Performance Testing Guide
Contents
1. IOCITY Virtual Platform Disk Optimizer Performance Testing Guide V locity 3 Performance Testing Guide A Note to Product Testers The aim of this testing guide is to assist you in documenting the performance gains obtained from V locity 3 virtual platform disk optimizer and to publish a factually accurate and well informed recommendation of its benefits In addition we encourage you to use the V locity 3 User s Manual accompanying the product to familiarize yourself with all facets of V locity 3 We hope you ll find this guide useful and would like your feedback Understanding Fragmentation The problem of file system fragmentation has been with us for decades Direct attached storage systems were plagued by it first and then administrators of NAS solutions had to contend with it Now SAN infrastructures are suffering its effects inherited from the overlying file systems The nature of a decoupled storage model precludes the operating system from being aware of the type of storage it s using This greatly limits the operating system s ability to optimize for a particular storage model One of the most significant potential issues and probably the most unrecognized is fragmentation in a storage system With the implementation of a SAN in their storage environments many Windows Server administrators believe that file system fragmentation which they accepted and dealt with when using DASD Direct Attached Storage Device2. are copies of the same script but used concurrently they will cause fragmentation to occur much faster speeding up this test Here is the procedure to execute the workload script on a freshly formatted 40GB Thin provisioned VM a Rename the extension to vbs and then run Create_Folder vos This VB Visual Basic script will create SQL_ Data and SQL_Log folders in the 40GB test volume When SQL scripts are running the data and logs will get stored in these folders b Run DropCreateDB sal This script will delete the databases from the server if it exists and create new databases TestlabDB1 TestlabDB2 TestlabDB3 c Rename the extension to cmd and then run ScriotBatch cmd This Batch file will select all three SQL scripts and run them simultaneously NOTE Before running this script you need to modify the location of below SQL scripts in this batch file Perform a Volume Analysis with V locity 3 and save the Job Report Evaluate snapshot deduplication Thin provisioning etc growth via SAN management tools mechanisms Restore the test image As applicable restore the LUN to its original state e g reclaim space etc Enable IntelliWrite in V locity 3 Run the workload script again gt PN ppp p Open the V locity 3 Guest and observe statistics in the Dashboard Notice the number of fragments prevented by IntelliWrite 4 Diskeeper 7 Vlocity 3 Performance Testing Guide IntelliWrite Fragmentation Prevention Inte
3. 3 is done so that a steady state of the disk can be analyzed as a baseline before running V locity 3 After completing all installation steps make sure to save the image of the test volume for repeated tests using an imaging product like Norton Ghost Full volume block based imaging software must be used so true comparisons can be made Test 1 Fragmentation Impact on SAN Disk Performance The first few steps involve measuring file performance on a fragmented file The procedure is then repeated on a non fragmented file At the end these results will be compared The test procedure is as follows 1 Run lometer The test file need not be larger than 10 20GB JE BURWST003 Worker 1 L Worker 2 E E BURWSTD03 Starting Disk Sector 1 HE Worker 1 j L Worker 2 ea er of Outstanding I Os orker Worker 2 j En Test Connection Rate r Transactions per connection 4 Diskeeper 5 corporation Vlocity 3 Performance Testing Guide 2 While lometer is creating the test file use the FragmentFile exe utility This will fragment the lometer test file and can be done by running the following command from the command line prompt FragmentFile e Nobw tst 1 The use of 1 in the parameters is to ensure significant fragmentation of the file which in turn will ensure that lometer is accessing fragments of its test file 3 Once lometer is complete perform a manual analysis of the test
4. 578 147 0827 145 375139 3 747541 1 89094 1856601 292 4578 0 3 417895 4 35749 2 39 DISK E SAN LUNI 292 4578 147 0827 145 375139 3 747541 1 89094 1 856601 292 4578 0 3 417895 4 35749 2 40 Time Stamp 41 2010 11 30 16 12 21 420 42 Aborted test Figure 4 Example Fragmented File Report 31 End access specifications 32 Results 33 Target Type Target Name Access Specification Name Managers Workers Disks Ops Read lOps Write lOps MBps Read MBp Write MB Transactic Connectic Average R Average R Aver 34 ALL All Allin one 1 1 1 577 23687 290 995951 286 240919 7 398929 3 729688 3 669241 577 2369 0 1 731392 2 592307 0 8 35 MANAGER BURWSTOO3 All in one 1 1 577 23687 290 995951 286 240919 7 398929 3 729688 3 669241 577 2369 0 1 731392 2 592307 0 8 36 PROCESSOR CPU 0 37 PROCESSOR CPU 1 38 WORKER Worker 1 Allin one 1 577 23687 290 995951 285 240919 7 398929 3 729688 3 669241 577 2369 O 1 731392 2 592307 0 8 39 DISK E SAN LUN1 577 23687 290 995951 286 240919 7 398929 3 729688 3 669241 577 2369 0 1 731392 2 592307 0 8 40 Time Stamp a MIN n nrn 9 ARQ ln Figure 5 Example Non Fragmented File Report 4 Diskeeper 6 corporation V locity 3 Performance Testing Guide Test 2 Fragmentation Prevention This test demonstrates the effectiveness of IntelliWrite in preventing fragmentation largely offsetting the need to run after the fact defragmentation This technology was developed in order to ensure compatibility wit
