It was like looking into an archaeologist's vault. Remember Mammoth and SDT and earlier drives? They were all there. Winter and his team have had a lot of practise at getting data off dead formats and re-writing it onto newer formats or external hard drives.
The fluttering ribbon of tape was sucked onto the reel which started revolving until the tape began wrapping over itself, was gripped and Presto! A working tape system with a readable and writeable reel. Impressive. . .but now defunct.
If a customer discovers some old StorageTek tape reels in a forgotten corner of their office, Kroll's tape archaeologists can read them, extract the information and copy it to newer media. They can do this for more than 200 different formats.
How do they get the drives? According to Winter, eBay was a big help, with firms moving on from a tape format selling equipment for whatever they can get.
Tape won't just die with a big bang and become extinct through some catastrophic event. Tape will die because it won't renew itself, and it won't renew itself because customers up and down the kingdom take individual decisions to stop writing data to tape and put it on disk or optical disks instead. It will take years to die.
Here's how it goes. For a while the data on a tape format -- Exabyte's first VXA format, say -- is accessible. But gradually, while stored in a cupboard gathering dust, the server with the VXA drives is replaced. The tapes are retained but are in limbo, in tape's halfway house -- not yet dead but not fully alive. The software that was used to write to them is replaced too. Now the business has no drive to set the tapes spinning and no software to interpret and read the data from them either.
But Winter's team does. They have the old drives and the old software and a variety of servers and interconnects so they can disinter information buried in cold, dead tape reels and cartridges.
The trouble with tape is that there is no graceful exit. Once you have accumulated tens, hundreds, thousands, or even tens of thousands of tapes and you want to move on, can you justify the weeks and months of time it will take to transfer the information to a new medium? Often the answer is no, so you carefully store the tapes in a cupboard just in case, some time later you move to a new job, your replacement is appointed, the department is re-organised, another manager comes along and, one day, the cupboard is opened and somebody says, "Oh my, what do we have here?" Another person utters the dreaded word "compliance," and Richard Winter gets a call.
But businesses with tape libraries face a bigger scale of problem altogether. If you have a StorageTek Powderhorn, you'll have thousands of tapes and when that massive archive reaches end-of-life, you have a large disposal problem.
There needs to be a way to make a more graceful exit from large-scale tape infrastructures than just having a few old reels or cartridges in a forgotten cupboard somewhere. It should be given serious consideration when decommissioning and be a carefully executed project. We could see organisations like Kroll, and maybe Iron Mountain, starting up tape preservation temples. They'll need all the libraries needed to read and handle a tape format, and offer long-term tape reel/cartridge storage and data migration services.
But, there are no signs of it happening yet at the larger end of the tape-using market. Tape is still unmatched in terms of cost per gigabyte stored when dealing with terabyte volumes of data. But that day will surely come. There's no Moore's Law here, no immutability about tape. It will die. In ten or 20 or 50 years the very last vast tape library will be switched off and its robotics will cease their darting, hissing movements.
If you have a modern tape library look on it and think of Richard Winter and his team at Epsom. The chemistry of tape recording layers is fantastic, a triumph of technology, and tape is an amazingly efficient use of data centre space. But it will die.
Once people looked at the vacuum-inducing, open-reel StorageTek drives and thought the day was bright, the future golden and technology wonderful. So it was, but technology is refreshed and refreshed and today's state-of-the-art examples, the LT04 drive costing tens of thousands of pounds and next year's LTO5, are both going to end up in Richard Winter's cupboard. Just more relics for the tape archaeologists to fetch out, use a few times and then put back in the vault.
Tape's future, its long winter, is in a cupboard in Epsom. Get used to the idea.
Chris Mellor is storage editor for The Register.