[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [ogfs-dev]Lustre's DLM [was Lustre vs. OGFS]



On Mon, 2003-08-25 at 21:03, Kip Macy wrote:
> See www.lustre.org - as best as I can tell they 
> are trying to solve different problems.
> 
> GFS is geared towards a relatively small cluster, whereas Lustre
> is claimed to be able to serve clusters with 10,000's of nodes.
> The configuration of Lustre appears to be quite a bit more complicated
> as well.
> 
> 
> 			-Kip

I did just go look at the Lustre factsheet:

    http://www.lustre.org/docs/lustre-datasheet.pdf

It definately looks interesting, but not competitive to ogfs.  Much more
like microsoft's dfs, not really a cfs at all in my opinion.

One very interesting line was:

"The MetaData servers provide enormous scalability and performance
through intent lock operations introduced by CFS and adopted in Linux
2.6."

Does anyone know what the above is talking about and if it has any
relevance for ogfs.

Also, per the factsheet Lustre has a full-fledged Posix compliant DLM
that is GPL'ed.

Since OpenDLM seems to be be dead and to have problems, maybe we should
evalute using the DLM from Lustre?

If I read the website correctly, they recently dropped 2.4 support so
their DLM may work with 2.6 kernel only.

Greg
-- 
Greg Freemyer






-------------------------------------------------------
This sf.net email is sponsored by:ThinkGeek
Welcome to geek heaven.
http://thinkgeek.com/sf
_______________________________________________
Opengfs-devel mailing list
Opengfs-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/opengfs-devel

[Kernel]     [Security]     [Bugtraq]     [Photo]     [Yosemite]     [MIPS Linux]     [ARM Linux]     [Linux Clusters]     [Linux RAID]     [Yosemite Hiking]     [Linux Resources]

Powered by Linux