General discussion

Locked

Disaster Preparedness

By Jakomeit ·
I am looking at options to minimize our downtime. Right now, our file and print server is about 16 months old. It has 7 hard drives running a raid 5 array. We had a hardware failure this summer and the part was sent overnight to us, but we were down for about 20 hours by the time I replaced the part. I want to minimize our downtime to less than an hour if something like this were to happen again so I basically came up with 2 options, buy another server with the exact config so I can swap the hard drives (which I can?t find the same server anywhere), or implement server clustering. Can you think of any other options I might have? Server clustering involves getting w2k advanced server on top of hardware and that will about triple my costs. Any other suggestions you have would be appreciated!

This conversation is currently closed to new comments.

9 total posts (Page 1 of 1)  
| Thread display: Collapse - | Expand +

All Comments

Collapse -

Disaster Preparedness

by timwalsh In reply to Disaster Preparedness

20 hours sounds a bit extremem for a RAID 5 array. The idea behind Raid 5 is that (aaumming you aren't maxed out on space) if a drive in the array fails, array can regenerate itself and continue to function after you pull a failed drive. When youhave the replacement in hand you can install it and regenerate the array again. The method to do this will be dependent on the Raid adapter used. Dig out the documentation for your adapter and check out the exact process you need to use.

Collapse -

Disaster Preparedness

by Jakomeit In reply to Disaster Preparedness

thanks for your response. It was not a hard drive failure, it was the voltage regulator module so the server was not able to power up at all without the replacement.

Collapse -

Disaster Preparedness

by timwalsh In reply to Disaster Preparedness

As you are discovering, true fault-tolerance costs money. You have essentially answered your own question as far as options.

For option 1 (second server as hot backup), if you can't find the exact same server, build one with as close to the same specs as you can get. The critical componenets will be the RAID controller (if you don't want to just move that with the drives) and CPU(s) (as this would involve the HAL). Even if you have to install a few different drivers after you move the array to the "backup" server, you can probably accomplish this within the hour time-frame you have set. In fact you can probably improve on that by doing a practice run on a weekend (when you don't affect any users). Once you do the swap once, the drivers will already be installed. The nxt time you HAVE to do a swap (because of equipment failure), Windows will recognize the differing hardware, and finding the appropriate drivers already installed, Windows will start that much faster (and you won't have to run around searching fro drivers).

Hope this gives you some ideas.

Collapse -

Disaster Preparedness

by Jakomeit In reply to Disaster Preparedness

Poster rated this answer

Collapse -

Disaster Preparedness

by MikeTalonNYC In reply to Disaster Preparedness

Hi:

First, check out the Disaster Recovery TechMails from TechRepublic, I address this type of issue all the time in my columns there =)

Specifically to answer your question, I would suggest using a replication tool (do a Yahoo search for "byte-level real-time replication") to keep all the data from one server on a second server. The benefit here is that the servers need not be identical. You simply configure the software on both machines, and the replication tool keeps all the data in sync - therefore in a failure situation you just failover to the second box.

Granted, this software does increase costs, but not nearly as much as cluster-class hardware and software would. Also, you can use non-cluster-aware applications with replication tools, something you can't do with a cluster.

As usual, I can't reccomend a specific tool becuase this is the area of expertise I write the columns about, but that Yahoo search will turn up quite a few different products for you to choose from.

Best of Luck
Mike Talon
miketalonnyc@yahoo.com

Collapse -

Disaster Preparedness

by Jakomeit In reply to Disaster Preparedness

Poster rated this answer

Collapse -

Disaster Preparedness

by Jose Mir In reply to Disaster Preparedness

Check Novell.
You are going to save big money in the long run, and you are going to add great stability, availability and scalability.
I know ? I know, many people think that because of previous decisions there is no way to roll back. Even if you use of MS-SQL and/or MS-Exchange you need to know about other possibilities. Oracle, DB2 and GroupWise are good examples.
Even if you want to keep mix environments you still may find benefits in using Netware and Win2K. Keep in mind that eDirectorywill make your life easier.
By the way, Netware 6 allows for 32 nodes clustering, at no extra cost. The same occur with multiprocessor support, you may have servers with up to 32 processors each.
And if your concerns are now about storage, Novell offers multi-platform NAS solutions.
Just take a look.

Regards,

Jose P Mir
jpm@jpmir.net

Collapse -

Disaster Preparedness

by Jakomeit In reply to Disaster Preparedness

Poster rated this answer

Collapse -

Disaster Preparedness

by Jakomeit In reply to Disaster Preparedness

This question was closed by the author

Back to Networks Forum
9 total posts (Page 1 of 1)  

Related Discussions

Related Forums