5. Corporation logo V locity logo V locity InvisiTasking V Aware CogniSAN and IntelliWrite are registered trademarks or trademarks owned by Diskeeper Corporation in the United States and or other countries All other trademarks are the property of their respective owners 4 Diskeeper corporation
6. h SAN features such as de duplication snapshots Thin provisioning etc The purpose of this test is to validate that IntelliWrite does not incur any unwanted side effects associated with SAN advanced functionality For the purposes of this test enable any and all SAN features that are believed to experience unwanted side effects from defragmentation of SAN attached Windows systems e g growth of a Thin provisioned VM This procedure involves running a workload script that simulates typical disk activity on the system i e web browsing downloading and creating mail editing documents downloading and deleting media files etc It may include installing several system updates The workload needs to be scripted so that it is repeatable and it should reflect typical system activity over a substantial amount of time i e a week Attached here is a sample workload for SQL Server requires that application be installed on the test system SQL Workload zip We can provide additional sample workload scripts for other processes e g Microsoft Office activity Please contact us if we can provide any alternate workload scripts Here is the procedure to test IntelliWrite 1 Use the test volume s image created in the Test Environment Setup stage 2 Run the workload script The three SQL scripts are SQLQuery4a sql adds data to TestlabDB1 SQLQuery4b sql adds data to TestlabDB2 SQLQuery4c sql adds data to TestlabDB3 The SQL scripts
7. lliwrite increases system performance by preventing fragmentation before it happens System Statistics for All Volumes Since 12 a m 6 5 2011 This section shows a summary of fragmentation handled by Intelliwrite Instant Defrag and Automatic Defragmentation B Fragmentation Prevented by Intelliwrite 2 036 Fragments 95 Ratio of Fragments E Fragmentation Eliminated by Instant Defrag 61 Fragments 3 preven ted vs E Fragmentation Eliminated by Defragmentation 40 Fragments Des j Defrag mented Total Fragmentation Handled within Timeframe 2 137 Fragments Average time to handle fragments using Intellwrite and Instant Defrag 0 00 seconds Disk Accesses I Os Saved for All Volumes Scroll to Top Diskeeper estimates the amount of disk accesses I Os saved on your system This optimization reduces the number of disk accesses required to read or write files With this improved efficiency disks are accessed less frequently resulting in increased reliability and longer disk lifespan This number is formulated by comparing the amount of Os your system uses with Diskeeper to the amount of Os your system would have needed if no defragmentation solution was used Disk Accesses I Os Saved 2 150 468 System Fragmentation Prevention Graph Scroll to Top This graph shows fragmentation prevented by Intelliwrite every second within the last minute for your entire system Click here for help EH System fragmentation prevented each sec
8. locity 3 Host Agent and connect to the Host Service Review the list of VMs synchronizing resource usage with the Host and verify that it includes all VMs in which V locity 3 Guests are installed 9 V locity 2 Host This test step would only be employed if you have installed the V locity 2 Host on a Hyper V or ESX hypervisor Go to the list of virtual disks and perform disk compaction the disks may not be in use by running VMs to perform compaction Verify that compaction succeeds and the unused space is reclaimed 4 Diskeeper 1 YE 10 11 12 locity Performance Testing Guide DAILY Perform a Manual Analysis to see that the volumes are in fact being defragmented automatically Save the Volume Maps both File Performance and File Structure views and Job Reports for each volume DAILY Copy the log file named ControllerLog txt located at C Program Files Diskeeper Corporation V locity on each VM Volume Maps Job Reports and Job Logs Please report any problems with detailed information on how to reproduce them Feature Testing System Performance Testing ee SS PS Create fragmentation on 40GB volume with moderate fragments Copy a large size file 2 GB in size and note down the time taken to copy the file Using the readfile exe tool measure the read time to read the above large file Delete the copied file and reboot the VM Repeat steps 2 4 Install V locity 3 Guest and defragment the volume Repeat s
