<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content="text/html; charset=iso-8859-1" http-equiv=Content-Type>
<META name=GENERATOR content="MSHTML 9.00.8112.16421">
<STYLE>@font-face {
font-family: 宋体;
}
@font-face {
font-family: Verdana;
}
@font-face {
font-family: @宋体;
}
@page Section1 {size: 595.3pt 841.9pt; margin: 72.0pt 90.0pt 72.0pt 90.0pt; layout-grid: 15.6pt; }
P.MsoNormal {
TEXT-JUSTIFY: inter-ideograph; TEXT-ALIGN: justify; MARGIN: 0cm 0cm 0pt; FONT-FAMILY: "Times New Roman"; FONT-SIZE: 10.5pt
}
LI.MsoNormal {
TEXT-JUSTIFY: inter-ideograph; TEXT-ALIGN: justify; MARGIN: 0cm 0cm 0pt; FONT-FAMILY: "Times New Roman"; FONT-SIZE: 10.5pt
}
DIV.MsoNormal {
TEXT-JUSTIFY: inter-ideograph; TEXT-ALIGN: justify; MARGIN: 0cm 0cm 0pt; FONT-FAMILY: "Times New Roman"; FONT-SIZE: 10.5pt
}
A:link {
COLOR: blue; TEXT-DECORATION: underline
}
SPAN.MsoHyperlink {
COLOR: blue; TEXT-DECORATION: underline
}
A:visited {
COLOR: purple; TEXT-DECORATION: underline
}
SPAN.MsoHyperlinkFollowed {
COLOR: purple; TEXT-DECORATION: underline
}
SPAN.EmailStyle17 {
FONT-STYLE: normal; FONT-FAMILY: Verdana; COLOR: windowtext; FONT-WEIGHT: normal; TEXT-DECORATION: none; mso-style-type: personal-compose
}
DIV.Section1 {
page: Section1
}
UNKNOWN {
FONT-SIZE: 10pt
}
BLOCKQUOTE {
MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px; MARGIN-LEFT: 2em
}
OL {
MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px
}
UL {
MARGIN-TOP: 0px; MARGIN-BOTTOM: 0px
}
</STYLE>
</HEAD>
<BODY style="MARGIN: 10px; FONT-FAMILY: verdana; FONT-SIZE: 10pt">
<DIV><FONT color=#000080 size=3 face=Verdana>thanks, Mohit.</FONT></DIV>
<DIV><FONT color=#000080 size=3 face=Verdana>yes, I run "dd" on what you
said.</FONT></DIV>
<DIV><FONT color=#000080 size=3>what's the max known ops of
gluster?</FONT></DIV>
<DIV><FONT color=#000080 size=3 face=Verdana>I will try with your advice, more
threads and clients.</FONT></DIV>
<DIV><FONT color=#000080 size=2 face=Verdana></FONT> </DIV>
<DIV><FONT color=#c0c0c0 size=2 face=Verdana></FONT> </DIV>
<DIV><FONT size=2 face=Verdana><STRONG>>From:</STRONG> Mohit Anchlia
</FONT></DIV>
<DIV><FONT size=2 face=Verdana><STRONG>>Send Time:
</STRONG> 2011-05-18 12:25:45 </FONT></DIV>
<DIV><FONT size=2 face=Verdana><STRONG>>TO:</STRONG> nuaa_liuben
</FONT></DIV>
<DIV><FONT size=2 face=Verdana><STRONG>>CC: </STRONG> gluster-users
</FONT></DIV>
<DIV><FONT size=2 face=Verdana><STRONG>>Subject:</STRONG> Re: Re:
[Gluster-users] Gluster 3.2 for lots of small files </FONT></DIV>
<DIV><FONT size=2 face=Verdana></FONT></DIV>
<DIV><FONT size=2 face=Verdana><STRONG>> </STRONG>
<DIV>single threaded testing will always limit your ability to get any</DIV>
<DIV>higher throughput. Can you try multiple threads of say 10 from one</DIV>
<DIV>client and add/remove as you see fit?</DIV>
<DIV></DIV>
<DIV>How did you arrive at 30MB/s throughput requirement?? Reason I ask is</DIV>
<DIV>that in most cases it's not well thought through and you try to test</DIV>
