I would consider doing either a 2 node or 4 node cluster with an external
RAID array. Clustering requires Windows 2000 Advanced Server and SQL Server
2000 Enterprise Edition; there is third party software for implementing
clustering but, I have not used them. There are two types of cluster
active/active and active/passiver; each has pros and cons. Be aware,
clustering provides failover support not load balancing. In the event
SERVERA should fail SERVERB will take over, end users should not notice
anything happened.
Another, less costly solution is log shipping. It is a process in which a
database's transaction log is automatically copied and restored to another
server called a "standby". There is a potential for data loss here, in the
event SERVERA should fail before the standby receives a transaction log.
There is also more human intervention required to make the standby server
the primary server. I have not used log shipping.
Personally, I like to configure my databases with the data files (.mdf) on
RAID 5 and transaction logs on RAID 1. Given budget constraints, I have
found it to give me the best cost/performance ratio.
After the hardware is in place, the next phase of the contigency plan is
creating a backup schedule. I usually do weekly full backups, nightly
differentials and periodic transaction logs during the day. In the event
there is a catastrophic failure -- meaning all servers in the cluster have
failed and/or the external RAID is blown -- if possible, attempt to get a
transaction log backup then proceed to restore everything from the last full
backup, most recent differential and subsequent transaction logs to another
server. (This last paragraph could possibly be more information than you
wanted).
A good resource on SQL Server availability is the December 2001 issue of SQL
Server Magazine. Its topic is SQL Server high availability. Here is a link
http://www.sqlmag.com/Articles/Index.cfm?IssueID=513
Regards,
Bianca
-----Original Message-----
From: Thirumudi, Dhodduraj [mailto:dhodduraj.[Email Address Removed] Monday, August 19, 2002 9:19 PM
To: Bianca Blount; [Email Address Removed] RE: Disaster Recovery Planning
Hi Bianca Blount,
The budget is around 1000K. Delivery time frame will be 4 to 6 weeks. The
project needs to be
completed within next 3 months. We have enough resources. Now what we need
is a solution to implement.
Our goal is to bring the uptime to 99% and in a disastrous situation like
Hardware failure we should
be in a position to switch to other server and the downtime should be less
as much possible 10 to 15 minutes.
Rgds,
Thiru.
-----Original Message-----
From: Bianca Blount [mailto:[Email Address Removed] Monday, August 19, 2002 9:22 PM
To: [Email Address Removed] RE: Disaster Recovery Planning
Hi,
The hardware/software configuration will depend on the budget, the amount of
availability needed, staff available and delivery timeframe. Answering
these questions would facilitate an answer to your question.
Regards
-----Original Message-----
From: Thirumudi, Dhodduraj [mailto:dhodduraj.[Email Address Removed] Monday, August 19, 2002 6:19 AM
To: SQL Server DBA
Subject: Disaster Recovery Planning
Hi DBA's,
Our company is currently reviewing the Disaster Recovery Planning procedures
and its requirements.
I would like to know what would be the best Hardware/Software configurations
required to minimse the downtime and
how soon this can be implemented.
Your input on this will be highly appreciated.
With thanks & Rgds,
Thiru.
---------------------------------------------------------------------
To unsubscribe, e-mail: mssqldba-[Email Address Removed] additional commands, e-mail: mssqldba-[Email Address Removed] unsubscribe, e-mail: mssqldba-[Email Address Removed] additional commands, e-mail: mssqldba-[Email Address Removed]MS Sql Server LazyDBA home page