9. ntelliWrite is effective in preventing the vast majority of fragmentation which is a much more efficient way of handling fragmentation compared to reactive fragmentation removal More importantly it demonstrates full compatibility with block based CDP Continuous Data Protection provisioning and de duplication solutions The tests described in this section rely on using a couple of industry standard benchmark tools The first such tool is lometer I O measurement and characterization tool to measure system performance The tool as well as relevant documentation can be obtained from the following location http www iometer org Diskeeper Corporation has a number of methods for creating fragmented data in a testing capacity In order for lometer tests to represent the impact of fragmentation the lometer test file iobw tst must be fragmented FragmentFile exe is one such method to ensure that the file is fragmented There are additional methods and benchmarking tools that can be used to measure the impact of fragmentation So long as the file s used by the benchmarking tool e g SQLIOSim Jetstress etc can be fragmented a before and after comparison can be evaluated 4 Diskeeper 3 V locity 3 Performance Testing Guide Test Environment Setup The test should be performed on a VM installed over a new SAN LUN Logical Unit Number Install Iometer and the V locity 3 Guest on the test VM During V locity 3 installation disable Aut
10. omatic Defragmentation and Automatic Space Reclamation both are enabled by default This is detailed with graphics below With revolutionary InvisiTasking technology locity works automatically in the background With no negative performance impact Automatic defragmentation improves computer performance without slowing other operations and is recommended to maintain peak performance Your computer s performance degrades anytime Automatic Defragmentation is disabled Enable Automatic Defragmentation on all volumes on this computer Important configuration information For SAN attached systems Click here InstallShield Figure 1 Uncheck the Enable Automatic Defragmentation and Enable Automatic Space Reclamation options during installation After installation open V locity 3 and disable IntelliWrite Right click on the test volume and select IntelliWrite from the dropdown menu Computer MSALES2K65 1Db4 Figure 2 Select IntelliWrite from the volume dropdown menu 4 Diskeeper i corporation V locity 3 Performance Testing Guide Intelliwrite Automatic Optimization Hoot Time Defragmentation JT Prevent fragmentation on this volume Disabling this feature will disable Intelliv rite on selected volumes only Figure 3 Uncheck the option to Prevent fragmentation on this volume disabling IntelliWrite Disabling these automatic real time operating features of V locity
11. ond by Intelliwite for all volumes Number of fragments prevented this second 22 BN HUSN SEREN le ee a 5 Fragments prevented per second 5 Proactive prevention is the most energy efficient method to eliminate increased storage power consumption caused by fragmentation Figure 6 Real time Fragmentation Prevention in the V locity 3 Dashboard 10 Perform Volume Analysis and save the Job Report Compare with the report obtained in Test Step 3 The test ohase with IntelliWrite will have far fewer fragments 11 Evaluate snapshot deduplication Thin provisioning etc growth via SAN management tools mechanisms The increase in activity space for these SAN solutions should be identical to their activity in the first IntelliWrite disabled phase Equal activity storage growth for the two phases of the test with and without IntelliWrite would validate that the technology is able to eliminate the majority of potential fragmentation without any negative side effects Conclusion Thank you for testing V locity 3 We hope that you found this Performance Testing Guide useful and we would like your feedback Should you have any questions please contact us directly We look forward to reading all about your results and recommendations If you are interested in participating in a joint Case Study utilizing your performance test results with V locity 3 please let us know 2011 Diskeeper Corporation All Rights Reserved The Diskeeper
