Evolution to a SAN Environment

Incremental Steps to a Second-Generation SAN

The creation of a SAN envisioned in the recommendation can be approached on an incremental basis:

1. Select backup/remote disk mirroring software . The current backup software should be reviewed to ensure it addresses the near real-time disk performance required for the remote disk mirroring. Since software tends to evolve version to version, the evaluation point is likely to be whether the software has the feature when you need it.

2. Upgrade the distribution switch to support SAN throughput speeds . Minimum speed will be Gigabit Ethernet. Some vendors will enter at the 10 Gigabit Ethernet speed. This throughput requirement may force the replacement of the existing distribution switch within your infrastructure.

3. Install a standalone iSCSI disk system and HBA . Second-generation iSCSI disk systems and HBAs were introduced in 2002. This step allows the iSCSI interface to be proven without any other components in the mix. Implementation on a nonmission-critical server should provide the experience needed for further evolution.

4. Convert standalone iSCSI disk system/HBA to the distribution switch . This step allows the iSCSI interface to prove itself through the distribution switch. Use of the nonmission-critical server from Step #3 should provide the experience needed to assure confidence.

5. Implement a second host on the distribution switch to create a SAN . This step creates a SAN in the central office. Again, adding a second nonmission-critical server is recommended. Once proven, the inclusion of mission-critical servers can be considered .

6. Implement remote storage with communications . Additional outlays will have to occur to implement the remote disk space with communications links. The most likely link will be Gigabit Ethernet.

7. Implement remote Disk Mirroring over the communications link . This step may force you to revisit the software question considered in Step #1.

These steps would provide easily managed steps with little or no risk of failure on a large expenditure. With the potential exception of Step #6, the other steps would fit into the current replacement process typically employed within Information Technology.