cosbench-openstack.pdf
4 likes3,672 views
The document describes COSBench, a benchmark tool for evaluating the performance of cloud object storage services. It provides an overview of COSBench's key components, including its configurable workload definition file, controller for managing tests, and drivers for generating load. The document also shares sample results from using COSBench to measure the throughput and response times of OpenStack Swift in different configurations. It found that the proxy node's CPU was the bottleneck for larger workloads on one setup. The goal is to open source COSBench to help storage providers optimize performance.
cosbench-openstack.pdf
- 1. COSBench: A benchmark Tool for Cloud Object Storage Services Jiangang.Duan@intel.com 2012.10 Updated June 2012
- 2. Agenda • Self introduction • COSBench Introduction • Case Study to evaluate OpenStack* swift performance with COSBench • Next Step plan and Summary
- 3. Self introduction • Jiangang Duan • Working in Cloud Infrastructure Technology Team (CITT) of Intel APAC R&D Ltd. (shanghai) • We are software team, Experienced at performance • To understand how to build an efficient/scale Cloud Solution with Open Source software (OpenStack*, Xen*, KVM*) • All of work will be contributed back to Community • Today we will talk about some efforts we try to measure OpenStack* Swift performance
- 4. COSBench Introduction • COSBench is an Intel developed Benchmark to measure Cloud Object Storage Service performance – For S3, OpenStack Swift like Object Storage – Not for File system (NFS e.g) and Block Device system (EBS e.g.) • Requirement of a benchmark to measure Object Store performance • Cloud solution provider can use it to – Compare different Hardware/Software Stacks – Identify bottleneck and make optimization • We are in progress to Open Source COSBench COSBench is the IOMeter for Cloud Object Storage service
- 5. COSBench Key Component Config.xml: • define workload with flexibility. Web Console Controller: Controller • Control all drivers Config.xml • Collect and aggregate stats. COSBench Driver: • generate load w/ config.xml parameters. Driver • can run tests w/o controller. Driver Web Console: Controller • Manage controller Node • Browse real-time stats • Communication is based on HTTP Storage Cloud (RESTful style) Storage Node
- 6. Web Console Driver list Workload List History list Intuitive UI to get Overview.
- 7. Workload Configuration Flexible load control object size distribution Read/Write Operations Workflow for complex stages Flexible configuration parameters is capable of complex Cases
- 8. Performance Metrics • Throughput (Operations/s): the operations completed in one second • Response Time (in ms): the duration between operation initiation and completion. • Bandwidth (KB/s): the total data in KiB transferred in one second • Success Ratio (%): the ratio of successful operations
- 9. Easy to be Extended • Easily extend for new storage system: • Support AuthAPI – OpenStack Swift Auth – Amplistor* – Adding More Context PUT GET DELETE StorageAPI Easy execution engine is capable of complex cases.
