Exploiting SAN-based tape backup

Posted on November 01, 1999

RssImageAltText

Exploiting SAN-based tape backup

Enabling LAN-free, server-less, and zero-window backup, SANs are expected to pave the way for easier backup.

Steve Richardson

No other IT issue has generated as much interest recently as storage area networks (SANs). SANs are poised to dramatically change the way organizations manage and distribute data, impacting IT infrastructures with a force comparable to that of LANs more than a decade ago. In fact, the move to embrace SAN technology underscores a paradigm shift that moves storage resources and storage management into the forefront of corporate decision-making.

The emergence of SAN technology is part of a fundamental shift from the long-standing server-centric storage model (disk and tape resources are directly attached to a server) to a storage-centric model (storage resources are centralized and shared among servers).

Gateway to the future

A SAN is a separate computer network, typically a high-speed Fibre Channel connection that links storage devices to a heterogeneous set of servers on a non-dedicated basis. This means that any server can access any storage device on the SAN. A SAN can also enable direct storage-to-storage connectivity, allowing data movement across the SAN without impacting server performance.

In addition to performance and availability advantages, SANs streamline the administration of storage resources-perhaps the most critical issue facing IT managers. By liberating storage resources, SAN technology opens a new era of enhanced IT operations. Significant user benefits include:

•Availability. SAN architectures can enable high-availability computing environments, such as server clustering. The combination of externalizing storage pools for shared access and high-bandwidth make clustering much more feasible on SANs. These clusters are commonly configured with hot-server failover for the high-availability demands of 24x7 environments. The Storage Networking Industry Association (SNIA), for example, has proposed a SAN-based hot backup model that uses a failover clustered server approach to deliver guaranteed availability for backup operations.

•Performance. In addition to availability, server clustering can dramatically increase application performance. Again, the key is the ability to make storage assets available on a shared basis to multiple servers.

•Centralized management. Consolidating storage resources onto a SAN instead of attaching them to individual servers streamlines the storage management process dramatically. Separating storage from the server allows administrators to focus exclusively on optimized storage resource management without impacting the server.

•Scalability. Externalized storage enables administrators to add storage resources independent of server assets. Storage purchases can be shared among multiple servers, making it a more affordable option. Additionally, storage and server purchases can be optimized since they are upgraded separately.

•Dynamic resource allocation. For peak demand periods, SAN-based installations enable administrators to allocate additional resources to priority applications and servers. While server re-allocation is possible without a SAN, it is far less useful since storage resources cannot be shared. Dynamic server allocation with the ability to add or change storage resources without pre-determination is a powerful combination.

•Universal access. A SAN also enables universal access to multiple libraries, rather than one larger library. The advantage of multiple smaller libraries is redundancy (therefore uptime) and cost savings. Large silos are costly to buy and maintain and they take up valuable floor space. Also, if they fail, the backup process is interrupted.

Key applications

Most of the attention on SANs has centered on the performance benefits of a dedicated gigabit network that relieves conventional LANs of data movement loads. From a tape automation perspective, however, SANs have other significant advantages, such as improved storage manageability, more reliable and flexible backup operations, and shared storage resources among multiple servers. These technologies will enable SAN-based tape libraries to be accessed across multiple platforms, paving the way for easier backup, data migration, remote copy, and virtual tape applications.

By relocating storage resources to an independent network, the heavy overhead that conventional storage architectures place on LANs and network file servers is eliminated. Such CPU-intensive processes like backup will also be able to be performed without involving servers. First-generation SAN-based applications, however, will be built around a new generation of backup technologies such as LAN-free, server-less, and zero-window backup.

LAN-free backup. In SAN installations, enterprise storage resources reside on an independent gigabit-speed network. All data movement occurs over this high-speed dedicated network, not over a standard Ethernet LAN. The net effect of SAN-based, LAN-free backup is an immediate improvement in LAN performance. LAN-free backup is likely to be the first widely deployed storage management application on storage area networks.

As shown in the figure on p. x, LAN-free backup technology gives multiple servers access to a single tape library connected to the SAN. All backup operations are now routed through the gigabit-speed Fibre Channel SAN rather than the conventional Ethernet LAN.

A new generation of SAN-aware backup software supports this architecture. The software coordinates between servers to allocate tape library resources among tape libraries, eliminating data movement over the LAN.

Server-less backup. Server-less backup takes LAN-free backup one step further, enabling direct data movement between SAN devices. Data moves directly from RAID disk storage to the tape library, removing the server bottleneck.

Server-less backup leverages two key technical developments: the SCSI-3 block copy command, or "third-party copy," and Network Data Management Protocol (NDMP)-compliant software to manage communications between the server and the tape library.

The term "server-less" backup is actually somewhat of a misnomer for this application, however. The server still plays a role in the backup operation using NDMP to manage communications between SAN storage devices and to ensure backup is successfully completed. Nonetheless, server intervention is minimized since all data is sent directly over the SAN. The result is significant increases in performance and reliability.

Conventional LAN backup operations, in contrast, consume a vast amount of resources, including server CPU cycles, I/O busses, and LAN bandwidth as data moves from a server-attached RAID device to a local or network-attached tape library. All aspects of server and network operations are affected, imposing a significant performance hit.

Zero backup window. The combination of NDMP-compliant backup applications, SCSI copy, and Legato`s Celestra agent has the potential to enable another powerful SAN-based application: zero backup window. Celestra creates a "snapshot"--that is, a point-in-time virtual mirror requiring only a small fraction of the disk space needed to create an actual mirror of the data--of the backup data. This allows the applications to be returned to production status almost immediately instead of remaining unavailable for the duration of the backup operation.

The snapshot directs the backup software to the disk location of the original data for backup. If a write command is issued to update the backup data set, the Celestra agent intercepts the command and the update is written to a new section on the disk, maintaining the integrity of the original data. When the backup is completed, the snapshot is deleted.

The case for open standards

At this early stage of development, multiple standardization efforts are in effect. Efforts by SNIA, IEEE, and others have lagged technology advances, making vendor testing and certification the only guarantee of interoperability. Nonetheless, a core group of standards, such as "third-party copy" and NDMP, provide the basis for scalable, open SAN installations. Though neither ISO or IEEE has officially sanctioned the standards, they have gained de-facto status and enable first-generation SAN applications.

The challenge now is to deliver open interoperable SANs to meet the huge demand for this innovative architecture. The combination of vendor certification and an adoption of de-facto market standards will speed the development of SANs.

Click here to enlarge image

LAN-free backup (left) on a SAN eliminates data traffice from a conventional Ethernet LAN by moving backup data across a dedicated high-speed Fibre Channel connection.

Click here to enlarge image

Server-less backup (right) liberates network servers from backup operations that affect CPU performance, I/O processes, and LAN bandwidth by enabling direct data movement between storage devices on the SAN.

Steve Richardson, vice president of marketing at Overland Data (overlanddata.com), in San Diego, CA.


Comment and Contribute
(Maximum characters: 1200). You have
characters left.

InfoStor Article Categories:

SAN - Storage Area Network   Disk Arrays
NAS - Network Attached Storage   Storage Blogs
Storage Management   Archived Issues
Backup and Recovery   Data Storage Archives