Cisco Systems OL-12518-01 manual Conclusion, Remote Tape-Backup Applications

Chapter 4 FCIP over IP/MPLS Core

Conclusion

Table 4-2	Application Requirements
Disk mirroring		Varies depending on	< 1–5 ms for	Synchronous or	Synchronous
		storage array.	synchronous.	asynchronous.	applications are very
		Typically maximum 50	Asynchronous		sensitive to delay.
		MB per storage array.	replication tolerates		Asynchronous are
					less sensitive to
			higher latency (100
					delay.
			ms).
File access		OS dependent.			Depends on the OS
					and application
					above it.

In addition to those listed in Table 4-2, other requirements include provisioning, error monitoring, and end-to-end management.

Remote Tape-Backup Applications

In an enterprise network, certain sites (remote branches or small offices) can have a small SAN that connects a few servers to the storage arrays. Backing up and restoring these servers over the WAN is a fundamental component of disaster recovery operations. Extending tape backup over a wide area imposes stringent requirements for efficient tape backup and recovery operations. These requirements include no data loss, low latency and jitter, monitoring of the link, and high security.

Slow wide area links can increase backup time and can make it impossible to complete backup within the allocated time period (or “window”). Distance is not a major limitation for backup to tape applications as long as it is possible to predict delay requirements. For backup to tape to be as efficient as possible, Cisco recommends sustaining a certain speed so that a continuous stream of data is sent to tape. Backup performance has been found to be best when the tape can accept a continuous stream. Backup to tape transfer over the WAN is asymmetrical in nature. The asymmetrical nature of tape-backup data transfer creates unique challenges when designing SAN extension networks.

Tape “pipelining” technology helps to extend tape drives thousands of kilometers, thus making remote tape backup an essential component of business continuance and disaster recovery applications. The efficiency is achieved by implementing buffering and error-recovery mechanisms. The concept is similar to spoofing; even though the server and tape controller are separated by a large distance, they behave as if they are co-located. The tape pipelining technique relaxes the design constraints of SAN extension technologies.

A typical solution includes transport over MPLS, which provides all the necessary QoS requirements required by tape backup applications. The Cisco solution provides necessary provisioning, management, bandwidth optimization, and performance parameters that are critical to implement tape backup applications. The Cisco solution can scale as bandwidth requirements increase and still maintain the QoS requirements required to support this application.

Conclusion

IP is becoming a protocol of choice to transport storage traffic across WANs and MANs. The IP/MPLS networks of SPs can be used to transport FCIP and iSCSI efficiently for disaster recovery and business continuance solutions. SPs can leverage their current infrastructure with out much modification to the network elements to transport FCIP and iSCSI. By providing storage services, SPs can increase the utilization of their network while providing value-added services.

	Data Center High Availability Clusters Design Guide
4-30	OL-12518-01