- 10. Setup-SATA has higher CPU power System Configuration Setup-SAS has faster disks Client Node Client Node Client Node Client Node Client Node 2GbE 2GbE 2GbE 2GbE 2GbE Setup-SATA Both Setup - CPU: 2 * Client - CPU: 2 * 2.93GHz (4C/8T) Ethernet 2.7GHz (8C/16T) Network - MEM: 12GB DDR3 1333MHz 10GbE - NIC: 2 * Intel 82579 1GbE - MEM: 32GB DDR3 1333MHz - NIC: Intel 82599 10GbE bonding (mode=rr) Proxy Setup-SAS Node - CPU: 2 * Setup-SAS 2.3GHz (8C/16T) 10GbE - 12 * 70GB SAS (15000 rpm) - MEM: 64GB DDR3 1333MHz Storage - NIC: Intel 82599 10GbE Ethernet Network Setup-SATA - 14 * 1T SATA (5,400 rpm) 2GbE 2GbE 2GbE 2GbE 2GbE Storage Storage Storage Storage Storage Node Node Node Node Node
- 11. 128KB-Read • SLA: 200ms + 128KB/1MBps = 325ms # Worker 95%-ResTime Throughput Setup-SAS 128KB-Read ms op/s Throughput Avg-ResTime 95%-ResTime 5 20.00 369.49 6.0k 1000 10 20.00 711.24 5.02k 5.00k 4.95k 4.84k 4.69k response time (ms) 800 20 20.00 1383.30 throughput (op/s) 4.5k 3.66k 40 30.00 2517.94 600 3.0k 2.52k 80 46.67 3662.71 400 1.38k 160 56.67 4693.97 1.5k 0.71k 200 0.37k 320 106.67 5019.85 0.0k 0 640 230.00 4998.13 5 10 20 40 80 160 320 640 1280 2560 1280 470.00 4947.15 Total Number of Worker 2560 923.33 4840.19 The bottleneck was identified to be the proxy’s CPU -- The CPU utilization at that node was ~100%! -- The peak throughput for setup-SATA was 5576 op/s (640 workers) For more complete information about performance and Better CPU results in higher throughput benchmark results, visit Performance Test Disclosure
- 12. 128KB-Write • SLA: 200ms + 128KB/1MBps = 325ms # Worker 95%-ResTime Throughput Setup-SAS 128KB-Write ms op/s Throughput Avg-ResTime 95%-ResTime 5 40.00 219.73 2.0k 1.87k 1.89k 4000 1.77k 1.77k 10 40.00 391.14 1.59k response time (ms) 3200 throughput (op/s) 20 50.00 668.19 1.5k 1.33k 2400 40 70.00 1022.07 1.02k 1.0k 80 100.00 1333.34 0.67k 1600 0.5k 0.39k 160 143.33 1594.12 0.22k 800 320 370.00 1769.55 0.0k 0 640 1223.33 1773.12 5 10 20 40 80 160 320 640 1280 2560 1280 1690.00 1871.58 Total Number of Worker 2560 3160.00 1886.81 The Disks at storage nodes had significant impact on overall throughput -- The peak throughput for setup-SATA was only 155 op/s (20 Workers) -- even after we had put account/container DB files on SSD disks! For more complete information about performance and benchmark results, visit Performance Test Disclosure
- 13. 128KB-Write To fully understand the write performance … -- From the Disk side: were storage disk a bottleneck? -- in setup-SATA, all SSD 1621 op/s compared with 155 op/s -- Need do more tests to understand the disk impact -- From the NIC side: were storage network a bottleneck? -- in setup-SAS, two set of object daemon, NO performance change (why? TODO) -- in setup-SATA, all SSD + 32/64KB objects Storage node CPU bottleneck (why? TODO) -- From the SW side: were container updating a bottleneck? -- in setup-SATA, 1 container, put account/container from HDD to SSD 316% ↑ -- in setup-SATA, 100 containers, put account/container from HDD to SSD 119% ↑
- 14. 10MB-Read • SLA: 200ms + 10MB/1MBps = 1200ms # Worker 95%-ResTime Throughput Setup-SAS10MB -Read ms op/s Throughput Avg-ResTime 95%-ResTime 5 270.00 34.69 80 69.58 70.18 71.41 72.90 60,000 10 320.00 51.87 65.48 67.37 68.69 59.91 50,000 response time (ms) 20 480.00 59.91 throughput (op/s) 60 51.87 40,000 40 900.00 65.48 40 34.69 30,000 80 1636.67 67.37 20,000 160 3093.33 68.69 20 10,000 320 5950.00 69.58 0 0 640 11906.67 70.18 5 10 20 40 80 160 320 640 12802560 1280 24090.00 71.41 Total Number of Worker 2560 52090.00 72.90 The bottleneck was identified to be the clients’ NIC BandWidth Double client receive bandwidth can double the throughput For more complete information about performance and benchmark results, visit Performance Test Disclosure
