Hello,

thank you all for a fast reply.

On Mon, 23 Apr 2012 14:40:23 -0400 Becky Ligon wrote:
> Please try the scenario where you have 2 servers, each handling data and
> metadata, no dedicated metadata server, and see if you get better
> performance in the 2 server scenario.

For 2 servers serving both data and metadata performance is about the
same, though distribution dispersion is high:
   KB  reclen   write rewrite    read    reread
 1024       4    6197    6400     6449     6371
 1024       8   10171   10715    10577    10420
 1024      16   14098   14711    18712    17967
 1024      32   24087   25862    31756    31394
 1024      64   37654   41712    46038    46936
 1024     128   57386   64197    67032    69325
 1024     256   64670   79092     4022     3969
 1024     512   80214   93225     3998     3991
 1024    1024   91134  102513   104098     4080
10240       4    5990    5976     5904     5941
10240       8   10448   10531    10408    10529
10240      16   14929   15087    17863    17846
10240      32   26269   26505    30825    31056
10240      64   40208   41481    45328    45401
10240     128   60009   61285    67542    66948
10240     256   77151   78960     5026     5235
10240     512   90214   91380    17142    20148
10240    1024  100260  102248    13044     9001
10240    2048  105222  108549     9187    14234

Full test results are in the attachment.

Best regards,
Andrew Savchenko

Attachment: orangefs-benchmarks2.tar.xz
Description: Binary data

Attachment: pgpgzL4gSQ8V1.pgp
Description: PGP signature

_______________________________________________
Pvfs2-users mailing list
[email protected]
http://www.beowulf-underground.org/mailman/listinfo/pvfs2-users

Reply via email to