<html xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office" xmlns:w="urn:schemas-microsoft-com:office:word" xmlns:m="http://schemas.microsoft.com/office/2004/12/omml" xmlns:mv="http://macVmlSchemaUri" xmlns="http://www.w3.org/TR/REC-html40">
<head>
<meta http-equiv="Content-Type" content="text/html; charset=us-ascii">
<meta name="Title" content="">
<meta name="Keywords" content="">
<meta name="Generator" content="Microsoft Word 15 (filtered medium)">
<!--[if !mso]><style>v\:* {behavior:url(#default#VML);}
o\:* {behavior:url(#default#VML);}
w\:* {behavior:url(#default#VML);}
.shape {behavior:url(#default#VML);}
</style><![endif]--><style><!--
/* Font Definitions */
@font-face
{font-family:"Cambria Math";
panose-1:2 4 5 3 5 4 6 3 2 4;}
@font-face
{font-family:Calibri;
panose-1:2 15 5 2 2 2 4 3 2 4;}
@font-face
{font-family:Consolas;
panose-1:2 11 6 9 2 2 4 3 2 4;}
/* Style Definitions */
p.MsoNormal, li.MsoNormal, div.MsoNormal
{margin:0in;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:"Times New Roman";}
a:link, span.MsoHyperlink
{mso-style-priority:99;
color:blue;
text-decoration:underline;}
a:visited, span.MsoHyperlinkFollowed
{mso-style-priority:99;
color:purple;
text-decoration:underline;}
p
{mso-style-priority:99;
margin:0in;
margin-bottom:.0001pt;
font-size:12.0pt;
font-family:"Times New Roman";}
span.apple-style-span
{mso-style-name:apple-style-span;}
p.emailquote, li.emailquote, div.emailquote
{mso-style-name:emailquote;
margin-top:0in;
margin-right:0in;
margin-bottom:0in;
margin-left:1.0pt;
margin-bottom:.0001pt;
border:none;
padding:0in;
font-size:12.0pt;
font-family:"Times New Roman";}
span.EmailStyle20
{mso-style-type:personal-reply;
font-family:Calibri;
color:windowtext;}
span.msoIns
{mso-style-type:export-only;
mso-style-name:"";
text-decoration:underline;
color:teal;}
.MsoChpDefault
{mso-style-type:export-only;
font-size:10.0pt;}
@page WordSection1
{size:8.5in 11.0in;
margin:1.0in 1.0in 1.0in 1.0in;}
div.WordSection1
{page:WordSection1;}
--></style>
</head>
<body bgcolor="white" lang="EN-US" link="blue" vlink="purple">
<br>
John,<br>
<br>
Just to clarify a little further. Your comment about the inode lock is correct (keeping in mind the comment from Andreas about more recent changes), but Lustre (in buffered, rather than O_DIRECT mode)returns to user space before the data is down on the server.
It maintains the pages in the page cache on the client until it knows for sure it will not have to resend them. This means that - in the inode lock case - while only one process can be actively writing data to the page cache at a given moment, Lustre returns
to userspace before the data is written out.<br>
<br>
This means that while (in the inode lock case) the 'writing pages in to the page cache and setting them up to go out' (which is nearly all of most write syscalls) can't overlap, the network and disk portions or Lustre I/Os can overlap. That's how multiple
writes can be in flight at once even with single entry for writing processes.<br>
<br>
This also applies to one process - it can get many I/Os in flight at once, and achieve much higher bandwidth by overlapping that time they spend in flight. So, buffered writing is generally >> direct writing. Extremely large write sizes can mitigate this,
as Andreas suggested.<br>
<br>
- Patrick<br>
<hr style="display:inline-block;width:98%" tabindex="-1">
<div id="divRplyFwdMsg" dir="ltr"><font face="Calibri, sans-serif" style="font-size:11pt" color="#000000"><b>From:</b> Dilger, Andreas <andreas.dilger@intel.com><br>
<b>Sent:</b> Friday, October 14, 2016 4:38:19 PM<br>
<b>To:</b> John Bauer; Riccardo Veraldi<br>
<b>Cc:</b> lustre-discuss@lists.lustre.org; Patrick Farrell<br>
<b>Subject:</b> Re: [lustre-discuss] Lustre on ZFS pooer direct I/O performance</font>
<div> </div>
</div>
<div>
<div class="WordSection1">
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri">John, with newer Lustre clients it is possible for multiple threads to submit non-overlapping writes concurrently (also not conflicting within a single page), see LU-1669 for details.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri">Even so, O_DIRECT writes need to be synchronous to disk on the OSS, as Patrick reports, because if the OSS fails before the write is on disk there is no cached copy of the data on the client
that can be used to resend the RPC.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri">The problem is that the ZFS OSD has very long transaction commit times for synchronous writes because it does not yet have support for the ZIL. Using buffered writes, or having very large
O_DIRECT writes (e.g. 40MB or larger) and large RPCs (4MB, or up to 16MB in 2.9.0) to amortize the sync overhead may be beneficial if you really want to use O_DIRECT.
