HDD hybrid pools Manual mode management support for dynamic extent pool merge and dynamic volume relocation (DS8000) Easy Tier 2 (DS8700/DS8800 R6.1) Automated cross-tier performance or storage economics management for hybrid pools with any 2 tiers (SSD/ENT, SSD/NL or ENT/NL) Automated intra-tier performance management (auto-rebalance) in hybrid pools Manual mode management support for rank depopulation and optimized volume restriping within non-managed pools (manual volume rebalance) Easy Tier 3 (DS8700/DS8800 R6.2 and SVC/Storwize R7.3) Automated cross-tier performance and storage economics management for hybrid pools with 3 tiers (SSD/ENT/NL) Automated intra-tier performance management in both hybrid (multi-tier) as well as homogenous (single tier) pools (auto-rebalance) Thin Provisioning support for Extent Space Efficient (ESE) Volumes (DS8000) Easy Tier 4 (DS8800 R6.3 / DS8870 R7.0) Support for encryption capable environments DS8800 and DS8870 have FDE capable SSD and nearline drives Easy Tier 5 (DS8870 R7.1) Server Application Heat map transfer Usability improvements in DS8870 R7.2 and SVC/Storwize R7.3 Easy Tier 6 (DS8870 R7.3) Optimization of DS8870 Flash A brief overview of the different Easy Tier versions. Easy Tier v1 Provided automated cross-tier performance management for hybrid pools with SSDs and HDDs only. It also introduced manual mode management support for dynamic extent pool merge and dynamic volume relocation (DVR). DVR provided a significant new capability to DS8000 clients as there was not a non-disruptive method of dynamically moving volumes on the DS8000 previously. Easy Tier v2 Introduced support for hybrid pools with any two tiers, not only providing automated cross-tier performance management for extent pools with SSD/ENT or SSD/NL drive classes but also storage economics management for extent pools with ENT/NL drive classes. Note that ENT = Enterprise drives meaning FC or SAS drives = 10K or 15K RPM drives NL = Nearline drives meaning SATA or Nearline SAS = 7,200RPM ET2 introduced automated intra-tier performance management in hybrid pools (known as auto-rebalance) which automatically rebalances the workload evenly across the ranks within the same tier of a managed extent pool. ET2 also added rank depopulation support for the Easy Tier manual mode and provided an enhancement to the volume migration algorithm for optimized volume restriping within a non-managed pool, referred to as manual volume rebalance. Easy Tier v3 Introduced support for managed pools with 3 tiers, combining automated performance and storage economics management across all three drive tiers within the same pool. By adding support for automated intra-tier performance management for single-tier extent pools, Easy Tier v3 also manages single-tier extent pools using the auto-rebalance feature. Easy Tier v4 Introduced support for Encryption capable SSDs and NL-SAS drives on the DS8800. Clients now have the capability of using Encryption for the security they need and get the benefits of Easy Tier to optimize performance across the three tiers of storage.

9 between adjacent tiers
Easy Tier - Extent Migration Types DS8000 R6.2+ and SVC / Storwize R7.3+ Auto Rebalance Promote / Swap Move hot data to higher performing tiers Warm Demote Prevent performance overload of a tier by demoting warm extent to the lower tier Triggered when bandwidth or IOPS thresholds are exceeded Cold Demote Identify coldest data and move it to lower HDD tier Expanded Cold Demote Demotes appropriate sequential workload to the Nearline tier to better utilize NL tier bandwidth Auto-Rebalance / Storage Pool Balancing Re-distribute extents within a tier to balance utilization across ranks for maximum performance and capacity Move and swap capability SSD RANK 1 SSD RANK 2 SSD RANK n Highest Performance Tier Flash/SSD Auto Rebalance Warm Demote Promote Swap ENT Expanded Cold Demote This slide gives an overview of the Easy Tier extent migration types including: Promote & Swap for moving hot data to higher performing tiers Warm Demote which prevents performance overload of a tier by demoting warm extents to the lower tier and being triggered when bandwidth or IOPS thresholds are exceeded. Cold Demote on HDDs only, where coldest data is identified and moved it to the lower tier. Auto-Rebalance, which re-distributes the extents within a tier to balance utilization across ranks for maximum performance. And Expanded Cold Demote for HDD tiers only, which comes with R6.2. The goal of Expanded Cold Demote is to demote some of the sequential workload to utilize the lower tier’s bandwidth. The bandwidth capability between enterprise tier and SATA/Nearline tier is not significantly different. So the idea is that the available bandwidth on the SATA/Nearline tier can be utilized to balance sequential workloads. Expanded Cold demote will´only demote: Pure sequentially accessed extents Mostly sequentially accessed extents NL SAS Migrations happen between adjacent tiers