<DIV>way way more than what's required. Do you get 5000 operations/sec,</DIV>
<DIV>with 6k file and 30MB/s this is what you would see. If that's the case</DIV>
<DIV>I don't think gluster will scale unless you add more nodes and use</DIV>
<DIV>lots of clients to test.</DIV>
<DIV></DIV>
<DIV></DIV>
<DIV>Where did you run "dd"? On gluster client mount?</DIV>
<DIV></DIV>
<DIV>On Tue, May 17, 2011 at 7:24 PM, nuaa_liuben <nuaa_liuben@sina.com> wrote:</DIV>
<DIV>> Avg. files size is between 2 ~ 6KB, both text and binary files.</DIV>
<DIV>> 1) about 30 MB/s for singel client, this can be realized?</DIV>
<DIV>> 2) dd with 4KB can get about 30 - 40MB/s</DIV>
<DIV>> 3) multiple clients can near linearly increase performance</DIV>
<DIV>> 4) only use single thread for testing</DIV>
<DIV>></DIV>
<DIV>> BR,</DIV>
<DIV>> liuben</DIV>
<DIV>></DIV>
<DIV>>>From: Mohit Anchlia</DIV>
<DIV>>>Send Time: 2011-05-18 09:54:45</DIV>
<DIV>>>TO: nuaa_liuben</DIV>
<DIV>>>CC: gluster-users</DIV>
<DIV>>>Subject: Re: [Gluster-users] Gluster 3.2 for lots of small files</DIV>
<DIV>>></DIV>
<DIV>> You will find difficult to tune when using small files. But what you</DIV>
<DIV>> are seeing is really low. Is your average file size 10k? Are these</DIV>
<DIV>> text files?</DIV>
<DIV>> 1) of all what's your throughput requirements??</DIV>
<DIV>> 2) try running "dd" tests with oflag/iflag=direct</DIV>
<DIV>> 3) run multiple clients from separate hosts and see if you add a</DIV>
<DIV>> client does it linearly increase the performance.</DIV>
<DIV>> 4) when using small files increasing number of parallel threads from</DIV>
<DIV>> same client will not increase performance. lower the threads and see</DIV>
<DIV>> if that helps.</DIV>
<DIV>> Post results.</DIV>
<DIV>> On Tue, May 17, 2011 at 6:23 PM, nuaa_liuben <nuaa_liuben@sina.com> wrote:</DIV>
<DIV>>> Hi,</DIV>
<DIV>>></DIV>
<DIV>>> I am testing gluster 3.2 for LOSF (lots of small files) scenario recently,</DIV>
<DIV>>> with iozne and postmark benchmark tools. And the results show that LOSF</DIV>
<DIV>>> performance is really bad, random read/write rate is just only about 2 ~</DIV>
<DIV>>> 4MB/s. I tried RAID0, RAID1, RAID10, and it's the same result. And create</DIV>
<DIV>>> ops with postmark is 100 ~ 400 per second.</DIV>
<DIV>>></DIV>
<DIV>>> I deployed 4 bricks servers: Dell R510, 4 1TB SATA disks, 1 GiE NIC, 8GB</DIV>
<DIV>>> Mem, and test machine is also R510 with 2 SAS disks and 16GB memory.</DIV>
<DIV>>></DIV>
<DIV>>> Any good ideas on gluster3.2 tunning for LOSF?</DIV>
<DIV>>> Why dbd translator is removed from gluster 3.2?</DIV>
<DIV>>> thanks in advance.</DIV>
<DIV>>></DIV>
<DIV>>></DIV>
<DIV>>> BR,</DIV>
<DIV>>> liuben</DIV>
<DIV>>> _______________________________________________</DIV>
<DIV>>> Gluster-users mailing list</DIV>
<DIV>>> Gluster-users@gluster.org</DIV>
<DIV>>> http://gluster.org/cgi-bin/mailman/listinfo/gluster-users</DIV>
<DIV>>></DIV>
<DIV>>></DIV>
<DIV></DIV></FONT></DIV></BODY></HTML>