Tuesday’s HostingCon activity closed with a return to the keynote presentation theatre, to the panel discussion and the topic of cloud computing, as representatives from Parallels, Rackspace, Mezeo and SoftLayer weighed in on the subject.
They started out by defining cloud computing, which I’m going to skip over, rather than write out for the third time today – there’s a general sense of agreement. It’s a pool of computing resources that is scalable, easily provisioned and billed like a utility, etc.
Jack Zubarev from Parallels makes the interesting point that the business model should perhaps be disassociated from the technology, since the nature of the billing (whether billed monthly or hourly) isn’t really a clear enough distinction.
Frankly, there’s a lot of repetition here from the cloud panel earlier today. Interestingly, however, it feels like there’s a bit of a shift between this afternoon’s panelists (Parallels, Mezeo, Softlayer) that “anyone can offer or deliver a cloud service” than this morning, when the agreement was definitely more on the “this is out of the reach technologically of the average provider” (Microsoft, Google, Salesforce.com).
The common thread is Emil Sayegh of Rackspace, who seemed to be more in agreement with the morning’s group than this afternoon, but doesn’t seem to disagree with the afternoon folks quite enough for there to be any kind of argument.
I asked about it, and I think the response from across the panel seemed to be to clarify that smaller hosts have a lot of opportunity to partner with another provider, whether that provider is a company like Rackspace, with a fully-deployed cloud, or a company like Parallels or Mezeo, with a fully-developed piece of hosting software. A host can build a cloud based on that technology, paying roughly the same amount for hardware that any company with moderate scale would pay – the infrastructure being basically a commodity at this point.
One thing Zubarev brought up was who cloud computing is for – a question a hosting provider should ask when they’re looking at whether they should move into the cloud. So, cloud is particularly applicable to hosting customers who might expect traffic spikes, for instance. But there are all kinds of hosting customers (picture your couple-page brochure-ware site) that will never see a traffic spike.
The question this raises for me is how many companies can realistically expect to keep making a living serving brochure-ware customers. Those are customers that are going to find a cheaper (if not free) provider somewhere down the road.
