SAN Management: The Basics
Storage area networks require storage management and network management disciplines.
By Scott McIntyre
Storage area networks (SANs) represent a fundamental shift in the way distributed computing environments relate to storage resources--a shift that has important ramifications for the way storage is managed. This article examines how data protection can be managed in a SAN today. Next month, we`ll examine the potential of improving the management and protection of storage with SANs.
Long an afterthought in IT buying decisions, storage is now a primary purchasing consideration, providing strategic competitive advantages. The emergence of Fibre Channel--a high-speed, low-latency serial interconnect that can be used to attach storage to servers--and the application of networking concepts to server-storage connectivity have made it possible to build new storage infrastructures.
It is not surprising, then, that SANs--high-performance networks (usually based on Fibre Channel) dedicated to storage--have attracted so much interest among IT managers. SANs provide any-to-any connectivity, which means any server can potentially talk to any storage device. This communication allows users to share storage resources. SANs also ease the task of centralized management, improve performance, and allow for longer connectivity distances via Fibre Channel. SAN technology is driving the adoption of remote management and data protection strategies, storage consolidation, and cross-platform data sharing.
Just as SANs marry storage and networking technologies, managing a SAN combines two disciplines: storage and network management. As in any network, the physical network resources must be managed, including such tasks as configuration and topology management, resource monitoring, failing path isolation, port control, and performance management. And as in any storage configuration, data needs to be managed, including such tasks as logical volume management, storage subsystem management, performance and environmental monitoring, backup, hierarchical storage management, and media management.
For many users, one of the immediate reasons for implementing a SAN is data protection. This article addresses such data protection issues.
Requirements for SAN Management
One of the first questions asked by potential SAN implementers is, Will I be able to back up and protect data in my SAN? To effectively protect your data, you`ll want to:
- Minimize network load--Early adopters of SANs often point to the ability to move backup off production networks and onto storage networks as a reason for implementing a storage network. Indeed, for users who are currently backing up large databases or application servers over production networks to backup servers, the ability to back up data directly through the storage network is significant.
- Accomplishing this requires a storage management architecture that enables large servers to back up data directly to tape while maintaining centralized management of the process. The backup load should be spread across the network and should allow large production servers to be backed up directly to secondary storage. In a storage network, this capability enables backup to be moved from the LAN to the SAN.
- Maximize application availability--Taking backup traffic off production networks keeps networks available for intended business processing. Equally important is "hot," or on-line, backup, providing uninterrupted access to business-critical applications and databases. This is particularly important for applications such as SAP R/3, Microsoft Exchange, and Lotus Notes.
- Support heterogeneous platforms. Because the SAN concept is relatively new, initial implementations will probably be homogeneous. However, networking technology is about connectivity, and that implies heterogeneity. Since SAN environments will likely become more homogeneous as they mature, effective SAN management software will need to support any server and any storage product, hosting any database, application, or file system, backing up to any tape drive or library through any switch, hub, router, or bridge. Applications that attempt to lock users into a particular hardware or software product set will limit the management scalability of the environment.
- Centralize management. The connectivity distances of Fibre Channel make it easier to deploy remote sites for business continuance and disaster recovery purposes. This will lead to increased use of remote backup, vaulting, and mirroring techniques.
- Share removable media libraries. One of the benefits of SAN connectivity is the ability to share resources (e.g., a large tape library) among multiple backup servers. This enables administrators to consolidate backups (often done from many different servers to locally attached tape drives) into one tape library. For many users, the ability to connect a library to multiple backup servers via a SAN will justify the expense of automation. However, simple connectivity to the tape library does not necessarily mean effective management. Effective management means managing access to the media, managing the allocation of drives among servers, and managing the library for any application, not just backup, requiring access to tape or optical storage.
- High-performance backup and recovery. One of the key early benefits of a SAN is the performance boost provided by Fibre Channel. Though you might think Fibre Channel`s increased bandwidth would automatically increase backup and restore performance, the key to high-performance backup is flowing enough data to the tape devices to keep them streaming. Failure to do so results in "backhitching" of the drive mechanism, which quickly degrades backup performance. Therefore, backup software must be able to drive data out to the tape drives with maximum efficiency.
In most cases, PC/workstation clients and small departmental servers won`t be directly attached to the storage network because they don`t have high data bandwidth requirements necessitating a direct Fibre Channel connection and because they depend on various file and application servers for their data needs. These clients will continue to be backed up over the messaging network, without putting undue strain on network bandwidth.
On the other hand, servers that own large amounts of data (such as large database servers, business-critical-application servers running SAP R/3, Microsoft Exchange, or very large file servers,) should be backed up directly over the SAN to a central tape library. In fact, one of the first benefits of SAN backup is taking the backup of these large servers off the production network and handling it within the storage network.
Some servers will be connected to the SAN for fast access to data, but will actually have relatively small amounts of data to back up, especially when incremental backups are considered. For these servers, backup over the messaging network to a central backup server may still be appropriate, though backing up directly to the storage network is possible. For these servers, a cost/benefit analysis should determine the benefit of direct backup versus the cost of the necessary hardware and software.
Given these considerations, it is likely that multiple servers (e.g., multiple large production servers and a client backup server) will be backing up data in the storage network. Therefore, the architecture of the data protection product will need to centrally manage all servers simultaneously, and the media management product will need to provide simultaneous access to these servers to shared tape library resources.
In summary, managing SAN backup today is not that much different than managing backup in a traditional storage environment if your storage management software has a robust management architecture. The benefits are immediate: improved performance through off-loading of backup traffic from the messaging network and optimal use of Fibre Channel bandwidth and shared storage resources. Next month, we`ll examine future developments in SAN management.
In this SAN configuration example, backup is split between the production network and the storage networks.
Scott McIntyre is business line manager for storage networking at Legato Systems, in Palo Alto, CA.