Re: A question about the "ClusterData2011_2 traces"

[john wilkes]

Hi Binlei.  The data in the trace is what you have to work with.  You can try to deduce which jobs are batch jobs by rereading the Borg paper (carefully!) and looking at how long jobs run for, as well as their relative priorities and the scheduling class (called appclass in the Borg paper)  But all this is deduction.  We cannot provide you with additional information.

  john

On Sun, Nov 26, 2017 at 7:39 AM, adsert <ads...@126.com> wrote:

Dear John Wilkes, 

My name is Binlei Cai, and I'm a PhD student of Tianjin University. 

Now, I am analyzing the "ClusterData2011_2 traces" to motivate our study. I want to analyze the online services (e.g., BigTable, web search) and batch jobs,respectively. However, according the "Google cluster-usage traces: format +schema" , I can not distinguishs online services and batch jobs.

Can you help me, and tell me how to extract the information of online services from "ClusterData2011_2 traces" ?

Kind regards,

Binlei Cai

--

蔡斌雷

天津大学 计算机科学与技术学院

MT 15822219156 13075301359

通信地址：天津市 津南区 海河教育园区 天津大学北洋园校区