Container port
Image by dannybirchall via Flickr

I took a quick trip to the Catalan city of Barcelona earlier this month; surely the first time I’ve been to that city without seeing the sea. I was there as a guest of HP, to hear some of the latest happenings in their storage business.

Over two days, we heard plenty about the P9500 (one can only presume that, although the numbering sort of fits the product line, the product manager hadn’t really absorbed Arthur C Clarke‘s point), the acquisition of 3Par, and the importance of intelligently shuffling data from fast but expensive storage to slower but cheaper varieties of disk.

As others who were there, such as Chris Evans, Chris Mellor and Ilja Coolen covered much of the news, there’s no need for me to rehash that. And a special episode of the Infosmack podcast took shape in a Spanish hotel room for those who’d like to hear some immediate impressions.

I thought it was interesting to hear honest impressions from HP staff involved in the partnership with Hitachi that resulted in HP’s new P9500; impressions that seemed to be at odds with those some of my colleagues heard at a very similar Hitachi event just weeks earlier. Everyone seems to be clear that the P9500 (and the very similar VSP that Hitachi’s HDS sells) is essentially a piece of Hitachi hardware. There’s nothing particularly surprising about that. The quibbles are around the extent to which HP engineers were involved in the process that designed, refined and tested the pieces that ended up forming the final box. Things are complicated still further by competing assertions as to the value that some additional HP code delivers to their customers. Chris Mellor has more on the story over at The Register, quoting a response by HP’s Simon Brassington to Mellor’s original review of the P9500;

“your description of ‘slapping a label‘ on the VSP doesn’t do justice to the collaborative partnership that HP and HDS have enjoyed in developing these products over the last 12 years.”

Hitachi subsidiary HDS (which sells the VSP) appears to disagree, with Mellor going on to highlight very public differences of opinion as to who did what.

Possibly some genuine misunderstanding. Possibly some enthusiastic spin on incremental features. Possibly the realities of life in a global corporation, where individuals have incomplete views of work being done elsewhere in the hierarchy. Hopefully no attempt on either side to deliberately mislead anyone. Whatever, these partners maybe need to take a deep breath, grow up, and remember the benefits that each continues to gain from the other.

Away from sniping about who was most important in deciding the colour of the P9500’s paint job, I was intrigued to learn more about HP’s ongoing efforts to cram computers into shipping containers. They’re not the first to do this, but one of the announcements in Barcelona did suggest that the company is the first to take the process mainstream. With much talk of Henry Ford, they unveiled POD-works assembly line plants in Texas and Scotland. The US plant can spit out a server every 12 seconds, and deliver a fully configured POD container in as little as six weeks. Impressive stuff, and a conversation (in a POD, naturally) with HP’s Eva Beck suggests that there are plenty of customers interested in the proposition. According to Beck, around half of POD deployments are in the open air. The other half go into buildings that protect them from the elements; but these buildings are much cheaper to build and run than a traditional data centre with its air conditioning, raised floors, fire suppression systems and the like. In many cases, they’re simply empty warehouses.

Despite use of the internationally standardised shipping container (both those big ones and the wee ones you don’t see as often, to be terribly technical about it), there seems to be little interest in moving these things around very much. This makes some sense as, although the container itself can fit on the back of a lorry, servers tend not to like being moved very much… and the container has significant requirements for power and water that your average truck stop might struggle to meet.

I liked the idea behind the POD, and it was interesting to see the way in which HP – and real customers – are putting it to work. In many ways, though, I wonder if the shipping container is maybe too big? The economies of scale kick in when you ring up HP and order a complete pod, stuffed to the rafters with servers. That’s an awful lot of computing power, and it doesn’t play so well in an environment where incremental enhancements are being made to an existing IT infrastructure.

Surely there’s a market for something similar, but smaller? There may also be a market for something a little more rugged, so that it really can be used in different locations? Why not deploy a baby POD to a sporting event, a new office, or even a disaster area? Maybe we’ll get there one day. In the meantime, I look forward to hearing more stories of ways in which the current POD concept is being stretched.

And given all the talk of Henry Ford, why were none of the shipping containers we saw, either on screen or in the flesh, painted black?

Disclosure: HP paid my expenses on this trip. There was no expectation that I write this in return, and the words and sentiment are entirely my own.