<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri">Riccardo,<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri">The other potential issue is that you have 20 OSTs on a single OSS, which isn't going to have very good performance. Spreading the OSTs across multiple OSS nodes is going to improve your
performance significantly when there are multiple clients writing, as there will be N times the OSS network bandwidth, N times the CPU, N times the RAM. It only makes sense to have 20 OSTs/OSS if your workload is only a single client and you want the maximum
possible capacity for a given cost.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri"><o:p> </o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri">Is each OST a separate VDEV and separate zpool, or are they a single zpool? Separate zpools have less overhead for maximum performance, but only one VDEV per zpool means that metadata
ditto blocks are written twice per RAID-Z2 VDEV, which isn't very efficient. Having at least 3 VDEVs per zpool is better in this regard.<o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri"><o:p> </o:p></span></p>
<div>
<div>
<div>
<p class="MsoNormal"><span style="font-size:10.5pt;font-family:Calibri;color:black">Cheers, Andreas<o:p></o:p></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:10.5pt;font-family:Calibri;color:black">-- <o:p></o:p></span></p>
</div>
<div>
<p class="MsoNormal"><span style="font-size:10.5pt;font-family:Calibri;color:black">Andreas Dilger<o:p></o:p></span></p>
</div>
</div>
<div>
<p class="MsoNormal"><span style="font-size:10.5pt;font-family:Calibri;color:black">Lustre Principal Architect<o:p></o:p></span></p>
</div>
</div>
<p class="MsoNormal"><span style="font-size:10.5pt;font-family:Calibri;color:black">Intel High Performance Data Division</span><span style="font-size:11.0pt;font-family:Calibri"><o:p></o:p></span></p>
<p class="MsoNormal"><span style="font-size:11.0pt;font-family:Calibri"><o:p> </o:p></span></p>
<div>
<div>
<div>
<p class="MsoNormal" style="margin-left:.5in"><span style="font-size:11.0pt;font-family:Consolas;color:black">On 2016/10/14, 15:22, "John Bauer" <<a href="mailto:bauerj@iodoctors.com">bauerj@iodoctors.com</a>> wrote:<o:p></o:p></span></p>
</div>
</div>
</div>
<div>
<p class="MsoNormal" style="margin-left:.5in"><o:p> </o:p></p>
</div>
<div>
<p class="MsoNormal" style="margin-left:.5in">Patrick<o:p></o:p></p>
</div>
<div id="AppleMailSignature">
<p class="MsoNormal" style="margin-left:.5in">I thought at one time there was an inode lock held for the duration of the direct I/O read or write. So that even if one had multiple application threads writing direct, only one was "in flight" at a time. Has that
changed?<o:p></o:p></p>
</div>
<div id="AppleMailSignature">
<p class="MsoNormal" style="margin-left:.5in">John<br>
<br>
Sent from my iPhone<o:p></o:p></p>
</div>
<div>
<p class="MsoNormal" style="mso-margin-top-alt:0in;margin-right:0in;margin-bottom:12.0pt;margin-left:.5in">
<br>
On Oct 14, 2016, at 3:16 PM, Patrick Farrell <<a href="mailto:paf@cray.com">paf@cray.com</a>> wrote:<o:p></o:p></p>
</div>
<blockquote style="margin-top:5.0pt;margin-bottom:5.0pt">
<div>
<div id="divtagdefaultwrapper">
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black">Sorry, I phrased one thing wrong:<br>
I said "transferring to the network", but it's actually until it's received confirmation the data has been received successfully, I believe.<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black"><o:p> </o:p></span></p>
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black">In any case, only one I/O (per thread) can be outstanding at a time with direct I/O.<o:p></o:p></span></p>
</div>
<div class="MsoNormal" align="center" style="margin-left:.5in;text-align:center">
<hr size="2" width="98%" align="center">
</div>
<div id="divRplyFwdMsg">
<p class="MsoNormal" style="margin-left:.5in"><b><span style="font-size:11.0pt;font-family:Calibri;color:black">From:</span></b><span style="font-size:11.0pt;font-family:Calibri;color:black"> lustre-discuss <<a href="mailto:lustre-discuss-bounces@lists.lustre.org">lustre-discuss-bounces@lists.lustre.org</a>>
on behalf of Patrick Farrell <<a href="mailto:paf@cray.com">paf@cray.com</a>><br>
<b>Sent:</b> Friday, October 14, 2016 3:12:22 PM<br>