- 15. 10MB-Write • SLA: 200ms + 10MB/1MBps = 1200ms # Worker 95%-ResTime Throughput Setup-SAS 10MB-Write ms op/s Throughput Avg-ResTime 95%-ResTime 5 536.67 13.12 23.71 25 21.38 22.21 23.19 23.23 21.55 250,000 10 936.67 17.50 20.28 21.30 response time (ms) 20 200,000 20 1596.67 20.28 throughput (op/s) 17.50 13.12 40 2786.67 21.30 15 150,000 80 5133.33 21.38 10 100,000 160 9800.00 22.21 5 50,000 320 18623.33 23.19 0 0 640 41576.67 23.23 5 10 20 40 80 160 320 640 12802560 1280 102090.00 21.55 Total Number of Worker 2560 200306.67 23.71 The bottleneck might be the storage nodes’ NICs -- in setup-SATA, the peak throughput was 15.74 op/s (10 clients) -- in both setups, the write performance was 1/3 of the read performance For more complete information about performance and benchmark results, visit Performance Test Disclosure
- 16. Next Step and call for action • Open source COSBench (WIP) – • Keep develop Cosbench to support more Object Storage Software • Take COSBench as a tool to analyze Cloud Object Service performance (swift and Ceph) • Contact me (jiangang.duan@intel.com) if you want to evaluate COSBench – and we are glad to hear your feedback to make it better
- 17. Summary • New storage Usage model rises for Cloud Computing age, which need new benchmark • COSBench is a new benchmark developed by Intel to measure Cloud Object Storage service performance • OpenStack Swift is stable/high open source performance object Storage implementation, still need improvement
- 18. Disclaimers • INFORMATION IN THIS DOCUMENT IS PROVIDED IN CONNECTION WITH INTEL PRODUCTS. NO LICENSE, EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. EXCEPT AS PROVIDED IN INTEL'S TERMS AND CONDITIONS OF SALE FOR SUCH PRODUCTS, INTEL ASSUMES NO LIABILITY WHATSOEVER AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO SALE AND/OR USE OF INTEL PRODUCTS INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT. • A "Mission Critical Application" is any application in which failure of the Intel Product could result, directly or indirectly, in personal injury or death. SHOULD YOU PURCHASE OR USE INTEL'S PRODUCTS FOR ANY SUCH MISSION CRITICAL APPLICATION, YOU SHALL INDEMNIFY AND HOLD INTEL AND ITS SUBSIDIARIES, SUBCONTRACTORS AND AFFILIATES, AND THE DIRECTORS, OFFICERS, AND EMPLOYEES OF EACH, HARMLESS AGAINST ALL CLAIMS COSTS, DAMAGES, AND EXPENSES AND REASONABLE ATTORNEYS' FEES ARISING OUT OF, DIRECTLY OR INDIRECTLY, ANY CLAIM OF PRODUCT LIABILITY, PERSONAL INJURY, OR DEATH ARISING IN ANY WAY OUT OF SUCH MISSION CRITICAL APPLICATION, WHETHER OR NOT INTEL OR ITS SUBCONTRACTOR WAS NEGLIGENT IN THE DESIGN, MANUFACTURE, OR WARNING OF THE INTEL PRODUCT OR ANY OF ITS PARTS. • Intel may make changes to specifications and product descriptions at any time, without notice. Designers must not rely on the absence or characteristics of any features or instructions marked "reserved" or "undefined". Intel reserves these for future definition and shall have no responsibility whatsoever for conflicts or incompatibilities arising from future changes to them. The information here is subject to change without notice. Do not finalize a design with this information. • The products described in this document may contain design defects or errors known as errata which may cause the product to deviate from published specifications. Current characterized errata are available on request. • Contact your local Intel sales office or your distributor to obtain the latest specifications and before placing your product order. • Copies of documents which have an order number and are referenced in this document, or other Intel literature, may be obtained by calling 1-800-548-4725, or go to: http://www.intel.com/design/literature.htm%20 • This document contains information on products in the design phase of development. • Intel and the Intel logo are trademarks of Intel Corporation in the U.S. and/or other countries. • *Other names and brands may be claimed as the property of others. • Copyright © 2012 Intel Corporation. All rights reserved.