November 5, 2001 at 7:13 pm
Hi all
I currently have a 2-way server with 512Mb RAM as my prod database server connected
to a disk farm via a 2m RAID controller. The idea was that we can cluster up as
need be in the near future as high availability becomes a concern for us with another
similar config server.
The current server is performing very well with the work load at present.
We also need a reporting DB, that will run some OLAP and batch reporting. I am not keen that it runs on the existing server though.
What I am thinking of is this:
(web-server)
|
|
Current Reporting
Server <--------> DB Server
(active/passive cluster)
|
|
#disk farm#
We will buy another server, and run in Active-Passive mode from a clustering
perspective. This will cater for the HA if and when we need it. In the mean
time the other server will have another instance on SQLServer for its reporting
DB functions to keep it busy.
Ideas/Thoughts??
As I am new to clustering, if we went to an active-active cluster, can I skip
the load balancing functions in our front-end code to force some jobs to
run over the reporting DB? the reason I asked is that
the reporting DB server will have a batch scheduler program running
that is not clusterable and we will need to submit jobs to this
specific server via some front-end DLL's on the webserver.
Another option I was considering (due to the cost of the cluster) was this:
(web-server)
|
|
Current Reporting
Server -----replication-------> DB Server (local db)
|
|
#disk farm#
Cheers
Chris
Chris Kempster
www.chriskempster.com
Author of "SQL Server Backup, Recovery & Troubleshooting"
Author of "SQL Server 2k for the Oracle DBA"
November 7, 2001 at 8:43 am
You can't remove the load balancing if you want your app to pull from ether server. In an active/active mode you basicly have two sql servers with two diffrent names they don't pass back and forth from each other.
November 7, 2001 at 11:26 am
It seems like you are trying to do two things. If you want to load balance, then you would do that by directing some queries to one server instead of another.
If I understand correctly, you could run active-active and setup your database on shared disk. Then create a 2nd database on the servers that receives replication. You can still direct queries to SQL1 or SQL2. In a failure, the remaining node woudl take over both servers, though it would be loaded.
Not a bad idea, though I'd spend some $$ to get the hardware beefed up.
Steve Jones
November 7, 2001 at 11:28 am
I was also thinking you could do a horizontal view across both servers.
November 7, 2001 at 12:41 pm
What would that do in a clustered scneario? Probably still work.
In that scenario, if you have SQL 2000, you might even want to investigate a 3 or 4 server cluster. that way the load would be shared across even more than one server if one failed.
Steve Jones
Viewing 5 posts - 1 through 4 (of 4 total)
You must be logged in to reply to this topic. Login to reply