[Top][All Lists]
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Fwd: Re: [Gluster-devel] features/mirror resp cluster/automatic-file-rep
From: |
Krishna Srinivas |
Subject: |
Fwd: Re: [Gluster-devel] features/mirror resp cluster/automatic-file-replication |
Date: |
Thu, 1 Feb 2007 20:16:19 +0530 |
---------- Forwarded message ----------
From: address@hidden <address@hidden>
Date: Feb 1, 2007 8:09 PM
Subject: Re: Re: [Gluster-devel] features/mirror resp
cluster/automatic-file-replication
To: Krishna Srinivas <address@hidden>
Dear Mr. Srinivas,
we are running a Linux Virtual Server with 9 servers and 2 load
balancers based on keepalived.
Underlying file system is OpenAFS with a replication/failover-handling
based on drbd-heartbeat.
Applications: Tomcat, Twiki, Catalyst/HTML-Mason, MTA (courier), Web
Services (SOAP::Lite), database (MySQL), file serving (CIFS/samba)
We are validating different cluster filesystems to replace OpenAFS due
to its bad write performance.
Backup software should be dirvish (hard links necessary)
gfarm: no automatic replication during write, metadata can become
inconsistent, no hardlinks, good performance
lustre: hard links ok., failover only via heartbeat and iSCSI, bad
performance for grep operations (TWiki), many kernel patches!!!
glusterfs: looks promising, hardlinks, good performancs, easy
administration, good to medium performance for grep (Twiki), overall
good performance, roadmap promising
A failover handling is necessary, because we need a 6x24 availibilty
for our CNC-machines and worldwide locations.
Best Regards / Mit freundlichen Grüßen
R. Rohmfeld
____________________________________________
EMUGE - Werk Richard Glimpel GmbH & Co. KG
Fabrik für Präzisionswerkzeuge
D-91207 Lauf, Germany
Dr. Rainer Rohmfeld
e-mail: address@hidden
http://www.emuge.de