Hardware beats out software for top honors.
SANs are well entrenched in larger corporations, but were suffering from the SAN island syndrome. Consolidation projects are big right now, and:
* Connect remote SANs using fast data transport and central management software
* Physically combine smaller SANs into larger central data centers, and/or
* Migrate data from several storage targets to high-capacity arrays and tape libraries.
NAS and SAN Convergence
NAS and SAN moved closer together as complementary technologies. There is no one-size-fits-all hybrid yet, but NAS gateways and SAN file systems allow firms to use storage networks for both block and file-based traffic.
Replication, mirroring and snapshots have always copied disk to disk, but companies are using disk for regular backups as well. IT likes the speed, performance and availability of disk.
Disk and tape hybrids: Tape libraries are still very much in the picture, and are responding to disk-to-disk's speed by presenting tape library back-ends with fast disk cache front-ends.
Intelligent switches: Switches are getting a lot smarter, and so-called backbone switches might take over a majority of SAN processing. Intelligent switches are important in both Fibre Channel and iSCSI storage networking environments.
iSCSI storage: IP-based storage networks are getting more popular, serving smaller companies and niches than Fibre Channel but showing great promise.
SMI-S: The standard-known-as-Bluefin bloomed under SNIA's care. Part of the larger CIM/WBEM initiative, vendors hope it will help to integrate multiple-vendor storage networks.
Consolidation is busting out all over. Scott Drummond, program director of storage networking at IBM said, "We're seeing a tremendous amount of storage consolidation happening. And the area we're attacking most aggressively, because we've already got consolidation in the UNIX space, is Windows and NT servers." Consolidation can be physical and logical--physical consolidation into large data centers and arrays, or logical connections between remote SANs using centralized management software.
IT departments consolidate SANs to decrease management overhead and to eliminate numerous smaller servers with long backup windows. Newer storage devices also have high availability standard features such as multiple global hot spare drives and robust power redundancy features, and present great value for the money in centralized systems.
NAS and SAN Convergence
NAS and SAN are quite different technologies, and there is no approach that offers file and block-based data sharing across multi-vendor storage networks. However, vendors have still managed to come close by putting NAS gateways in front of the SAN and using SAN filers.
NAS heads and gateways are popular integration technologies. They allow NAS filers to sit at the edge of the SAN and share storage space. This leverages NAS value by expanding its storage options, and leverages SAN value by giving SAN users a file-based storage option. SAN filers offer similar advantages by offering file serving within the high-speed storage network. For example, EMC's Highroad runs on Celerra, intercepting file requests and passing them on to the Symmetrix. No one has a universal file system yet--SAN file system technology tends to be limited to proprietary devices, since it is a huge challenge to share block level and file level data. But global SAN file system technology is promising, since it would enable SANs to serve single pools of block and file data, using both NIFS and CFS across multiple OS's, in heterogeneous systems.
NAS is also getting better at hosting block-based data, meaning IT can safely use qualifying NAS as database storage. IT has been using NAS for data-base storage for years--it can certainly be done--the question is how reliably? The problem has been NAS' inability to directly access block-level data on SCSI-attached devices or Fibre Channel SANs. (NAS can still serve block data, but must route the connection through a file system.) Many database applications and data types don't expect or appreciate the routing and may have an unacceptable error rate. This is now changing with new NAS technologies that can bypass its own file system.
Tape is great for long-term, off-site archival but online backup prefers disk to disk. The challenge for backup product makers is to provide disk-based long-term storage that supports multiple backup applications, multiple versions of backups, uses disk's random access capability for reliable restores, and doesn't cost an arm and a leg. Disk-to-disk speed allows a fast backup and restore, which makes it an excellent candidate for online backups, replication, mirroring and snapshots.
However, disk-to-disk is not a no-brainer. Keeping long-term backup copies in hard-disk storage is cost-prohibitive in terms of disk capacity, data center space demands, and power consumption. (Constantly spinning disk takes up a lot more power than tape cartridges residing in vaults.) Many consultants--and certainly the tape vendors--suggest that tape should still serve as the long-term archive repository, perhaps as part of a tiered storage system (primary storage, near-line storage, off-line storage).
