[Disksim-users] 回复: The question about Request Size in DiskSim/FlashSim

shzb89 at yahoo.cn
Sun Jan 6 02:56:00 EST 2013


Hi,Wang Wei. thank you for your reply.
I have some questions about the "average request size","write ratio","sequential ratio" and "average inter-arrival time" of one trace. As most papers had cited, write ratio and average inter-arrival time can be get from the simulation output file. And they are the same as the one get from the original trace. So, why the average request size can not get from the output file? I think the simulator just give the statistics about the trace. But I don't know why the average request size is wrong.
In addition, how can i get the "sequential ratio" of one trace? what's the metrics for sequential request? That is, how to judge whether a request(or requests) is sequential?
Sorry for my poor English. 


________________________________
 发件人: Wei Wang <wangw8210 at gmail.com>
收件人: 沈 <shzb89 at yahoo.cn> 
抄送: "disksim-users at ece.cmu.edu" <disksim-users at ece.cmu.edu> 
发送日期: 2013年1月6日, 星期日, 12:46 下午
主题: Re: [Disksim-users] The question about Request Size in DiskSim/FlashSim
 

average request size is calculated by the statistics of original trace. it has nothing to do with the simulator.

Wei


On Sat, Jan 5, 2013 at 6:12 PM, 沈 <shzb89 at yahoo.cn> wrote:

Hi,all.
>Recently,I use FlashSim to do some simulations. Now, I want to know the average request size of Financial1 trace. From the paper such as "WAFTL: A Workload Adaptive Flash Translation Layer with Data Partition","DFTL",they give the average request size, I don't know how they get the value.I have two ways to get the value, but I don't know if it is right. 
>1.From the output file of Flashsim, there are "Request size stats","Read Request size stats", "Write Request size stats",I can get the average value of average request size, I think the unit size is the same as the ascii trace, i.e.,512bytes(in the statdefs file, the Request size entry, the Scale is 1).
>2.I get the value through caculate the average of the "bcount" in the trace.
>I don't know which is right.
>
>
>In addition, I found that the request size is just not the same as the bcount in the trace. For example,
>Overall I/O System Request size distribution
>   =  1    =  2    =  3    =  4    <  7    <  9    < 13    < 17    < 21     21+
>      0       0       0      24       0       0       0  158986  474238  422200 
>but in Financial1 trace,after I convert it into the ascii format, the value of bcount can be 1,2,3,4 and so on. So the request size in DiskSim/FlashSim is not the same as the one in ascii trace, is that right?
>Can anyone help me? Thank you in advance.
>_______________________________________________
>Disksim-users mailing list
>Disksim-users at ece.cmu.edu
>https://sos.ece.cmu.edu/mailman/listinfo/disksim-users
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.andrew.cmu.edu/pipermail/disksim-users/attachments/20130106/1fe70c5c/attachment.html>


More information about the Disksim-users mailing list