In addition to the regularly scheduled recovery drills, the lead DBA should randomly choose at least one time every month for a live disaster drill. This disaster drill is addressed and treated just like a real emergency. This is a date that should be selected by the lead DBA and approved by management. No one else on the DBA staff should know when it is going to occur. Since this is treated just like a live disaster, this means nothing interferes with this drill. Support gets put on hold or handled exactly the way it would in a normal emergency. Any meetings a person might be involved in are rescheduled or ended early if appropriate in order for them to participate. This last part of the disaster recovery program obviously needs to be approved by and supported by all levels of management. The entire purpose of a drill of this kind is for practice in a live situation.

This process gives you a highly trained DBA team that can quickly and accurately diagnose and recover from any problem in your environment. With the practice also comes familiarity with recovery procedures and the specific processes for your environment. This allows you to set and be able to expect specific levels of service and outage time frames.

All databases should be dumped daily to both disk and to tape. DLT drives are highly recommended due to capacity and speed. These are native SQL Server dumps. This gives you media protection in case a single media type fails. Further, your Enterprise backup software should then come through and backup the disk dump to tape to give an extra measure of redundancy.

The tapes should then be rotated offsite as per the tape rotation policy.

Server Inventory

The server inventory is obviously specific to your environment. This will contain a complete listing of all servers and their configurations. I generally lay this out in an outline format, but you can utilize any format you choose as long as all pertinent data is captured. This is your paper track to reconfigure a server exactly the way it was in the event of a complete server failure. An example is included below:

ServerA

Model: Compaq 7000 rack mount

Processor: 4 – 450MHz Xeon processors with 1MB cache

Memory: 8 – 256MB DIMMs (2GB)

RAID controller: Dual ES3100 controllers

NIC: Dual Netelligent 3122

Power Supply: 3 - hot swap

Fans: 3 - hot swap

Internal Storage: 2 – 4.3GB 10,000RPM UltraWide SCSI,

10 – 18.2GB 10,000RPM UltraWide SCSI

External Storage: 24 – 18.2GB 10,000 RPM UnltraWide SCSI in 3 external cabinets

RAID Configuration: 2 - 4.3GB mirrored. 10 –18.2GB in a RAID 5 (1 configured as a spare), 24 – 18.2GB in a RAID5 (second controller with 1 configured as a spare)

Partitions: C – 4.3GB (RAID1), D – 378.4GB (RAID5), E – 137.6GB (RAID 10)

Software: NT Server 4.0 Enterprise Edition, SQL Server 7.0 Standard Edition, WinZip 7.0, Insight Agent, SNMP services, SSD

Backup Overview 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32