[DRAFT] Virtual Machine Performance Comparison of Public IaaS Providers in China
发表时间：2012-06-21 09:15:52This is a paper submitted for consideration of lecture presentation on the IEEE Asia Pacific Cloud Computing Congress 2012 in Shenzhen.
Abstract — We compare the virtual machine (VM) performance of two public IaaS providers (GrandCloud and Aliyun) in China. UnixBench and Hadoop wordcount are utilized to provide benchmark data for the comparison. It is found that VM specifications such as the number of CPU cores and the amount of memory can no longer be used as reference for VM performance. In both UnixBench and Hadoop wordcount tests, as the VM gets bigger, the performance / price ratio gets lower. In Hadoop wordcount tests, a cluster with several smaller VMs provides much better performance as performance / price ratio as compared to a bigger VM. It is recommended to practice horizontal scaling when an application needs more computing resource.
Keywords — Public IaaS, Aliyun, GrandCloud, performance comparison, UnixBench, Hadoop wordcount, vertical scaling, horizontal scaling
Download Full Paper: IEEE_Cloud_Computing_Conference_John
| 上一篇 | 下一篇 |
| 姓名： | |
| 评论： | |
|
请输入下面这首诗词的作者姓名。 杯汝来前！老子今朝，点检形骸。 |
|
| 答案： | |
云与清风常拥有，
冰雪知音世难求。
击节纵歌相对笑，
案上诗书杯中酒。
2000.12.31 于 洛杉矶
单凭这两个公有云服务提供商上进行的两组测试就得出“VM specifications such as the number of CPU cores and the amount of memory can no longer be used as reference for VM performance”，个人认为太仓促了。造成小型VM性价比高于大型VM的原因很多:比如定价策率，workload本身性质（是否能随着core/mem线性scale)。阿里云的定价策率本身就决定了小型VM更合算，那么在比较时候应该去掉这个因素；对于workload本身性质，只能依赖于加上native同配置，本地虚拟机同配置测试数据来说明问题。有不少workload是对core/mem scale不友好的(受限于应用程序本身，比如锁优化不好等），即使workload没问题，那测试环境的配置是否合理（OS，应用本身的配置可能造成不scale），这类因素也必须排除在外。 另外，workload的选择是否有代表性也非常重要，unixbench在我看来就不是什么很有代表性的应用。在公有云上测试，应该关注web server这种应用。 个人意见，仅供参考。
我正在考虑加入如下内容：（1）测试数据在24小时中的变化，（2）MySQL 基准测试，（3）MemCache基准测试，（4）Web Server 基准测试
你好请问如何联系，我们有一款Iaas产品，希望也请你测试下。谢谢。
quhaifeng@gmail.com
[...] http://www.qyjohn.net/?p=2180 //qyjohn主要在阿里云和盛大云上评测了UnixBench和Hadoop [...]