12. size 5 Repeat steps 1 4 on a 10GB max provisioned size volume Dynamic Thin Disk Growth Testing 1 Create a volume with moderate fragmentation in a Dynamic Hyper V Thin ESX ESXi Disk 2 Note the disk size 3 Enable Automatic Defragmentation and let it finish 4 Note the size of the disk Growth should be minimal Testing V locity 3 Advanced This document describes test procedures for evaluating performance improvements from using V locity 3 in a SAN environment These tests will take 4 6 hours of actual hands on involvement for a test engineer spread out over a couple of days allowing for time to create the environment let the benchmark job complete etc The tests outlined in this document are proof of concept_ and are not intended to directly replicate a real world environment The performance benefit of virtual platform disk optimization is always proportionate to the amount of file fragmentation and the frequency that fragmented data is accessed The first set of tests demonstrates performance improvements from handling fragmentation in a SAN environment Specifically the tests show that a system affected by fragmentation performs less efficiently than a system on which fragmentation has been handled by V locity 3 The second set of tests evaluates IntelliWrite fragmentation prevention technology involving comparing a system with V locity 3 and without V locity 3 under the same workload This test demonstrates that I
13. storage has gone away Testing V locity 3 Simple 1 Install V locity 3 Host for Hyper V Microsoft Hyper V or V locity 3 Host Agent VMware ESX ESXi 4 04 a Hyper V Installation The V locity 3 Host for Hyper V installs on the Microsoft Hyper V Hypervisor b ESX ESXi 4 0 Installation The V locity 3 Host Agent installs on any Windows OS system which can communicate with both the VMware Host vCenter and the V locity 3 Guests This can also be a VM Virtual Machine 2 Install V locity 3 Guest on several VMs hosted by the server 3 Runa Manual Analysis on each volume on all guest VMs and save both the Volume Map and Job Report for each volume Save these files as part of your testing 4 Verify that V locity 3 works correctly and successfully removes fragmentation on the guest VMs 5 Inthe V locity 3 Guest Dashboard verify that V locity 3 Guests running in VMs synchronize their resource usage with the Host 6 Verify that InvisiTasking technology is working properly and that there is no negative impact on the performance of the Host or any of the VMs from defragmentation 7 Verify that V Aware and CogniSAN technologies in the V locity 3 Guests on VMs are aware of the resource usage at the virtual host and shared storage layers For example heavy CPU or disk I O activity in one VM over the same host storage should affect defragmentation on V locity 3 Guests on other VMs 8 Open the V locity 3 Host for Hyper V or V
14. teps 2 4 twice and tabulate the readings InvisiTasking Testing gt VI ff p FL p Create fragmentation on 40GB volume with moderate fragments Copy a large size file 2 GB in size and note down the time taken to copy the file Delete the copied file and reboot the machine Repeat steps 2 3 Install V locity 3 Guest and enable Automatic Defragmentation Repeat steps 2 4 twice and tabulate the readings V Aware and CogniSAN Testing 1 2 oO SY DY This test involves two separate test VMs labeled A amp B each provisioned with 40GB volumes and running over the same physical data store Create fragmentation on 40GB volumes of test VMs A amp B with moderate fragments Copy a large size file 2 GB in size on test VM A and note down the time taken to copy the file Delete the copied file and reboot the machine Repeat steps 3 4 Install V locity 3 Guest on test VM B and enable automatic defragmentation Repeat steps 3 4 twice and tabulate the readings Diskeeper 2 corporation Vlocity 3 Performance Testing Guide Space Reclamation Testing 1 Create fragmentation on 40GB volume with moderate fragments and record virtual disk size 2 Delete a significant portion of the files on the disk and record size of deleted files 3 Install V locity 3 Guest and enable Automatic Optimization 4 Perform a Quick Storage Migration Hyper V or Storage VMotion ESX ESXi of the test VM and afterwards record virtual disk
15. volume from V locity 3 Save the Job Report which shows the state of fragmentation of the volume and the degradation it is causing on the read access times on the volume 4 Optional Run any SAN optimization routines Such as reallocate and then rerun the lometer benchmarks These operations will not affect the degree of fragmentation of the iobw tst file 5 Delete the LUN and clear the SAN cache otherwise the comparison result when a non fragmented file is tested will be skewed 6 Create anew LUN that maps to the same location as best as possible 7 Re run lometer with the same settings This time do not run the file fragmentation utility 8 Once lometer testing is complete run a manual fragmentation analysis and save the Job Report again This is done to validate that little to no fragmentation exists on the iobw tst file 9 Now compare the before and after data from lometer and the V locity 3 fragmentation analyses 31 End access specifications 32 Results 33 Target Type Target Name Access Specification Name Managers Workers Disks lOps Read Ops Write Ops MBps Read MBp Write MB Transactic Connectic Average R Average R Ave 34 ALL All 1 1 1 292 4578 147 0827 145 375139 3 747541 1 89094 1856601 292 4578 0 3 417895 4 35749 2 35 MANAGER BURWSTOO3 Allin one 1 1 292 4578 147 0827 145 375139 3 747541 1 89094 18560601 292 4578 0 3 417895 4 35749 2 36 PROCESSOR CPU 0 37 PROCESSOR CPU1 38 WORKER Worker 1 Allin one 1 292 4
Download Pdf Manuals
Related Search