Gimme shelter

In 1979 - two years before William H. Gates III said '620K should be enough for anyone' - Messrs Egan, Marino, and Collins...

In 1979 - two years before William H. Gates III said '620K should be enough for anyone' - Messrs Egan, Marino, and Collins founded a storage company which, because it happened to coincide with their surnames and that of another bloke whose name began with C, but who pulled out between registration and the commencement of operations, they called EMC2.

With 38 per cent of a market, estimated by IDC to be worth $46bn (£28.8bn) by 2003, the lads have done well, and their company has accordingly been targeted with the sort of friendly competitive banter traditionally reserved for IBM. EMC, say the knockers, locks in the unfortunates who purchase its system, doesn't conform to everyone else's standards, generally goes its own way, sells more than we do, and it's not fair - as Ryal Poppa had it when he headed rival STK (StorageTek) 'that's typical EMC - if we ain't got it, you don't need it'.

EMC's view of the matter is unsurprisingly different, and UK and Ireland marketing director Nigel Ghent points out that the existing fibre channel standard was put in place by Ansi - 'nothing to do with us, it was a done deal'. EMC submitted a 'blueprint proposal' to the IETF (Internet Engineering Technology Forum), the SNIA (Storage Networking Industry Association) is only an advisory body, but in any case it invited an EMCite to be its current chairperson, and his company's products are compatible with 'thirty six different server and operating system environments.'

So a punter with EMC on his San could just bung on some STK kit and then sit back and admire the harmony? Mr Ghent says if one is 'on the podium' then one should anticipate carping from the masses and no, not exactly, as STK doesn't appear to be on his list. Yes, EMC can appear to be priced at 50, a hundred, even 150 per cent more than competing solutions, but he reasonably points out that one does not necessarily complain that a Ferrari costs more than a Ford just because of the similar amounts of metal involved. The hardware, he emphasises, is unimportant in the sense that EMC just picks the best bits available, and makes them work.

On to STK, where marketing director Robert Huntley says they would say that wouldn't they; did you know Lufthansa now backs up in 13 seconds what used to take 13 hours, guess whose system they use, and quotes STK's 'open San offerings' document which says that 'without cross vendor, platform and storage interoperability, the network environment becomes chaotic and unreliable'. So a punter could have STK and EMC on the same San? Not exactly, but STK stuff will work with just about everything else, and then there's virtual storage.

To the ignorant the concept of virtual storage may seem strange. Virtual memory, yes - it goes in and out rather like the legs in the hokey cokey; but storage is, well, storage; how would one render it virtual? Mr H explains patiently about tape technology leadership, and eventually concedes that virtual storage might not be unadjacent to a technologically hip version of asking the librarian to load a tape; the medium, he says, is unimportant - it's the way it's managed that counts.

If there is a consensus in the fascinating world of storage it would appear to be that advances will be software based, while the hardware will somehow or other lope along, and continue to provide a bottomless pit for the waxing hoards of the world's computerised squirrels.

Chuck Foley, vp of marketing at 'enterprise network infrastructure technologies supplier' Inrange, recalls US installations where tape couriers 'whizzed around the storage racks on roller skates', and confirms the opinion that virtual storage is 'pretty much the same thing, except you've got an automated gopher'. The physical technology, he thinks, is unlikely to change in the near future; 'basically, you've still got brown disks, magnetic heads and tapes, but the advances are going to be in the way that data is managed.'

Paul Ellis and Nick Tebellion at Amdahl Software took a similar line to Chuck Foley; both pushed openness, whilst adding the increasingly familiar 'EMC and the rest' side swipe. Neither anticipated an imminent quantum leap in storage hardware, neither thought it mattered as 'technology always keeps up somehow', and Paul Ellis emphasised that Amdahl Software was 'no longer a cost centre, but a separate company with its own P and L'. Which may sound trite, but don't forget Gene Amdahl knocked out the architecture for the IBM 360/50, in APL, on three sheets of American quarto.

Brocade Communications is 'in storage area networking', claims '90 per cent of the fibre channel market', and EMEA (which you've undoubtedly guessed stands for Europe, Middle East, and Africa) marketing manager is Paul Trowbridge.

Brocade, it appears, provides the glue which makes all the other bits stick together; 'infrastructure' gets mentioned a lot, but the fibre channel doesn't need to be made of fibre - a couple of copper wires would do because 'the fibre channel' is, 'well, just a name for a solid connection'; and no, 'a leap of the relays to chips variety isn't just around the corner'.

At this stage - particularly with A Well Known Vendor asked to quantify an exabyte and responding 'we have here the best techie in the country, he ... er .... hasn't got a clue, but he knows it's a seriously big number' - a few definitions might be in order. It's very simple. A thousand bytes is a kilobyte, and a thousand of those is a megabyte; then it carries on stepping in thousands, and you get giga, tera, and peta bytes up to exabytes, which is ten to the eighteenth; and, like the man said, is a seriously big number.

Robert Huntley at STK says you can get one point two terabytes into something the size of a Coca-Cola machine. Logic would seem to dictate that, if storage requirements are doubling every 18 months, today's one machine user will be requiring a floor space upgrade to accommodate four Coca-Cola machines in three years' time. Not necessarily.

Northern European sales and marketing manager Mark Hiley claims LSI is the biggest storage OEM in the world; they could get 8 terabytes into a Coca-Cola machine; they can get 73 gigabytes into a one-inch high disk; next year it'll probably be 146 gigabytes, and there's no way the world's going to run out of storage 'until they reach the molecular level of chips'. Hyperbolic or not, he's rather nice about EMC whose product he describes as 'a San in a box'.

Ray Rice is business manager at CMS Peripherals; his PR agency bills him as 'having been in the storage business for 15 years', and he knows about Ian Dury being wrong (since you ask, it was the far superior B-side of Hit me with your Rhythm Stick) on atoms being the littlest. Mr R can also name every one of the Blockheads but, in passing, mentions quarks which are the really tiny bits that comprise atoms, and which - he says this has been mentioned in New Scientist - whilst spinning, represent a possible new wave in storage technology. Mr Rice describes this as 'serious vapourware', and reckons conventional storage hardware will last 'about another five years'.

Ray Rice's view is shared by Ideal Hardware's Ian Lockhart: 'There has to be a parametric saturation point', and 'there's just so many times you can keep cranking the handle; I'd give it no more than five years.' StorCase Technology director Ajay Lukha makes the point that 'as disk capacities increase so do the performance demands on the hard disk drive and its mechanical capabilities, such as heat dissipation. We are now seeing spindle speeds of up to 15,000 rpm, which is 3 times faster than 5 years ago.'

There is, surely, a limit to the number of quarts that even the smartest software can squeeze from the storage pint. The world's first (1980) gigabyte capacity disk was the IBM 3380 which - roughly the size of a respectable fridge - weighed in at 550 pounds, and cost $40,000. Today's one gig IBM Microdrive is about the same diameter as a 2p piece, weighs less than an ounce, and 'will be available for less than $500'.

Dietmar Wendt, IBM's vp for storage solutions, expects the Microdrive to be the subject of 'a major mobile phone deal'. Great - something else to amuse the serial idiots when they're not saying 'hello darling, I'm on the train'. Could the Microdrive capacity hit two gigabytes? Possibly, but Mr Wendt is 'not excited by hardware' which, he says, 'is not a big issue - what's important is giving people what they want, which is manageable, secure, and affordable solutions.'

Big Blue pandering to the punters? 'Yes; we've learned our lesson, you don't get anywhere trying to lock people in, and we're wide open.'

Like everyone else?

'Almost, EMC's holding back a bit.'

Read more on Networking hardware