10 Easy Tier – Intra-tier Auto Rebalance / Storage Pool Balancing DS8000 R6.2+ and SVC / Storwize R7.3+
Auto rebalance applies to Single or Multi-tier Pools Extents moved between array/ranks of the same tier to balance workload Balances IO skew and capacity utilization Workload performance Additional capacity added to the Pool Proactively avoids hot spots Eliminates need to manually re-stripe extents Performance rebalance begins after 6 hours by default Adding capacity triggers Auto Rebalance to start within minutes Enabled by default on SVC / Storwize Do not need an Easy Tier license on any Storwize system to utilize Storage Pool Balancing Extents migrated to balance workload across ranks – Performance and Capacity balancing 1 2 3 Array 1 Array 2 Array 3 Flash / SSD or Enterprise Disk or Nearline Disk

11 DS8000 Easy Tier – Automatic Mode
4/22/2010 DS8000 Easy Tier – Automatic Mode Extent Migration Plan built based on I/O statistics collected (heat) Dynamically relocates a logical volume’s extents Hot/warm extents relocated to higher performance class of disk (NL SAS -> Enterprise  Flash/SSD) Cold extents relocated to lower performance class of disk (Flash/SSD -> Enterprise  NL SAS) Extent level promotion/demotion requires mixed technologies in a merged extent pool (between any two or three tiers), for example: Flash/SSD + Enterprise + NL SAS Flash/SSD + NL SAS or Enterprise + NL SAS DS8000 Extent Size 1GB for FB 3390 Mod 1 (0.94 GB) for CKD

12 DS8000 Easy Tier – Manual Mode
4/22/2010 DS8000 Easy Tier – Manual Mode Easy Tier Manual Mode allows a user to perform the following actions: Dynamic Volume Migration User can change a logical volume’s storage technology by dynamically relocating between extent pools User can change a logical volume’s extent allocation algorithm (EAM) (e.g. can re-rotate extents within the target extent pool) Dynamic Extent Pool Merge User can merge two existing extent pools without moving data Consolidate extent pools with equivalent disks Merge extent pools with to create a mixed technology extent pool for Automatic Mode Rank Depopulation Storage Administrator can ask that a rank be removed from an extent pool Automatic, non-disruptive and transparent to host access, the used extents will be reallocated to other ranks in the pool and rank freed Manual Volume Rebalance Redistributes a volume’s extents within a non-managed single-tier pool Same pool and can change EAM

13 DS8000 Easy Tier 5 - Easy Tier Server Overview
Host Caching Client IBM Server IBM Server Easy Tier Caching Server DS8870 Flash Cache SAN SSD HDD IBM POWER AIX / DS8000 integration Easy Tier Server is an architecture: A Host flash-cache based on IBM research technology using IBM flash hardware (Easy Tier Caching Client) Easy Tier based algorithms for automatic application aware system-wide optimization (Easy Tier / Easy Tier Caching Server) A proprietary protocol provides in band (SAN) communication and coherency Hosts (Caching Clients) Work independently to cache their applications IO streams providing real-time performance enhancement Use the Cooperative Caching protocol to have cooperative system-aware caching that interfaces with IBM Storage running the Cooperative Caching Server As reads are coming in, 5 minute interval but the client is able to determine and promote extents as well as ET doing so. In general, DS8000 behaves as normal Statistics collection knows the difference of front end DS8000 cache vs. backend No need for extents to move to SSD first before moving to host. The caching server code is looking at the data access vs. where it physically lives. This host is reading this data allot so promote that data to the host flash cache. When data is copied to host flash cache, remove the workload from the SAN/DS8000 so they don’t have to serve up the data. Data has been offloaded to the host. Writes will therefore be optimized. No immediate way to discard the data from the DS8000 cache since we don’t know if another host may need to access it. Host will request 1MB of data and that data has to come through the DS8000 Data will typically live much longer in host Flash Cache vs. typical DS8000 DRAM cache For R1, no way to have portion of one LUN on one Flash Cache and portion of the same LUN on a different Flash Cache. Likewise, can’t have the same data on different Flash Caches in R1 13

