<HTML>
<HEAD>
<TITLE>Re: [Lustre-devel] How store HSM metadata in MDT ?</TITLE>
</HEAD>
<BODY>
<FONT FACE="Calibri, Verdana, Helvetica, Arial"><SPAN STYLE='font-size:11pt'><BR>
<BR>
<BR>
On 7/11/08 8:32 AM, "Jacques-Charles Lafoucriere" <<a href="jc.lafoucriere@cea.fr">jc.lafoucriere@cea.fr</a>> wrote:<BR>
<BR>
</SPAN></FONT><BLOCKQUOTE><FONT FACE="Calibri, Verdana, Helvetica, Arial"><SPAN STYLE='font-size:11pt'><BR>
<BR>
Peter Braam wrote:<BR>
</SPAN></FONT><BLOCKQUOTE><FONT FACE="Calibri, Verdana, Helvetica, Arial"><SPAN STYLE='font-size:11pt'> <BR>
</SPAN></FONT><BLOCKQUOTE><FONT FACE="Calibri, Verdana, Helvetica, Arial"><SPAN STYLE='font-size:11pt'> <BR>
</SPAN></FONT><BLOCKQUOTE><FONT FACE="Calibri, Verdana, Helvetica, Arial"><SPAN STYLE='font-size:11pt'> <BR>
3. how are multiple coordinators and agents utilized to distribute load so<BR>
that the HSM can keep up with massive small file creation?<BR>
<BR>
<BR>
</SPAN></FONT></BLOCKQUOTE><FONT FACE="Calibri, Verdana, Helvetica, Arial"><SPAN STYLE='font-size:11pt'> <BR>
One coordinator per MDT.<BR>
<BR>
<BR>
</SPAN></FONT></BLOCKQUOTE><FONT FACE="Calibri, Verdana, Helvetica, Arial"><SPAN STYLE='font-size:11pt'> <BR>
<BR>
No - these must be independent considerations. A coordinator may be much<BR>
slower than an MDS node in handling a single file. I say this because this<BR>
has been the experience in the industry so far - with small files the HSM<BR>
can not at all keep up.<BR>
<BR>
</SPAN></FONT></BLOCKQUOTE><FONT FACE="Calibri, Verdana, Helvetica, Arial"><SPAN STYLE='font-size:11pt'><BR>
Why a coordinator should be on a slower node than a MDS ?<BR>
Coordinator is a Lustre service like other Lustre services so it will be on a right hardware<BR>
<BR>
</SPAN></FONT></BLOCKQUOTE><FONT FACE="Calibri, Verdana, Helvetica, Arial"><SPAN STYLE='font-size:11pt'>I see no reason whatsoever to couple them to MDT’s. Keeping them de-coupled is more flexible and can scale better. Am I missing issues here?<BR>
<BR>
As for performance the following: a coordinator may have a lot of work to do to track which files still need to be handled by agents and which are in progress already and may have a fair amount of interaction with agents (not with HSM to tape but when a coordinator is handling a re-striping migration it will). But its interaction with the MDT’s is very limited. So we could degrade the performance of an MDS by placing this on the same node. Let’s keep this flexible please.<BR>
<BR>
Peter<BR>
</SPAN></FONT><BLOCKQUOTE><FONT FACE="Calibri, Verdana, Helvetica, Arial"><SPAN STYLE='font-size:11pt'><BR>
<BR>
Do you mean a coordinator is not part of the Lustre cluster ?<BR>
<BR>
JC<BR>
<BR>
</SPAN></FONT></BLOCKQUOTE>
</BODY>
</HTML>