[gpfsug-discuss] LROC Express

Sanchez, Paul Paul.Sanchez at deshaw.com
Tue Jun 23 15:10:31 BST 2015


Hi Sven,

Yes, I think that fileset level include/exclude would be sufficient for us.

It also begs the question about the same for write caching.  We haven’t experimented with it yet, but are looking forward to employing HAWC for scratch-like workloads.  Do you imagine providing the same sort of HAWC bypass include/exclude to be part of this?  That might be useful for excluding datasets where the write ingest rate isn’t massive and the degree of risk we’re comfortable with potential data recovery issues in the face of complex outages may be much lower.

Thanks,
Paul

From: gpfsug-discuss-bounces at gpfsug.org [mailto:gpfsug-discuss-bounces at gpfsug.org] On Behalf Of Sven Oehme
Sent: Monday, June 22, 2015 7:14 PM
To: gpfsug main discussion list
Subject: Re: [gpfsug-discuss] LROC Express


Hi Paul,

just out of curiosity, not that i promise anything, but would it be enough to support include/exclude per fileset level or would we need path and/or extension or even more things like owner of files as well ?

Sven

------------------------------------------
Sven Oehme
Scalable Storage Research
email: oehmes at us.ibm.com<mailto:oehmes at us.ibm.com>
Phone: +1 (408) 824-8904
IBM Almaden Research Lab
------------------------------------------

[Inactive hide details for "Sanchez, Paul" ---06/22/2015 03:57:29 PM---I can’t confirm whether it works with Express, since we]"Sanchez, Paul" ---06/22/2015 03:57:29 PM---I can’t confirm whether it works with Express, since we’re also running standard. But as a simple t

From: "Sanchez, Paul" <Paul.Sanchez at deshaw.com<mailto:Paul.Sanchez at deshaw.com>>
To: gpfsug main discussion list <gpfsug-discuss at gpfsug.org<mailto:gpfsug-discuss at gpfsug.org>>
Date: 06/22/2015 03:57 PM
Subject: Re: [gpfsug-discuss] LROC Express
Sent by: gpfsug-discuss-bounces at gpfsug.org<mailto:gpfsug-discuss-bounces at gpfsug.org>

________________________________



I can’t confirm whether it works with Express, since we’re also running standard. But as a simple test, I can confirm that deleting the gpfs.ext package doesn’t seem to throw any errors w.r.t. LROC in the logs at startup, and “mmdiag --lroc” looks normal when running without gpfs.ext (Standard Edition package). Since we have other standard edition features enabled, I couldn’t get far enough to actually test whether the LROC was still functional though.

In the earliest 4.1.0.x releases the use of LROC was confused with “serving NSDs” and so the use of the feature required a server license, and it did throw errors at startup about that. We’ve confirmed that in recent releases that this is no longer a limitation, and that it was indeed erroneous since the goal of LROC was pretty clearly to extend the capabilities of client-side pagepool caching.

LROC also appears to have some rate-limiting (queue depth mgmt?) so you end up in many cases getting partial file-caching after a first read, and a subsequent read can have a mix of blocks served from local cache and from NSD. Further reads can result in more complete local block caching of the file. One missing improvement would be to allow its use on a per-filesystem (or per-pool) basis. For instance, when backing a filesystem with a huge array of Flash or even a GSS/ESS then the performance benefit of LROC may be negligible or even negative, depending on the performance characteristics of the local disk. But against filesystems on archival media, LROC will almost always be a win. Since we’ve started to see features using tracked I/O access time to individual NSDs (e.g. readReplicaPolicy=fastest), there’s potential here to do something adaptive based on heuristics as well. Anyone else using this at scale and seeing a need for additional knobs?

Thx
Paul

From: gpfsug-discuss-bounces at gpfsug.org<mailto:gpfsug-discuss-bounces at gpfsug.org> [mailto:gpfsug-discuss-bounces at gpfsug.org] On Behalf Of Barry Evans
Sent: Monday, June 22, 2015 11:40 AM
To: gpfsug main discussion list
Subject: Re: [gpfsug-discuss] LROC Express

Hi Bob,

Thanks for this, just to confirm does this mean that it *does not* work with express?

Cheers,
Barry


Bob Oesterlin wrote:
It works with Standard edition, just make sure you have the right license for the nodes using LROC.

Bob Oesterlin
Nuance COmmunications


Bob Oesterlin


On Mon, Jun 22, 2015 at 10:28 AM, Barry Evans <bevans at pixitmedia.com<mailto:bevans at pixitmedia.com>> wrote:
Hi All,

Very quick question for those in the know - does LROC require a standard license, or will it work with Express? I can't find anything in the FAQ regarding this so I presume Express is ok, but wanted to make sure.

Regards,
Barry Evans
Technical Director
Pixit Media/ArcaStream



--

This email is confidential in that it is intended for the exclusive attention of the addressee(s) indicated. If you are not the intended recipient, this email should not be read or disclosed to any other person. Please notify the sender immediately and delete this email from your computer system. Any opinions expressed are not necessarily those of the company from which this email was sent and, whilst to the best of our knowledge no viruses or defects exist, no responsibility can be accepted for any loss or damage arising from its receipt or subsequent use of this email.
_______________________________________________
gpfsug-discuss mailing list
gpfsug-discuss at gpfsug.org<http://gpfsug.org/>
http://gpfsug.org/mailman/listinfo/gpfsug-discuss

_______________________________________________
gpfsug-discuss mailing list
gpfsug-discuss at gpfsug.org
http://gpfsug.org/mailman/listinfo/gpfsug-discuss


This email is confidential in that it is intended for the exclusive attention of the addressee(s) indicated. If you are not the intended recipient, this email should not be read or disclosed to any other person. Please notify the sender immediately and delete this email from your computer system. Any opinions expressed are not necessarily those of the company from which this email was sent and, whilst to the best of our knowledge no viruses or defects exist, no responsibility can be accepted for any loss or damage arising from its receipt or subsequent use of this email._______________________________________________
gpfsug-discuss mailing list
gpfsug-discuss at gpfsug.org
http://gpfsug.org/mailman/listinfo/gpfsug-discuss


-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://gpfsug.org/pipermail/gpfsug-discuss_gpfsug.org/attachments/20150623/59e9955d/attachment-0003.htm>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: image001.gif
Type: image/gif
Size: 105 bytes
Desc: image001.gif
URL: <http://gpfsug.org/pipermail/gpfsug-discuss_gpfsug.org/attachments/20150623/59e9955d/attachment-0003.gif>


More information about the gpfsug-discuss mailing list