14 Easy Tier Caching Client and Easy Tier Server work together
IBM Server IBM Server Easy Tier Cache Client The driver in the Host SCSI stack determines what to cache based on access patterns The cache makes per-io decisions of what to keep in cache and what to evict Selection algorithms achieve high performance by remembering the latest evicted tracks when selecting what to promote into the cache Host Caching Client Host Caching Client Flash Cache Flash Cache SAN Storage Coherency Easy Tier Cache Server While the Client Host SCSI stack driver does its own caching, it also receives ‘advice’ through Easy Tier Server specific communication The ‘advice’ is based on a unified view of the SAN storage, hosts, and their access patterns. It is a priority sorted list of extents The Client combines this advice with its own population list, resulting in both short and longer term cache population, higher hit ratio, and better storage solution optimization, including application aware storage. SAN based multi-host applications are enabled through the Easy Tier Server SAN Storage coherency protocol With coherency managed, storage advanced replication functions such as FlashCopy and remote mirroring are transparently enabled Caching Statistics Caching Server Global Heat & Statistics Data EasyTier SSD HDD IBM Storage 14

15 DS8000 Easy Tier 5 – Easy Tier Application
Databases and ISV solutions will be able to provide performance tips to Easy Tier Will guide what, when, where and how data is placed across the server and storage systems dynamically Enables Administrators to direct data placement based on business knowledge and application knowledge Provide volume level pin / unpin capability Tier placement Let’s start with the role automation plays in Storage Efficiency … IBM Easy Tier is an example of Efficiency technology invented by IBM Research, and brought to market as a capability practically anyone can set up and use. Easy Tier operates continuously, uses advanced analytics, and moves data to the right storage tier; all without disrupting applications. AND Easy Tier works faster than a Storage Administrator could, and learns as it grows, so it maintains efficiency over time. Notice that, as clients deploy Solid-state Storage to improve performance, and high capacity drives to reduce costs; there is less dependence on the traditional Fibre Channel drives that have been the mainstay for high performance applications. Fibre Channel drives are fast, but inefficient: High performance applications usually use low capacity drives with utilization well below 50%. Easy Tier helps clients integrate new storage tiers without adding complexity. IBM’s strategy is to continue building storage systems with advanced Storage Efficiency capabilities built in and highly automated, so they’re more accessible to clients. ***** Additional Information which are very detailed and might be covered in the plenary sessions *********** Easy Tier now supports high capacity 3TB drives to further reduce storage costs while improving application performance. Hybrid environments where performance is automatically optimized across high-speed Solid State Drives and low-cost 3TB disks can now be implemented. On the DS8000, Easy Tier capability is enhanced with 3 tier support. Now, Easy Tier can automatically place data across 3 tiers of storage, creating more ways for clients to optimize performance and efficiency. With 3 tier support, clients can reduce dependency on high cost, low capacity fibre channel drives because: Hot data moves to solid-state drives for faster performance Less active data moves to near-line storage, without archiving As Fibre Channel drives disappear from the configuration, you can reduce power and cooling costs while freeing up capacity inside the storage array. We’re also announcing support for 3 Terabyte drives in Easy Tier, which gives the DS8800 over 2 Petabytes of capacity per system. You can implement Best Practices for information lifecycle management for applications without changes to user processes or applications. Easy Tier Application Data Placement 15

