[Beowulf] OS for 64 bit AMD
Many of your questions may have already been answered in earlier discussions or in the FAQ. The search results page will indicate current discussions as well as past list serves, articles, and papers.
Craig Tierney ctierney at HPTI.comFri Apr 1 11:25:22 PST 2005
- Previous message: [Beowulf] Compiling HPL
- Next message: [Beowulf] OS for 64 bit AMD
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
On Thu, 2005-03-31 at 13:38, David Kewley wrote: > Joe Landman wrote on Thursday 31 March 2005 10:04: > > Actually Redhat now has HPC pricing per node. There are other good > > reasons to look elsewhere for HPC distributions though, specifically due > > to their lack of good high performance/scalable per-node file system. > > In your experience/knowledge, what are the issues that make e.g. xfs better > than ext3 on RHEL4? I ask specifically about RHEL4 because RH has worked > hard on ext3 in time for RHEL4. > > My understanding is that RH choose to support ext3 but not xfs because: 1) > they have in-house expertise for ext3 but not for xfs, and 2) they believe > that xfs has no real advantages over ext3. > > If customers show RH that there are real-life needs for xfs that are not > satisfied by ext3, then RH may well be willing to invest in in-house xfs > expertise. > Is the work they are doing in the kernel.org kernel yet? I know that the CFS (Lustre) has been working on improving ext3 as well. I have tried ext3 a bit under SuSE 9.2, and I still prefer ext3. Maybe RH has some different code that addresses my needs. 1) Deletes under ext3 are slow. Not a big deal for a few files but when you need to delete many at one time it is annoying. 2) Streaming performance of xfs for very high performance storage (hundreds of MB/s) is better. 3) xfs dynamically creates inodes, so I don't have to worry about that at creation time. This is a big deal for diskless cluster installations. I don't create filesystems all the time, and it only needs to be done once, but it is real annoying to wait 4-8 hours for a mkfs to finish under ext3 for multi-terabyte filesystems. For the 2.4 kernel, I found that NFS was faster with xfs as a base than ext3. It also held up under load better. It seemed like xfs dealt with failures better, but that was my observation more than fact. This may have changed though with 2.6 and Redhat's kernel. Since I have already committed to xfs, and ext3 isn't any better I don't need to look to swtich back. Craig > David > _______________________________________________ > Beowulf mailing list, Beowulf at beowulf.org > To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf
- Previous message: [Beowulf] Compiling HPL
- Next message: [Beowulf] OS for 64 bit AMD
- Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]
More information about the Beowulf mailing list