Disk and Tape Hybrids
The major tape library vendors are responding to disk-to-disk backup by making disk backup themselves (StorageTek) or by adding disk cache front-ends to their tape libraries (StorageTek, ADIC, Quantum ATL). Keeping fresh backed-up data in cache allows a fast backup and restore, which is useful for barely aging data. The library's management functions eventually migrate the data from the cache to the back-end tape cartridges, where it can remain online for some time. Tape capacity may be an issue with single drives and autoloaders, but enterprise tape libraries can host many drives and contain from hundreds to thousands of 100GB+ (native) cartridges.
Switch vendors are actively adding more intelligence at the fabric level. Developers are centering mostly on copy and move functions like replication and remote mirroring; virtualization, provisioning and storage management; and internetworking and utility-based models. Mark Stratton, McData's director of Solutions and Alliances programs, describes the switch vendor's viewpoint. "We're looking at managing more of what's in the fabric, also providing more information about relationships of what we're managing. We'll continue to be as aggressive as we can in this space." Copy and move functions include replication, remote mirroring, and online backup. Intelligent switches can also control tiered storage, which migrates data to different storage targets depending on characteristics. Virtualization is a big development center for switch makers, since virtualization significantly improves volume management and provisioning between different storage arrays.
Both established companies and start-ups are developing so-called backbone switches and investing them with heavy-duty storage management capabilities (not everyone uses the backbone terminology.) These backbones are meant to support consolidated enterprise SANs with huge port densities and large-scale scalability needs. Some of the new switches will work only on their own fabrics while others will work with existing switches from other vendors.
SCSI is the premiere data networking I/O bus. It's the backbone technology for data networks, but Fibre Channel supplanted it for high-speed storage networking. This is due to SCSI's inherent disadvantages, which include performance issues, single-server configuration, short distance limitations, availability risk, bus limitations and difficulties sharing storage among servers. In contrast, Fibre Channel's high performance, connectivity and block-level availability created a switched infrastructure for stored data.
Putting the two together resulted in the iSCSI protocol, which combines the SCSI command set and IP/Ethernet protocols with Fibre Channel storage networking features. This means that IP-based host systems can pass commands to multiple storage devices over an IP network infrastructure, not over direct SCSI cables or a Fibre Channel storage network. iSCSI has been around for several years, but Cisco's entry into the iSCSI arena raised the bar this year--the move is hugely significant given Cisco's long reach, networking dominance and rich partnerships.
SNIA took on a proposed storage specification named Bluefin and renamed it SMI-S (Storage Management Initiative Specification). Most storage vendors are looking to SMI-S as a standard to replace API sharing and improve management simplicity and functionality. SMI-S, which is the storage piece of the broader CIM/WBEM standard, provides a highly functional and interoperable interface for managing multi-vendor storage devices.
SNIA reasons that SMI-S will reduce the cost of storage management by providing a single management console for heterogeneous storage networks, eliminating the need to manage each device with a separate management application, and enabling any SMI-S compliant management application to discover and control conforming devices in the storage network. If it works as planned--and it looks like it will--then storage hardware vendors will be able to shift resources from API development and sharing to value-added functionality, will be able to present richer interfaces to users, and will greatly improve multi-vendor storage networks.
What's Ahead? Here's My Crystal Ball
2004 will see further gains in SAN consolidation, both physical--giant data centers, huge arrays and tape libraries--and remote, as SAN management software moves to next-generation capabilities and interoperability features. NAS and SAN consolidation will continue to center around gateways and SAN filers, while developers continue to work at a global SAN file system for 2005. Disk-to-disk is hot and getting hotter, so backup and storage management vendors will concentrate on compressing and managing backed up data on RAID. Tape vendors will continue to adopt disk technology, both within their own product lines and as front-end caches to tape libraries. iSCSI will make continued inroads into IP-based storage networking, and will be most useful in smaller environments without consolidation pressures. Vendors began introducing SMI-S compliant products in 2003, and will beef up their compliant product introductions into 2004 and beyond.
Storage hardware's interesting technologies and surprising developments--not to mention its never-ending competitive soap opera--will bear watching for a long time to come.
|Printer friendly Cite/link Email Feedback|
|Title Annotation:||Storage Networking|
|Author:||Chudnow, Christine Taylor|
|Publication:||Computer Technology Review|
|Date:||Dec 1, 2003|
|Previous Article:||CTR salutes those who made a difference in 2003: when the going got tough, the tough kept going.|
|Next Article:||Verbatim to deliver 8x DVD + R write-once media as drives hit market: new discs significantly reduce recording time, company ramps production as DVD...|