I want to thank everyone who has provided insight into my thread about clustering MySql. I kind of just sat back and watched it develop. I learned a lot from it all.
I have been reading all of the documentation on clustering provided by Centos/Red Hat, and find I travel in circles. I read one chapter and answer a self-imposed question but I end up asking myself another.
What I really want to do is have HA for any service I run (which is mostly HTTP, MySQL, FTP, and the common things like that). I want to run that to redundant storage somewhere that is real easy to expand by just adding more hardware (server or disk drive).
I started exploring this by using the Cluster Suite as a base and then looked into each aspect of the cluster and invariably got stuck on the storage side of this. I see how I can maybe set this up originally, but the expansion just doesn't seem to be there. I don't really want to go the route of Fibre channels and ISCSI, and would prefer to use common hardware (which sort of suggests GNBD).
If anyone cares to offer suggestions, with a pretty clear explanation trail (thanks Ken Price for your link to a step-by-step), I would really like to see it, as I'm not getting anywhere with the documentation. I hope to get some hardware to play with shortly, and maybe that'll make things clearer.
I'm sure it one of those deals where once I get it done, it'll be so obvious. I just need a little kickstart to help me get there.
Thanks,
Steve Campbell
On Fri, 14 Dec 2007 14:51:32 -0500 Steve Campbell campbell@cnpapers.com wrote:
I don't really want to go the route of Fibre channels and ISCSI, and would prefer to use common hardware (which sort of suggests GNBD).
Also take a look at ATA over Ethernet (aoe), either from coraid.com or roll-your-own with small aoe "server" called vblade.
Much simpler than iSCSI, but works on layer2, so no routing. Performance is about the same as you'd expect from old 1gbit fibrechannel.
Hi all,
I'm currently thinking about similar configurations, and (also for cost reasons :-) am also thinking about GNBD with two standard servers as a "poor man" redundant storage - but I'm wondering if that gives enough performance for running databases (in my case Oracle) on top of it. The configuration I'm thinking about would be two current Dell servers with hardware RAID 10 and connected by a dedicated 1 GBit crossover-cable, running both the cluster software and Oracle.
Does anyone use this in "real-World" scenarios and has some practical experience with it?
Best regards, __ /homas
-- Thomas Bleier, DI Information Management Austrian Research Centers GmbH - ARC HG Wien - FN 115980i - ATU14703506 2444 Seibersdorf, Austria
Mobile: +43 (664) 8251279 E-Mail: thomas.bleier@arcs.ac.at
-----Original Message----- From: centos-bounces@centos.org [mailto:centos-bounces@centos.org] On Behalf Of Steve Campbell Sent: Friday, December 14, 2007 8:52 PM To: centos@centos.org Subject: [CentOS] Expandable network storage
I want to thank everyone who has provided insight into my thread about clustering MySql. I kind of just sat back and watched it develop. I learned a lot from it all.
I have been reading all of the documentation on clustering provided by Centos/Red Hat, and find I travel in circles. I read one chapter and answer a self-imposed question but I end up asking myself another.
What I really want to do is have HA for any service I run (which is mostly HTTP, MySQL, FTP, and the common things like that). I want to run
that to redundant storage somewhere that is real easy to expand by just adding more hardware (server or disk drive).
I started exploring this by using the Cluster Suite as a base and then looked into each aspect of the cluster and invariably got stuck on the storage side of this. I see how I can maybe set this up originally, but the expansion just doesn't seem to be there. I don't really want to go the route of Fibre channels and ISCSI, and would prefer to use common hardware (which sort of suggests GNBD).
If anyone cares to offer suggestions, with a pretty clear explanation trail (thanks Ken Price for your link to a step-by-step), I would really
like to see it, as I'm not getting anywhere with the documentation. I hope to get some hardware to play with shortly, and maybe that'll make things clearer.
I'm sure it one of those deals where once I get it done, it'll be so obvious. I just need a little kickstart to help me get there.
Thanks,
Steve Campbell
_______________________________________________ CentOS mailing list CentOS@centos.org http://lists.centos.org/mailman/listinfo/centos
Hi Thomas,
On Dec 17, 2007 10:56 AM, Bleier Thomas Thomas.Bleier@arcs.ac.at wrote:
Hi all,
I'm currently thinking about similar configurations, and (also for cost reasons :-) am also thinking about GNBD with two standard servers as a "poor man" redundant storage - but I'm wondering if that gives enough performance for running databases (in my case Oracle) on top of it. The configuration I'm thinking about would be two current Dell servers with hardware RAID 10 and connected by a dedicated 1 GBit crossover-cable, running both the cluster software and Oracle.
It depends on the number of queries and how they use the indexes, this will generate the I/O that may or may not be too much for a poor man's solution.
Does anyone use this in "real-World" scenarios and has some practical experience with it?
I'm afraid I don't have a practical experience.
- Nicolas
-
Bleier Thomas -
Jure Pečar -
Nicolas Sahlqvist -
Steve Campbell