Sample Projects and Implementations

Objectives
The project was aimed at moving  a production database based on SQL Server 2000 to a cluster and configuring it in Active-Active mode. The whole operation including tests had to be carried out during a twelve- hour service window.
Implementation
After the computer to serve as a database server had been configured and started, Windows Server 2003 operating system was installed on it. Next, the clustering service and DTC service were installed and tested.
Two instances of SQL Server 2000 were installed on the computer thus prepared. After the latest Service Pack and Hot Fix add-ins had been installed, SQL Servers were allocated with relevant disk resources.
The functioning of the installed and configured SQL Servers was thoroughly tested, e.g. the system’s response to various failures was checked, including failures resulting with starting both SQL Server instances on the same cluster node.
Next, production databases were moved to SQL Servers, along with external resources necessary for the system’s work, including booting files. Having been tested, the system worked for 2 years on end, until, due to an air-conditioning failure, disk arrays suffered damage.
 Since the failure caused damage to over a dozen production databases, in the first place steps were taken  to gain access to the damaged databases. This was achieved within 2 hours, yet, unfortunately, the attempt at automatic recovery of the data stored turned out unsuccessful. In effect, following the customer’s decision, data were recovered from backup.
The array failure also resulted in damage to one of the SQL Server instances. In this case, the customer decided to have the server de-installed and then re-installed and configured. After this operation had been completed, the production databases on this server were recovered from latest backup.
 
 
Results

  1. Over a dozen databases of crucial importance to the customer were moved to a new cluster.
  2. The move was performed without impeding in any way users’ work.
  3. After a subsystem crucial for each cluster (the array) had been damaged, SQL Server’s work was back to normal within a few hours.
  4. The users and the administrators regained a limited access to the data lost due to the failure as soon as 2 hours after the failure had been reported.
  5. The problems with communicating with one of the servers using Linked Server mechanism, occurring in result of the damage caused to the operating system, were also diagnosed and troubleshot.