<b>To:</b> Riccardo Veraldi; <a href="mailto:lustre-discuss@lists.lustre.org">lustre-discuss@lists.lustre.org</a><br>
<b>Subject:</b> Re: [lustre-discuss] Lustre on ZFS pooer direct I/O performance</span>
<o:p></o:p></p>
<div>
<p class="MsoNormal" style="margin-left:.5in"> <o:p></o:p></p>
</div>
</div>
<div>
<div>
<div id="x_divtagdefaultwrapper">
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black">Riccardo,<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black"><o:p> </o:p></span></p>
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black">While the difference is extreme, direct I/O write performance will always be poor. Direct I/O writes cannot be asynchronous, since they don't use the page cache. This means Lustre cannot
return from one write (and start the next) until it has finished transferring the data to the network.<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black"><o:p> </o:p></span></p>
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black">This means you can only have one I/O in flight at a time. Good write performance from Lustre (or any network filesystem) depends on keeping a lot of data in flight at once.<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black"><o:p> </o:p></span></p>
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black">What sort of direct write performance were you hoping for? It will never match that 800 MB/s from one thread you see with buffered I/O.<o:p></o:p></span></p>
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black"><o:p> </o:p></span></p>
<p style="margin-left:.5in"><span style="font-family:Calibri;color:black">- Patrick<o:p></o:p></span></p>
</div>
<div class="MsoNormal" align="center" style="margin-left:.5in;text-align:center">
<hr size="2" width="98%" align="center">
</div>
<div id="x_divRplyFwdMsg">
<p class="MsoNormal" style="margin-left:.5in"><b><span style="font-size:11.0pt;font-family:Calibri;color:black">From:</span></b><span style="font-size:11.0pt;font-family:Calibri;color:black"> lustre-discuss <<a href="mailto:lustre-discuss-bounces@lists.lustre.org">lustre-discuss-bounces@lists.lustre.org</a>>
on behalf of Riccardo Veraldi <<a href="mailto:Riccardo.Veraldi@cnaf.infn.it">Riccardo.Veraldi@cnaf.infn.it</a>><br>
<b>Sent:</b> Friday, October 14, 2016 2:22:32 PM<br>
<b>To:</b> <a href="mailto:lustre-discuss@lists.lustre.org">lustre-discuss@lists.lustre.org</a><br>
<b>Subject:</b> [lustre-discuss] Lustre on ZFS pooer direct I/O performance</span>
<o:p></o:p></p>
<div>
<p class="MsoNormal" style="margin-left:.5in"> <o:p></o:p></p>
</div>
</div>
</div>
<div>
<p class="MsoNormal" style="margin-left:.5in"><span style="font-size:10.0pt">Hello,<br>
<br>
I would like how may I improve the situation of my lustre cluster.<br>
<br>
I have 1 MDS and 1 OSS with 20 OST defined.<br>
<br>
Each OST is a 8x Disks RAIDZ2.<br>
<br>
A single process write performance is around 800MB/sec<br>
<br>
anyway if I force direct I/O, for example using oflag=direct in dd, the <br>
write performance drop as low as 8MB/sec<br>
<br>
with 1MB block size. And each write it's about 120ms latency.<br>
<br>
I used these ZFS settings<br>
<br>
options zfs zfs_prefetch_disable=1<br>
options zfs zfs_txg_history=120<br>
options zfs metaslab_debug_unload=1<br>
<br>
i am quite worried for the low performance.<br>
<br>
Any hints or suggestions that may help me to improve the situation ?<br>
<br>
<br>
thank you<br>
<br>
<br>
Rick<br>
<br>
<br>
_______________________________________________<br>
lustre-discuss mailing list<br>
<a href="mailto:lustre-discuss@lists.lustre.org">lustre-discuss@lists.lustre.org</a><br>
<a href="http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org">http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org</a><o:p></o:p></span></p>
</div>
</div>
</div>
</blockquote>
<blockquote style="margin-top:5.0pt;margin-bottom:5.0pt">
<div>
<p class="MsoNormal" style="mso-margin-top-alt:0in;margin-right:0in;margin-bottom:12.0pt;margin-left:.5in">
_______________________________________________<br>
lustre-discuss mailing list<br>
<a href="mailto:lustre-discuss@lists.lustre.org">lustre-discuss@lists.lustre.org</a><br>
<a href="http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org">http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org</a><o:p></o:p></p>
</div>
</blockquote>
</div>
</div>
</body>
</html>