The wisdom of the crowd seems to have decided that both cloud computing and its sibling SaaS are cost plays. You engage a cloud or SaaS vendor to reduce costs, as their software utility has the scale to deliver the same functionality at a lower price point than you could do yourself.
I think this misses some of the potential benefits that these new delivery models can provide, from reducing your management overhead, allowing you to focus on more important or pressing problems, through to acting as a large flex resource or providing you with a testbed for innovation. In an environment where we’re all racing to keep up, the time and space we can create through intelligently leveraging cloud and SaaS solutions could provide us with the competitive advantage we need.
Could and SaaS are going to take over the world, or so I hear. And it increasingly looks that way, from Nicholas Carr‘s entertaining stories about Sameul Insull through to Salesforce.com, Google and Amazon‘s attempts to box-up SaaS and cloud for easy consumption. These companies massive economies of scale enable them to deliver commoditized functionality at a dramatically lower price point that most companies could achieve with even the best on-premises applications.
This simple fact causes many analysts to point out the folly of creating a private cloud. While a private cloud enables a company to avoid the security and ownership issues associated with a public service, they will never be able to realise the same economies of scale as their public brethren. It’s these economies of scale that enables companies like Google to devote significant time and effort into finding new and ever more creative techniques to extract every last drip of efficiency from their data centres, techniques which give them a competitive advantage.
I’ve always had problems with this point of view, as it ignores one important fact: a modern IT estate must deliver more than efficiency. Constant and dramatic business change means that our IT estate must be able to be rapidly reconfigured to support an ever evolving business environment. This might be as simple as scaling up and down, inline with changing transaction volumes, but it might also involve rewriting business rules and processes as the organisation enters and leaves countries with differing regulation regimes, as well as adapting to mergers, acquisitions and divestments.
Once we look beyond cost, a few interesting potential uses for cloud and SaaS emerge.
First, we can use cloud as a tool to increase the flexibility of our IT estate. Using a standard cloud platform, such as an Amazon Machine Image, provides us with more deployment options than more traditional approaches. Development and testing can be streamlined, compressing development and testing time, while deployed applications can be migrated to the cloud instance which makes the most sense. We might choose to use public cloud for development and testing, while deploying to a private cloud under our own control to address privacy or political concerns. We might develop, test and deploy all into the public cloud. Or we might even use a hybrid strategy, retaining some business functionality in a private cloud, while using one or more public clouds as a flex resource to cope with peak loads.
Second, we can use cloud and SaaS as tools to increase the agility of our IT estate. By externalising the the management of our infrastructure (via cloud), or even the management of entire applications (via SaaS), we can create time and space to worry about more important problems. This enables us to focus on what needs to happen, rather than how to make it happen, and rely on the greater scale of our SaaS or cloud provider to respond more rapidly than we could if we were maintaining a traditional on-premises solution.
And finally, we can use cloud as the basis of an incubator strategy where an organisation may test a new idea using externalised resources, proving the business case before (potentially) moving to a more traditional internal deployment model.
One problem I’ve been thinking about recently is how to make our incredibly stable and reliable IT estates respond better to business change. Cloud and SaaS, with the ability to shape the flexibility and agility of our IT estate to meet what the business needs, might just be the tools we need to do this.
I agree entirely, and have ranted on the topic before. In particular, I see deployment options as one of the biggest advantages. Releasing a new version of a system becomes a dream. You fire up new images containing the new software whilst the production version keeps on ticking. Perform some tests to ensure that the new version works, then switch the load balancer over to point to the new servers instead of the old ones. If anything goes wrong you simply switch back. Nothing could be easier.
I saw an interesting article the other day which talked about continuous deployment, which I think is also relevant: http://timothyfitz.wordpress.com/2009/02/10/con…
My original rant, available at http://www.brucecooper.net/2009/01/cloud-comput…
Reminds me of a post I've written last year “Don't focus too much on costs, cloud computing is about business agility” http://www.capgemini.com/technology-blog/2009/0…
This is a well worn argument, and I think anybody on the inside looking out already knows that it's all about 'I want it now' rather than $s. The only significant consideration on the cost side is that cloud computing and SaaS present a ramp function rather than a series of steps; and you can go off and have a lot of fun figuring out when the ramp is above or below the steps for various use uses and user populations.
