The cloud computing issues that confront enterprise systems architects today, are really not much different than...
the architectural issues we have been facing for decades, argues enterprise architecture expert David Linthicum. Service-orientation came about as a way to maximize the use and reuse of system resources. SOA aligned particularly well with Web applications and, according to Linthicum, the coming wave of cloud technologies is an extension of the core concepts behind SOA.
In June of 2011, David Linthicum presented a session entitled, "Handling data integration challenges in the cloud," and participated in the expert panel discussion, "What can go wrong in the cloud?" at the SearchSOA virtual event "SOA in Action, Navigating SOA, Integration and the Cloud." The main question from the panel discussion was – as the title suggests – about problems that enterprise organizations run into as they implement new cloud technologies as a part of their IT architecture. In this SOA in Action recap David Linthicum responds to SearchSOA.com Site Editor Jack Vaughan's question about where cloud computing implementations fail to contribute to the greater enterprise architecture.
SearchSOA.com: David, what can go wrong in the cloud?
David Linthicum: The biggest issue that can go wrong with cloud computing really is around architectural planning. I think that people who do not plan how clouds are implemented within their enterprise - public, private or hybrid, it matters not - are really going to be running into huge issues. So, you can’t move into cloud computing without dealing with architecture. And typically I have to approach architecture with more of a service-oriented approach. That’s the biggest thing that I see wrong with the cloud right now.
When I get into these enterprises sometimes there hasn’t been a lot of forethought and thinking around how they’re going to leverage cloud, how they’re going to move to this environment, and pick the right systems that need to be moved, and then find the right migration and transformation path into the cloud. Then they end up making mistakes and they have to back up and fix things, and that gets costly, and ultimately we’re moving into cloud computing to become more effective and efficient.
The second thing is people don’t consider performance. The cloud doesn’t always provide the best performance, believe it or not. I understand there are massive amounts of scalability there and you can do these queries in a matter of minutes, which may take a matter of hours on local systems, and those capabilities are available in the cloud. But if you port a rather chatty application to the cloud and it’s communicating down to a client, there’s latency involved and there can be some performance issues in doing that.
The other thing to consider is cost. Clouds aren’t always cost effective, [and that goes for] private, public and hybrid. So you have to sit down and do the business case model in terms of the viability of the cloud within a particular enterprise, within a particular problem. You need to ask what kind of business problems you are looking to solve, and then try to figure out what the real costs are, and what the real savings are when moving to this cloud computing environment.
Now one of the things I urge people to consider is the value that you get from the agility aspect of it. The ability to change your enterprise processes to accommodate the needs of the enterprise, in almost real-time. That ultimately is probably the greater value of the cloud.
The other things are all things that I’ve mentioned before. Security is always an issue, however, I think if you put some security planning into cloud computing you’ll be just fine. As long as you put some smarts into it, chances are you’re going to find cloud computing systems are as secure, if not more secure, than a lot of internal systems that you see every day.
So, people have a tendency to make two mistakes out there. What goes wrong in the cloud is they assume the cloud can do too much and they start throwing things out into the platform of the cloud and typically don’t get the benefit [they expect]. Or, they assume the cloud can do very little and it’s a security issue and a control issue, and therefore they don’t leverage the cloud, and they don’t take advantage of this technology. Either is a problem. Either is failure.
A lot of the issues that we have with existing internal systems and building these systems, really transfer into the cloud. I think that we’re able to make those mistakes a little faster.
See related SOA in Action Recap story
Melvin Greer on what can go wrong in cloud computing – SearchSOA.com