16 Technology Prototype Demonstration: zEC12 + DB2 for z/OS + DS8000 Directive Data Placement
Use Case DB2 for z/OS integration of Directive Data Placement capability in REORG Demonstrate DB2 is able to control the data placement of a Shadow data set based on the storage tier of a REORG Target data set Customer Value: Performance Assurance On Demand Just-In-Time Data Placement Based on unique knowledge about data processing task and data classification, application can timely choose data placement decision Easy Tier function continues to manage the rest of data placement decisions in a coordinated fashion Storage Performance Assurance Ensure a newly created Shadow dataset is placed at the same tier as the existing target dataset during a typical database operation such as REORG Assure new dataset is performance optimized immediately after the reorganization Demonstration Execution Apply special DB2 utility, DFSMS, and storage prototype support in demonstration environment Perform DB2 for z/OS REORG Confirm the storage tier assigned for a newly reorganized data matches original data storage tier 16 16 16

17 DS8000 Easy Tier 5 - Heat Map Transfer
Workload Hotspot Analysis MM/GC/GM Replication Key is the Workload Hotspot analysis. Heat map sent every 12 hours and merged into the next migration plan. EMC announced last year. HDS has not announced similar capability. EMC directly sends the information to the remote over the remote mirroring links. IBM requires s/w on a host and host reads/sends to the secondary. Additional IP isn’t required since use IP addresses already used for regular management. Design not based/require Easy Tier. This approach can be applied for XRC and beyond replication (eg. Preserve a heat map based on Quarter end and reapply the heat map at that time). Using out of band allows other s/w to direct, such as TDMF or for the target of a FlashCopy relationship. S/W such as TPC-R and GDPS will be using this capability. Recommended to have a similar Target configuration so performance will be the same as the primary in the event of a failover. The heat map that is sent goes into the migration plan based on what the configuration is. The heat map sent is for the replicated volumes only so if another workload on the secondary, those volumes will have their own heat and the transferred data will be merged. R7.1 does the secondary. Cannot specify the target volume in R7.1. GDPS solution will be offered later on in 2013 to support the FlashCopy and secondary. 17

18 DS8000 Easy Tier 5 - Heat Map Transfer Learning
Primary site: Optimize the storage allocation according to the customer workload Save the learning data Transfer the learning data from the Primary site to the Secondary site Secondary site: Without learning, only optimize the storage allocation according to the Replication work load With learning, Easy Tier can apply the checkpoint learning data from the primary site Following Primary storage data placement to optimize for the customer workload Heat Map Transfer Utility Available on Windows SLES11+/RHEL4+ TPC-R and GDPS supported TPC SPE for GDPS3.10 for CKD Included in GDPS3.11 for FB and CKD PRIMARY SECONDARY Remote Mirroring (MM, GM, GC) 18

19 DS8000 Easy Tier - Enabling License Enable Easy Tier
Order Easy Tier feature code #7083 Download license key from DSFA and apply it to the DS8000 Enable Easy Tier Storage Image Properties (DS GUI) chsi –Etautomode all | tiered | none storage_image_id Auto Mode all Automatically manage all single and multi-tier pools tiered Automatically manage multi-tier pools only none Do not automatically manage any pools chsi –Etmonitor all | automode | none storage_image_id Monitor Automode Only volumes managed by Easytier auto mode are monitored all All volumes in the system are monitored none No volume is monitored.

20 Flash / SSD Technology in DS8870
Up to 384 SSDs on DS8870 Recommended minimum of 16 SSDs 200/400/800GB SSDs Utilizes existing 2.5” enclosures Hybrid DS8870 1 to 4 1U High Performance Flash Enclosures in base (A) frame Up to GB eMLC encryption capable Flash Cards Up to 9.2TB usable RAID 5 per enclosure Supports up to 120 Flash cards ” HDDs Easy Tier Auto Rebalance will distribute the workload among traditional SSDs and Flash Enclosure according to their IOPS capacity All Flash DS8870 1 to 8 1U High Performance Flash Enclosures in base (A) frame Up to 73TB usable in all Flash DS8870 Supports up to 240 Flash cards 20

