Elastic computing

Is it feasible?

We had an internal technology session last night and I was on a panel where each of us was asked to come up with the IT aspects of an enterprise architecture for a small (100 people) financial services organisation. My approach was based upon building up a service-oriented architecture (SOA) and then deploying it onto the cloud, which can then scale with the business.

Elastic computing

Elastic computing makes a lot of sense in this context for several reasons, but the primary one is that information technology probably isn't (and arguably shouldn't be) the core focus of most small organisations. In other words, take advantage of software as a service, and have that scale with your business by dumping it on the cloud.

While some businesses are using this approach, there are some "interesting" questions to be answered about cloud computing before I think I'd actually propose it for some types of business. First off, there's data privacy and security. I don't want my data on the cloud and I'm sure that many businesses feel the same way. Additionally, there are some other issues around compliance and interoperability (see Cloud computing hysteria paralyzed by bolt of reality for more details on this).

Cloud computing might not be feasible for some applications at the moment, but it is another approach for us as architects to evaluate.

"King Cloud" photo by akakumo.

About the author

Simon is an independent consultant specializing in software architecture, and the author of Software Architecture for Developers (a developer-friendly guide to software architecture, technical leadership and the balance with agility). He’s also the creator of the C4 software architecture model and the founder of Structurizr, which is a collection of open source and commercial tooling to help software teams visualise, document and explore their software architecture.

You can find Simon on Twitter at @simonbrown ... see simonbrown.je for information about his speaking schedule, videos from past conferences and software architecture training.



Re: Elastic computing

Another issue with this form of computing is with the non-functionals and SLAs. Having spent the last few months working on a grid computation system, I often see fantastic throughput but terrible latency. e.g. getting 99% of the job done in minutes but waiting an hour for the last 1%. As it's in a 'cloud' you often lack control over *how* your work is actually executed.

These issues are solvable but hard..

Re: Elastic computing

That's a good point and, after the session last night, we were talking about whether the cloud could migrate your data/apps automagically, depending on where they were being accessed from. For example, imagine if the cloud could adopt a follow-the-sun pattern, moving stuff from Europe over to the US at close of business in Europe. Regardless, you can't get away from the latency, even *if* your data/apps reside in the same geographic region.

Guess we'll need a big cache. ;-)

Re: Elastic computing

There are so many problems with cloud computing with respect to a financial institution! You've mentioned security and compliance gets a good shoeing in the linked article too. Solutions for latency issues are little more than hosting providers currently offer - hardly elastic or cloudy.

You might be able to stop worrying about maintaining the clerical functions and even some of the analytical functions. You may well cut a lot off your capex and opex but only in a few areas.

Re: Elastic computing

My thoughts exactly ... at the moment, only the non-business critical applications (e.g. ops, etc) lend themselves to being dumped on the cloud. Out of interest, do you know of any financial organisations that don't own their own data centres/host their own server rooms?

Re: Elastic computing

If by "host their own" you mean own the facility then I think I know of one that might fit the bill but perhaps that a discussion for another time...

Re: Elastic computing

Why aren't people talking about clouds instead of the cloud? A banking cloud might stand a chance of being compliant and still providing some of the benefits we're talking about. It may also offer some very neat solutions for low-latency and existing compliance headaches.

It's also been pointed out that the nature of compliance is not fixed - it should adapt to suit the changing ways in which we do business rather than simply constraining them.


Add a comment Send a TrackBack