Cluster Service improvements for Storage Area Networks (SANs) (301647)

MORE INFORMATION

Earlier versions of Windows server clusters presumed that all communications to the shared disk should be treated as a SCSI bus. This behavior may be somewhat disruptive, and it does not take advantage of the more advanced features of Fibre Channel to both improve arbitration performance and reduce disruption.

The key enhancement is that the Cluster service issues a command to break a RESERVATION, and the port driver can do a targeted or device reset for disks that are on a Fibre Channel topology. In Microsoft Windows 2000 server clusters, an entire bus-wide SCSI RESET is issued. This causes all devices on the bus to be disconnected. When a SCSI RESET is issued, a lot of time is spent resetting devices that may not need to be reset, such as disks that the CHALLENGER node may already own.

Resets occur in the following order:

1. Targeted LUN
2. Targeted SCSI ID
3. Entire bus wide SCSI RESET

NOTE: Targeted resets require functionality in the host bus adapter (HBA) drivers. Contact the manufacturer of the HBA to determine if it supports target resets.

If the targeted resets fail, the Cluster service resorts to a Windows 2000-style bus-wide SCSI RESET. For more information about how server clusters communicate with the shared disk, click the following article number to view the article in the Microsoft Knowledge Base: For more information about SANs, visit the following Microsoft Web site: