A prospective customer has to investigate the cloud supplier. Is the supplier competent? And reliable? What do their existing customers make of the service? The service may need to be audited. Does the supplier take backups? Where are they stored? Are they any good in an emergency? What do they know about security? Who's going to get priority when there's an emergency? Are their staff trustworthy? Which jurisdictions do they operate in? Contracts have to be negotiated. Then someone has to work out all the scripts to transfer code and data into the cloud. Or from one cloud supplier to another.
And so when you run an OJEU procurement, you magically DON'T do any of this? These costs are inherent in ANY outsourced procurement; arguably they are minimised in G-Cloud because of the pre-selection, but even if not, they're no different from any competition. You could also argue that in cloud services, contracts *don't* have to be negotiated - you take or leave what is offered, and it is up to YOU to manage the risk - but then that's the same no matter what procurement route you follow.
Regarding up-front costs, my reference wasn't to due diligence, but to the capital costs of setting up a service in the "traditional" way - i.e. buying dedicated kit, delivery, installation, configuration, testing, even before you get to the application. This kit has to be specced to handle peak load (otherwise you get a debacle like the census launch), for a 5 year contract, and woe-betide if things change in that time period...
Oh, and you might want to ask somebody in Central Govt how much / how long it takes to run an OJEU procurement; typically the *shortest* cycle is 6 months, and big contracts can be 12+ months.
Regarding the Guardian article on the "*Amazon* marketplace for traders" (http://services.amazon.co.uk/services/sell-online/how-it-works-pro.html), this is NOTHING to do with cloud computing, it is for organisations selling goods and service to the public, using an online shop. This is no different from eBay changing their cut of sales, or a council upping local rates.
The Wired article is referring to the *AWS Reserved Instance Marketplace* (http://aws.amazon.com/ec2/reserved-instances/marketplace/), which allows Cloud users to sell virtual machine reservations they have bought bought but no longer need. See this article for more information about what a Reserved Instance is, and why you might use it: http://aws.amazon.com/ec2/reserved-instances/
Finally, there is also the AWS Marketplace (https://aws.amazon.com/marketplace), which is the cloud equivalent of the physical marketplace, to allow software companies to sell pre-installed VMs, support services, etc. AWS cloud customers have *no* requirement to even visit that marketplace, but Amazon provide it as a value-added service.
Could Amazon increase their cloud pricing? Yes, they could; however the 7+ year trend is in ONE direction, down - and in the event they do increase the price, the cloud model means that the customer can move - no long-term lock-in, no 5 year contracts to break, no waste of CapEx, just choose another provider.
Of course, portability between clouds can be problematic, but this is NO different from between two "traditional" providers, and in many cases, the use of standards such as OVF for VM migration means that cloud providers are ahead of the curve.
Friday, 3 May 2013
Cloud computing – a fairy responds
The following anonymous comment on Cloud computing – away with the fairies has been received and may be commented on later:
Posted by David Moss at 11:19