21 Flash / SSD Technology in SVC / V7000
SVC optimizes scalable FlashSystem family or traditional SSD FlashSystem 840 provides 4TB – 40TB usable RAID5 Integrated bundle available as IBM FlashSystem V840 Traditional SSD with CG8: Up to 4 SSDs per node Up to 8 nodes for a maximum capacity of 32 SSDs RAID level options are 0,1,10 only Traditional SSD with DH8: Up to 48 SSDs per node pair Up to 2 Expansion enclosures per IO Group RAID level options are 0, 1, 5, 6, 10 Storwize V7000 / FlashSystem family Previous V7000: supports up to 240 drives per I/O Group (up to 960 in a cluster) Plan for maximum capacity of around SSDs (depending on workload) New V7000 with R7.3: supports up to 504 drives per I/O Group (up to 1056 in a cluster) Plan for maximum capacity of around SSDs (depending on workload) First 24 drives in the Control Enclosure are on a dedicated SAS chain. Ensure remaining drives in Expansion Enclosures are split across the two other SAS chains for optimal SSD performance 200/400/800GB option RAID level options are 0,1,5,6,10

22 Update on SVC prefetch SVC has a prefetch algorithm to improve cache hit ratios especially for sequential workloads With some real world client workloads this algorithm can result in excessive prefetch with significantly higher MB/s from the backend than is read by the host Especially with Flash or Enterprise disk subsystems behind SVC it can be better to rely on the backend storage and to disable SVC prefetch Available with and later using the chsystem command Only use at direction of SVC support R7.3 cache changes address prefetch concern 22

23 SVC and DS8000 - Easy Tier Deployment Scenarios
With SVC R7.3, use Easy Tier at the SVC level as the default in these scenarios: Utilize DS8000 Flash, FlashSystem or new enclosures with DH8/R7.3 as they can benefit multiple SVC storage pools Requirement to pin a VDisk to a specific tier Able to exploit three tier and auto rebalance Plan on utilizing SVC Compression Use Easy Tier on DS8000 in these scenarios: zOS and IBM i environments Requirement for Easy Tier Application, Easy Tier Server, Heat Map transfer capability SVC environments prior to R7.3 You must have free extents for Easy Tier to function For DS8000, plan on approximately 10 free extents per rank in the pool For SVC/Storwize, plan on a minimum of 10 free extents per MDisk Use Easy Tier on SVC R7.3 for all other configuration implementations This is likely the most popular choice once at SVC R7.3 23

24 Workload skew drives Easy Tier benefits
50% of the extents do 10% of the MB and virtually no random IOPS! 58% of the random IOPS and 33% of the MB from about 5% of the extents! 24

25 Drive Selection in an Easy Tier Environment
SSD FlashSystem family 10-20% Flash/SSD, 80-90% Enterprise Provides performance similar to Flash/SSD only solution at significantly reduced costs All data guaranteed to have at least enterprise performance 3-5% Flash/SSD, 95-97% Enterprise Provides improved performance compared to single tier solution Removes requirement for over provisioning for high access density environments 3-5% Flash/SSD, 25-53% Enterprise, 40-70% NL SAS Provides improved performance and density to a single tier solution Significant reduction in environmental costs 20-50% Enterprise, 50-80% NL SAS Provides reduced costs and comparable performance to a single tier Enterprise solution Enterprise Nearline

26 SVC / V7000 Easy Tier Implementation
Prior to R7.3, Easy Tier supported two tiers: Generic SSD which is tier 0 Generic HDD which is tier 1 With R7.3, Easy Tier supports any three tiers and Storage Pool Balancing By default if multiple tiers of storage are present in a storage pool then Volumes in that pool will be measured and managed External MDisk performance setting defaulted by controller type New setting - easy_tier_load for external MDisks displays the expected performance capability of the MDisk within its particular tier Tier 0 Tier 1 Tier2 SSD Enterprise Nearline ---

27 V7000 Tier Selection for Internal Flash
By default, a Storage Pool with two or three tiers will be Easy Tier Active. Easy Tier can be disabled using CLI svctask chmdiskgrp –easytier off <Storage Pool/MDisk Group ID or name> Auto, on, off and measure supported settings With R7.3, On is the default and activates Storage Pool Balancing in single tier pool

