Please start any new threads on our new
site at https://forums.sqlteam.com. We've got lots of great SQL Server
experts to answer whatever question you can come up with.
| Author |
Topic |
|
mrkryz03
Starting Member
1 Post |
 Posted - 2004-10-11 : 10:45:27
|
| Hi everyone, I'm new to this forum and was hoping to get some advice on what the best method of replication would be to provide redundancy and fault tolerance.I just migrated our existing sql server (old hardware) to two new machines. Each new server runs windows 2003 standard + sql server 2000 enterprise w/sp3 and is configured with 2 Xeon - 2.8G cpu's, 2 gig of ram, harware raid 10 across 4 large drives, one 100Megabit internet facing nic, and one 1Gigabit Lan facing nic.The old platform housed 7 databases. 2 databases were approx. 1 Gig each the rest were <500M. Now that I have the two machines I split the load equally and reconfigured the clients to access data at the appropriate target server.My question is given the above information would anyone have suggestions on how to BEST ensure fault tolerence/redundancy/fail-over. Our budget is pretty stretched already after the hardware and licensing was purchased. The goal is to have the two machines run independently and in the event of machine or sql failure to autonomously fail over to the other machine. I have copied all the databases over so each is a mirror of the other however the databases that are meant to be failover targets are currently offline/detached. Server A Server Bdatabase1 - live database1 - offlinedatabase2 - live database2 - offlinedatabase3 - live database3 - offlinedatabase4 - offline database4 - livedatabase5 - offline database5 - livedatabase6 - offline database6 - livedatabase7 - offline database7 - liveI could really use some help in findign the best way to accomplish a truely autonomous failover architecture in the event server1 dies that server 2 goes online for server1's databases and vice-versa. I have some background (minimal at best) with replication and to me it seems that transactional replication is needed as these databases change significantly in a matter of a minute or two. How to acheive this is a bit confuxing lol . Ive read that transaction log exporting may be the best way, other sources site active/active replication, yet others reccomend 3rd party software etc... anyone have any similar experiences or ideas? any help would be very appreciated by this noobie.Thanks,Kryz |
|
|
nathan.russell
Starting Member
13 Posts |
Posted - 2004-10-13 : 05:28:46
|
I have a very similar requirement (and am in a similar 'newbie' type position !) - did you get anywhere with this ?I also get the feeling that the SQL gurus dont really like or recommend replication ? but I can't find out why ? The only thing I've heard is things like it is a pain to setup / maintain / monitor etc (which from my limited trials I have found to be true  )I'd be interested in anything you may have learnt on this subject.CheersNathan |
 |
|
|
jen
Master Smack Fu Yak Hacker
4110 Posts |
Posted - 2004-10-13 : 05:51:03
|
| what do you mean active/active replication?i think you meant cluster not replication. we implement active/active but make sure that if the failover happens, the other node can handle the load, otherwise you'll end up with 2 nodes not working. unless your license can accomodate 4.i am not sure with logshipping though. i believe you need the other server to be on standby and read-only and this is only possible for 1 database at a time per server. now if you have 7 db's then this is not your option.another option is to create your own logshipping, perform network backups of the: one full backup at the beginning of the day, then tlog backup every 15 minutes. you restore each file being backed up the moment it is received in your "failover" server which can be operational at the same time.just some ideas...--------------------keeping it simple... |
|
|
|
|
|
|
|
|
Replication/Availability Advice