| |||
|
|
||
|
|||
| |
|
Home > Storage Technology News > IBM unveils SAN File System |
| |
|
|
|
|
| Storage Technology News: |
|
||
|
|||
|
|||
|
|||
|
|||
|
|||
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||
Some might say that a networked storage architecture can not reach its true potential until data can be shared across all the servers and storage in a network. It appears, at least according to IBM, that time has arrived. Big Blue on Monday announced its long-awaited TotalStorage SAN File System (SAN FS) with big hopes that it will provide common file services in a storage network, with the help of a metadata appliance. Based on the Storage Tank technology developed by IBM's research unit, SAN FS is designed to provide a single, centralized point of control for managing storage devices and data. SAN FS enables policy-based file provisioning, distributed file locking and file-based FlashCopy to help automate and speed information-sharing and data-management tasks. Jai Menon, IBM's director and chief technologist for storage systems architecture and design, began the Storage Tank project five years ago, alongside the research team that created Tivoli Storage Manager. He said that SANs provide physical connectivity but not logical connectivity, which is a problem that needed to be solved at the file system level. Instead of installing a file system on each individual server connected to the SAN, SAN FS resides on a metadata server that runs out of band, meaning that data traveling among servers and storage does not travel through them. Traditionally, each server connected to a SAN is allotted its own set of disks within the storage network. The SAN FS technology lets all servers and applications access and store data anywhere on the SAN rather than being relegated to a specific logical unit. The SAN FS metadata server keeps track of each file's location in the SAN, but allows data traveling among arrays and servers to move across the SAN directly. "What we're really trying to do here is say that it's very limiting, from a management point of view, to manage [data] on a server-by-server basis instead of at the enterprise level," Menon said. "Think of it as providing common file services to all of your servers on a SAN. Now, when you create a file, the SAN File System determines where it goes. When data is deleted, that space is available to [all of] the other servers. You can create data from one operating system and read it back from another." Menon said an agent must be loaded onto each server to communicate with the metadata appliance. But, he added, IBM hopes to make its interface a standard, so that it can be built into operating systems, eliminating the need for installing a multitude of software agents. IBM is targeting its SAN FS at the financial services, retail and life sciences industries, as well as at end users with large data stores and grid computing applications. "The key thing to take away, however, is that file systems and this type of technology in general are very hard," said Steve Kenniston, a technology analyst with Enterprise Storage Group Inc., Milford, Mass. "People may give IBM a hard time for them taking a while to deliver this technology. However, this type of technology is not something that is just developed in a few months and thrown out into the market. IBM, with their install base, customer loyalty and trust, has the ability to now bring clustered file systems to the masses." Marc Farley, an author, president of Building Storage Inc., and a SearchStorage.com expert , said IBM's technology has the ability to change underlying storage for the file system on the fly, without downtime or performance degradation. "[The SAN File System] provides a single file system interface that is easy to manage and install [applications] on, but allows administrators to select which files and directories are protected by remote copy," he said. He said users that have problems managing storage growth for databases will also realize scalability and cost efficiency benefits, but they first need to migrate the database data into the file system. "So far, most of the excitement in storage has been at the networking and storing levels, but the filing level has been more or less ignored, probably because file systems and databases are inherently complicated," he said. "If you build a file system with rich metadata, as IBM has done, it's now possible to get a lot smarter about where to store data." SAN FS will be available Nov. 14. A starter configuration will be priced at $90,000 and will be available through resellers and from IBM's newly created storage software sales team, which was specifically built to sell SAN FS. The $90,000 SAN FS starting configuration includes the file system itself, two IBM eServer xSeries 345 Metadata Server engines with software preloaded and an eServer xSeries 305 server for the master console. It includes IBM Director and Tivoli Bonus Pack for SAN Management for 64 ports. Prices for the virtualization software scale based on the number of processors in the Metadata Server engines and on the number of the Connected Application Servers connected to the SAN FS. The entry prices above are for two Metadata Server engines with two processors each and one Connected Application Server with two processors. Let us know what you think about the story; e-mail: Kevin Komiega, News Editor. Live webcast this week: "IBM's StorageTank - It's finally here. Now, what?" IBM nails down dates, performance for storage virtualization
|
|
|
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||
|
|||||||||||||||||||||||||||
| |||||||||||||||||||||||||||
|
|
|
|||||||||||||||||||||||||||||
|
|||||||||||||||||||||||||||||||