28 SVC/V7000 Easy Tier Operational Considerations
Volume can have Easy Tier set to on/off/measure (default is on) Image and Sequential Volumes can be measured but not managed by Easy Tier Pool can have Easy Tier set to on/auto/off/measure (default is auto) External MDisk performance control Low, medium, high, very high Set by default based on type of external controller and can be changed via CLI Easy Tier is supported for FlashCopy source and target and Metro or Global Mirror primary and secondary Volumes MM or GM secondary Volumes do not inherit characteristics of primary Example: If secondary becomes primary and usage changes the Easy Tier will have to adjust the extents over time to accommodate the new access patterns Easy Tier is supported for Thin Provisioned Volumes Only the real storage will be subject to management Easy Tier manages extents of Compressed Volumes with R7.1+ Volume migrate Management of the Volume will be disabled during a migration Measurement of the Volume will reflect the status of the source storage pool until the migration completes

29 Easy Tier on Compressed Volumes
Previously support for Easy Tier with compression volumes was disabled Compression I/O looked random to Easy Tier and this prevented Easy Tier being able to detect hot extents The IOs generated by compression are never greater than 32 kb. Therefore all IOs seen by Easy Tier are “hot” IOs Resolved by adding extra metadata to internal I/O to give Easy Tier hints as to source I/O size, allowing Easy Tier to work correctly Easy Tier for compressed volumes will not count write IOs – only reads Support enabled in R7.1+

30 Thin Provisioning and Easy Tier on SVC/V7000
Thin Provisioned Volumes in Storage Pools using EasyTier must use a grain size of 64 KB or greater If grain size is default of 32K then all I/Os to TP volumes will be considered by ET algorithms since even large sequential I/Os from host will be broken up into 32K I/Os resulting in odd ET behaviour and performance issues See flash for more details R6.4 changed default grain size of a Thin Provisioned Volume to 256KB rather than 32KB © Copyright IBM Corporation 2011

31 Easy Tier – Supported Environment and Restrictions
4/22/2010 Easy Tier – Supported Environment and Restrictions Supported in SVC R6.1+, SVC R7.3, DS8700 R5.1+, DS8800 R6.1+ and DS8870 R7.0+ Can merge any two extent pools on DS8000 except: Both must be same extent type (CKD or Fixed Block) At most, one of two extent pools can have a space efficient repository At most, one of one of two extent pools can have virtual capacity Not allowed if one extent pool is homogeneous with Flash/SSD disks and additionally has space efficient repository or virtual capacity configured You must have free extents for Easy Tier to function For DS8000, plan on having approximately 10 free extents per rank in the pool For SVC/Storwize, plan on 10 free extents per MDisk 31 31

32 Easy Tier – Modeling Tools
Storage Tier Advisor Tool (STAT) Extracts heat data collected by Easy Tier for volumes that are being monitored Disk Magic Supports 5 predefined skew levels for prediction with Easy Tier Can utilize detailed Easy Tier data to generate actual client skew chart Tiered Solution Advisor available in V (open) / (zOS and IBM i) Utilizes either predefined or actual skew to predict the number of I/Os Higher skew results in a more aggressive sizing FLASHDA (z/OS only) Identifies what datasets and devices have the highest accumulated read-only disconnect time IBM Tivoli Storage Productivity Center (TPC)