A view on this that I took a little while ago is that 'cloud' is really all about management. The companies that are attracting attention right now (like Amazon, Google and SF.com) have achieved tremendous economies of scale (and scope) over a bunch of assets that they own by having better management platforms than the typical enterprise. An inversion is coming though, and people are bottling that management goodness into things that can be used at much smaller scale – right the way down to the individual device/user (something that I wrote about last month at http://wp.me/palij-3a). Something like Ubuntu Enterprise Cloud (which bottles the Eucalyptus flavour of AWS style management) lets you deploy AMIs onto your own machines.
Completely agree. Having spent the last 6 months pulling the plug on our datacenter and migrating to EC2 the benefits of cost are far outweighed by all the flexibility you get. Like for like, we cut costs by about 40%. That was mostly the reduced system administration burden we needed to carry (new breed of sys guys that are far more dev like) and no longer carrying heavy power and carrier charges. But why go like for like? We span up more environments, became more fault tolerant, could run more dev environments and try out new things. Before long we quickly realised that we had lumps of hardware kicking around that were worth not a lot, but that could more than happily run eucalyptus. Why should we pay amazon to run our images? We ran private cloud as well as public. For pretty much the same cost we got double the infrastructure across zones plus a shed load of dev environments. Our releases started happening fast and fluid and my dev resources started to want to release as it meant they weren't wrapped up in the red tape of the datacentre.
Nice! None of this is rocket science, but it does require us to change the way we think about IT.
It seems that the cloud bandwagon wants to push the cost angle in exclusion to everything else, as you point out. I even read an article on TechCrunch the other day, written by the CEO of cloud vendor, arguing that we've crossed the chasm and are in the tornado (there was a lot of metaphors in the article) and cost efficient cloud will take over the world.
It not that the ideas in our posts are new or original, but they do seem to be overlooked by a lot of pundets.
I'm not sure it's all “I want it now” as “it's not worth my time to worry about”. We're all time poor, and cloud is a nice tool to create time to worry about something other than infrastructure, even if it's only a private cloud. As you point out, cloud is a nice tool for bottling up the expertise require to run a virtualised environment.
As with all good arguments, there's nothing original in what I wrote; it's all just common sense. Unfortunately common sense is not all that common 🙁
So it's about time (whether we're talking about latency or duration). Maybe this finally proves that time!=money?
It's not just infrastructure that we don't have to worry about. If you can position yourself to be a big SaaS user then you can stop worrying about an awful lot of things. There is however some residue:
* Identity management remains a huge issue, though I have high hopes that serious progress will be made in 2010. There's a lot of good stuff that can be back ported from enterprise identity management, and not too many places that need to be touched to push past 80:20 coverage.
* Data portability, and particularly the 'what do I do when my provider goes bust' (or I fall out with them) remains an issue. This problem was never really fixed in the enterprise, and so remains somewhat intractable except for trivial cases like email.
* Security remains a double edged sword. Most service providers are doing security better than the users of their services could ever do it themselves (with the possible exception of large enterprises with well resourced security teams), but each service provider also represents a pool of risk aggregation.
Often we forget the little guy, the SMB, in our discussions of the comings and goings of the Internet marketing industry. Sure there are times like this when a report surfaces talking about their issues and concerns but, for the most part, we like to talk about big brands and how they do the Internet marketing thing well or not so well.
http://www.onlineuniversalwork.com
I've been thinking for a little while that we should stop saying “time = money” and start thinking “money = time”. (Let's assume that this operation is not commutative. 🙂
Your on the mark with the residue, which is are really the messy details of making cloud work. I don't see any of this as insurmountable (assuming that we're happy with a “good enough” solution).
[…] I’ve pointed out before, there’s a lot more to cloud than simply reducing costs. The biggest benefit is probably the agility that cloud can bring to your IT estate, leveraging a […]