<html>
<head>
<meta http-equiv="Content-Type" content="text/html; charset=utf-8">
</head>
<body text="#000000" bgcolor="#FFFFFF">
<p>FWIW, <br>
</p>
<p>For OSTs, I like to use ZFS as the backing filesystem and then
use an SSD/NVME as the cache disk(s) for a zpool.</p>
<p>The MDT, depends on the expected data. If it is all large, I do a
minimal size using SSD/NVME. If there will be a significant number
of smaller files, I plan for multiple MDTs with larger sizes
(still using zfs as the backing filesystem).</p>
<p>Those are just general starting points that seem to work well for
the installations I have done where I have a little more freedom
in the architecture.<br>
</p>
<p>YMMV<br>
</p>
<p>Brian Andrus<br>
</p>
<br>
<div class="moz-cite-prefix">On 2/6/2018 9:32 AM, E.S. Rosenberg
wrote:<br>
</div>
<blockquote type="cite"
cite="mid:CA+K1OzRc-4cVGLWo24EKSv6UkPds+YqztLMS1xw8TkqD81cJbw@mail.gmail.com">
<div dir="ltr">
<div>
<div>
<div>
<div>
<div>Hello fellow Lustre users :)<br>
<br>
</div>
Since I didn't want to take the "size of MDT, inode
count, inode size" thread too far off-topic I'm starting
a new thread.<br>
<br>
</div>
I'm curious how many people are using SSD MDTs?<br>
Also how practical is such a thing in a 2.11.x Data On MDT
scenario?<br>
</div>
Is using some type of mix between HDD and SSD storage for
MDTs practical?<br>
</div>
<div>Does SSD vs HDD have an effect as far as ldiskfs vs zfs?<br>
<br>
</div>
Thanks!<br>
</div>
Eli<br>
</div>
<br>
<fieldset class="mimeAttachmentHeader"></fieldset>
<br>
<pre wrap="">_______________________________________________
lustre-discuss mailing list
<a class="moz-txt-link-abbreviated" href="mailto:lustre-discuss@lists.lustre.org">lustre-discuss@lists.lustre.org</a>
<a class="moz-txt-link-freetext" href="http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org">http://lists.lustre.org/listinfo.cgi/lustre-discuss-lustre.org</a>
</pre>
</blockquote>
<br>
</body>
</html>