33 Storage Tier Advisor Tool (STAT)
System Summary Report Support pool configurations with up 3 tiers (Flash/SSD – ENT – NL SAS) Tier status: Existing Tier Flash/SSD-ENT-NL SAS is IOPS/BW overloaded/skewed Storage Pool Performance Statistics and Improvement Recommendation Existing Tier Status showing rank utilization for each tier Recommended Flash/SSD, Enterprise, NL Configurations Expanding 1-tier HDD and any 2-tier configurations to 3-tier configurations Expanding existing tiers when tiers are BW or IOPS overloaded Volume Heat Distribution Showing capacity and heat distribution by tier (hot data - warm data - cold data) Systemwide Recommendation Recommended Flash/SSD Configuration Recommended Enterprise Configuration Recommended NL SAS Configuration This Slide shows the Enhancements for the Storage Tier Advisor Tool (STAT). In R6.2 the Storage Tier Advisor Tool supported pool configuration with up to three tiers (SSD – ENT – NL) which are listed on the System Summary Report for each extent pool together with tier status messages such as SSD/ENT/NL is IOPS or BW Overloaded or Skewed. The Storage Pool Performance Statistics & Improvement Recommendation report provides information for each extent pool about the Existing Tier Status within the pool showing the rank utilization for each tier Recommended drive class configurations for SSD, ENT and NL drive classes based on the workload in the pool allowing to expand 1-tier HDD and any 2-tier configurations to 3-tier configurations, or expanding existing tiers when tiers are bandwidth or IOPs overloaded and the Volume Heat Distribution, now showing capacity and heat distribution by tier (hot data - warm data - cold data) It also provides a systemwide summary of all recommended extent pool configurations in the Systemwide Recommendation report. Download STAT (no charge) and run on Windows

34 DS8000 STAT – System Summary
Pools/volumes monitored How much hot capacity Current configuration of the pools Flash/SSD, Enterprise, NL or any combination Tier Status IOPS overloaded, BW overloaded Data Management Status

35 SVC/Storwize R7.3 STAT: System Summary
Pools/volumes monitored How much hot capacity Current configuration of the pools Flash/SSD, Enterprise, NL or any combination Tier Status IOPS overloaded, BW overloaded

36 DS8000 STAT - Intra-tier Rebalance
Intra-tier Status Ranks monitored for IOPS Current IOPS, utilization of the rank and projected

37 Intuitive Easy Tier reporting Simplifies Storage Administration
DS8000 R7.2 and SVC / Storwize R7.3 Easy Tier Workload Categorization New workload visuals help clients compare activity across tiers within and across pools to help determine optimal drive mix for current workloads Easy Tier Data Movement Daily Report New Easy Tier summary report every 24 hours illustrating data migration activity (5-min. intervals) can help visualize migration types and patterns for current workloads Easy Tier Workload Skew Curve Shows skew of all workloads across the system in a graph to help clients visualize and accurately tier configurations when adding capacity or a new system Clients can import data into Disk Magic This provides the ability to deliver recommendations on how to adjust configurations over time. I am pretty certain that this is unprecedented in any platform... The idea that it will tell you what to add as you reach critical levels of performance or capacity is a huge benefit: Takes out an enormous amount of labor and guess work by the storage staff for normal performance / capacity planning needs. This could be the ultimate reason to leave the data parked on your ds8k vs moving it to something "cheaper".. Cause it's costing more in complexity that saving you capital expense (which is plummeting anyway).... Assumes client already has DS8000 installed Previously, tool illustrated skew for each extent pool, but was not able to import into Disk Magic tool 37 Page 37

38 DS8000 STAT - Performance Improvements
STAT provides estimated performance improvements for adding different amounts of solid state drives This figure is related to the backend latency for both read and write IOs It does not translate directly into a host performance improvement It is based on the skew of the workload and a set of simple assumptions regarding the response time improvement of Flash/SSDs It is based on the 24 hour workload not any particular interval This figure can be useful to get a feeling for effective Easy Tier could be but needs to be considered with other information such as cache hit ratios, sequential percentages during batch etc.

