True data sharing - where any workstation can access any corporate data, irrespective of what it is stored on, what database it is held in, and what the server operating system is - is still a vision. Or, as EMC UK marketing manager Nigel Ghent puts it: 'True sharing of information and data is perhaps the vision beyond the vision.' But we are getting closer?
Only a few years ago, it seemed impossible. In the first half of the last decade, the traditional storage paradigm saw files held on disk attached to a specific server, stored in a format determined by the operating system, and usually held in a proprietary database management system. Data could only be accessed by a PC connected to that server, and equipped with the necessary software.
Storage developments since then have opened up the prospect of real data sharing as a realistic, albeit distant, goal. The first advance was the ability to share storage subsystems between heterogeneous servers. Typically this was done on the basis of partitioning the subsystem, where one server would be allocated part of the storage, and a second server the other part. Neither server could access the other's data, even though it was all stored on the same device.
One vendor took this concept further, and introduced a limited form of true data sharing. That was Encore, which introduced its Infinity SP subsystem in early 1995. This was the pioneer of the modern type of mainframe disk subsystem, with its own processors and Unix operating system, allowing the development of bespoke storage applications. It took IBM until 1999 to offer its mainframe users a disk subsystem with similar flexibility.
Technically, the main limitation of the Infinity SP was that data sharing was only possible with flat files, and not with database files. But the main reason why Infinity only sold in small numbers was commercial. Encore had no presence in the storage market, and never developed the marketing clout to build one.
As a result, its storage operation was eventually taken over by Sun Microsystems, which re-engineered and rebadged the Infinity SP, and now sells the evolved version as the A7000.
Sun's interest in Encore arose because it was trying to develop its own storage operation as an independent entity, rather than a division which existed to produce disk systems for Sun servers and workstations. Infinity seemed an ideal product to try to compete with IBM, EMC, and co in the mainframe market. But events overtook Sun before it really got started. The two important developments of the late '90s were, first, the arrival of Fibre Channel as an interconnect technology to replace SCSI, and second, the arrival of the storage area network, or San.
Sun quickly appreciated that this was the direction it had to go in, and concentrated on building up capability in these two areas. And very effectively, too: Sun had shipped over 3 petabytes of Fibre Channel storage by the end of 1999.
As a result, the concept of data sharing took a back seat. Other developments contributed to this, notably the arrival of replication technologies, such as StorageTek/IBM Snapshot, and EMC's SRDF and TimeFinder. Do you need data sharing, with all its complexity, when you can rapidly make copies of any data you want, and ship it quickly over fibre optic links to whoever wants it?
The answer, according to Mercury Computer Systems, was 'yes'. Mercury developed a product called SANergy, which allows file sharing with the necessary data locking between heterogeneous clients over a San. The product found a ready market, and has now sold into 3000 organisations worldwide.
The main European reseller of SANergy is Sagitta Performance Systems, which has over 100 customers in the UK alone. According to Sagitta managing director Andy Norman, most of the sales have been for niche applications - 'video and non-linear editing companies'.
He continues: 'The implementations today are mainly niche companies like BSkyB and Printcraft. They are building Sans for special functions, not for the enterprise. There are several reasons for that. Sky, for example, was keen to employ a new architecture, and was prepared to take the risk - they are a classic early adaptor.'
Last December, SANergy was bought from Mercury by Tivoli. Tivoli is keen to develop the product into an enterprise data sharing tool, and is taking a number of steps to achieve that.
First, it has repackaged the product so that, instead of being for one operating system (NT or Apple Mac or Unix), the same product supports all three. Second, Tivoli is extending the support to additional platforms and providing additional file system support.
Then, according to Tivoli Storage Management products manager Karen Dutch: 'We are putting a solution together - a combination of SANergy file sharing and Tivoli Storage Manager to do Lan - and server-free backup, restore, archiving, and HSM. Again in a heterogeneous environment - it doesn't matter what type of disks you've got, anything from JBODs, or mid range server disk, up to the most intelligent subsystem.'
All of these moves take the existing product and develop it to be more widely useable. The next step, known internally as the Storage Tank project, is more radical.
According to Dutch: 'We are looking to deliver in 2001 a brand new file system oriented towards enterprise San users. It will be a combination of what SANergy file sharing offers and a brand new file system, with features such as life cycle management. So it is not just file sharing; there will be policies to determine where a file should be placed, for backing up files, and for transferring files to tape. It is an extension of file sharing with full life cycle management.'
How will this work? IBM Worldwide SAN programme director Scott Drummond explains. 'The idea is data sharing by translating every byte of I/O into a common format, and then converting it into the form required by the operating system in question.' He adds: 'We have a prototype working today; the work that remains is to improve performance.' Tivoli plans to release this product, which will be known as Tivoli Data Management, next year.
Today, as HDS San and high availability solutions manager Vincent Franceschini puts it: 'The promise of any data, anywhere, any time is unrealistic.' But developments such as SANergy and Tivoli Data Management are bringing the goal closer, by moving data sharing from being a niche application, to an enterprise-wide capability. There is some way to go yet, especially as Sans themselves are still in their infancy, but the new storage world is visibly taking shape.l
Case study: BSkyB
BSkyB uses SANergy to create all its graphics for programmes on all the Sky television news, film and ports channels. The company found that the immense quantities of bytes needed for 3D images overwhelmed the IT infrastructure, which consisted of an Ethernet network with a Silicon Graphics host server.
According to visual effects supervisor Dave Sedgwick: 'If someone needed to work on a different workstation it would take longer to move the data than it would to complete the project. The whole process was very time-consuming and would leave us with very tight deadlines.'
Something more powerful was needed, and Sky found the answer in a San using a Gadzoox Gibraltar hub and SANergy file sharing software, put together by Sagitta Performance Systems. The storage is used by four Silicon Graphics Unix and three NT workstations.
Now, says Sedgwick, 'our 3D operators can work from any workstation, any time, and have direct access to the data they need.'
Case study: Printcraft
Printcraft is a digital pre-press company based in Dublin whose business, among other things, includes printing Microsoft technical manuals. Like BSkyB, Printcraft used to rely on an Ethernet Lan, but found it needed more disk capacity and faster file sharing speeds. The necessity to use the same workstation to complete projects was again a major limitation.
According to general manager Julian McDermott: 'We basically needed an entirely new, high-end installation that would provide us with excellent storage capacity and scaleability in preparation for future growth and centralised storage.'
Sagitta Performance Systems again provided the solution - a San using two Sagitta Fibre Storage Systems and a Gadzoox Capellix switch, shared between 16 workstations (12 Macs and four NTs). SANergy software again provides the necessary file sharing capability, allowing desktop publishing operators instant access to all centrally held files.
According to McDermott: 'The new San environment has allowed us to work from any computer, and has enabled us to achieve far more in less time.'