39 SVC/Storwize R7.3 STAT- Performance Improvement Recommendation
Focus on response time improvement Give suggestions how many hot data should be migrated to Flash/SSD tier and how many new Flash/SSD MDisks should be added

40 STAT – Volume Heat Distribution Updated
DS8000 R6+ and SVC / Storwize R7.3 40

41 Easy Tier - Summary Easy Tier operations
DS8000 SVC/Storwize family IO Skew and drive selection guidelines Storage Tier Advisor Tool

42 IBM Systems Lab Services and Training
Deep skills driving client satisfaction and technical effectiveness for smarter computing IBM Systems Lab Services and Training infuses intelligence into how global information technology works. is composed of experts who develop and deploy solutions across IBM’s systems family offerings. From in-depth product expertise, to training, to platform-specific hardware and software solutions, we’re here for you! IBM Systems Training IT infrastructure training that enables a Smarter Planet IBM’s award-winning classes and industry-renowned technical events helps you fully exploit the power of your IBM technology investment. Our comprehensive portfolio of skills and career accelerators are designed to meet all your training needs – individual, departmental and organizational. click to explore our offerings Visit us Partnered with IBM selected global training providers

43 Thank You

44 Trademarks and Disclaimers
Adobe, the Adobe logo, PostScript, and the PostScript logo are either registered trademarks or trademarks of Adobe Systems Incorporated in the United States, and/or other countries. IT Infrastructure Library is a registered trademark of the Central Computer and Telecommunications Agency which is now part of the Office of Government Commerce. Intel, Intel logo, Intel Inside, Intel Inside logo, Intel Centrino, Intel Centrino logo, Celeron, Intel Xeon, Intel SpeedStep, Itanium, and Pentium are trademarks or registered trademarks of Intel Corporation or its subsidiaries in the United States and other countries. Linux is a registered trademark of Linus Torvalds in the United States, other countries, or both. Microsoft, Windows, Windows NT, and the Windows logo are trademarks of Microsoft Corporation in the United States, other countries, or both. ITIL is a registered trademark, and a registered community trademark of the Office of Government Commerce, and is registered in the U.S. Patent and Trademark Office. UNIX is a registered trademark of The Open Group in the United States and other countries. Java and all Java-based trademarks and logos are trademarks or registered trademarks of Oracle and/or its affiliates. Cell Broadband Engine is a trademark of Sony Computer Entertainment, Inc. in the United States, other countries, or both and is used under license therefrom. Linear Tape-Open, LTO, the LTO Logo, Ultrium, and the Ultrium logo are trademarks of HP, IBM Corp. and Quantum in the U.S. and other countries. Other product and service names might be trademarks of IBM or other companies. Information is provided "AS IS" without warranty of any kind. The customer examples described are presented as illustrations of how those customers have used IBM products and the results they may have achieved. Actual environmental costs and performance characteristics may vary by customer. Information concerning non-IBM products was obtained from a supplier of these products, published announcement material, or other publicly available sources and does not constitute an endorsement of such products by IBM. Sources for non-IBM list prices and performance numbers are taken from publicly available information, including vendor announcements and vendor worldwide homepages. IBM has not tested these products and cannot confirm the accuracy of performance, capability, or any other claims related to non-IBM products. Questions on the capability of non-IBM products should be addressed to the supplier of those products. All statements regarding IBM future direction and intent are subject to change or withdrawal without notice, and represent goals and objectives only. Some information addresses anticipated future capabilities. Such information is not intended as a definitive statement of a commitment to specific levels of performance, function or delivery schedules with respect to any future products. Such commitments are only made in IBM product announcements. The information is presented here to communicate IBM's current investment and development activities as a good faith effort to help with our customers' future planning. Performance is based on measurements and projections using standard IBM benchmarks in a controlled environment. The actual throughput or performance that any user will experience will vary depending upon considerations such as the amount of multiprogramming in the user's job stream, the I/O configuration, the storage configuration, and the workload processed. Therefore, no assurance can be given that an individual user will achieve throughput or performance improvements equivalent to the ratios stated here. Prices are suggested U.S. list prices and are subject to change without notice. Starting price may not include a hard drive, operating system or other features. Contact your IBM representative or Business Partner for the most current pricing in your geography. Photographs shown may be engineering prototypes. Changes may be incorporated in production models. © IBM Corporation All rights reserved. References in this document to IBM products or services do not imply that IBM intends to make them available in every country. Trademarks of International Business Machines Corporation in the United States, other countries, or both can be found on the World Wide Web at ZSP03490-USEN-00

Download ppt "sDS05 - Performance Optimization is Easy with IBM Easy Tier"

Similar presentations

Ads by Google