Integration is at the center of major disruptions in IT today. SearchSOA.com recently spoke with David Linthicum, senior vice president at Cloud Technology Partners and SOA and cloud computing thought leader, about three key topics: integration, the cloud and "big data." The discussion ranged from cloud integration to security issues in the cloud and the workings of Hadoop.
In particular, Linthicum gave insight into cloud integration best practices, including how companies can determine which services are best for the cloud. The heart of his guidance focused on evaluating existing enterprise systems:
My advice is to look at your architecture -- your existing enterprise -- conceptually. Break them down into their component parts in terms of their systems; then break them down again into the core services that those systems provide. Then, evaluate each one of those services on their own, in terms of where they should reside.
To determine whether services should reside on-premise or in the cloud, Linthicum suggested looking at what the services do, how much compute power they require, and what information they consume and produce. He continued on to say that, in some cases, organizations can mix and match cloud and on-premise systems. That approach, he said, will provide for a much more functional, agile enterprise.
According to Linthicum, the requirements and challenges of cloud integration vary depending on the size of the enterprise. While large enterprises have multiple systems and a sizeable budget, small or startup enterprises generally have limited resources and fewer systems.
Big enterprises may take on more complex integration projects, but they may also face difficult, time-sensitive problems when it comes to integration strategy and enabling technology. Things are different for smaller enterprises, Linthicum said:
[Small companies are] typically going to do very simple integration scenarios. I don't think they're going to get into the very complex, event-driven, service-oriented things […] just because of the expense and complexity.
Data integration in the cloud
Security concerns follow closely on the heels of most cloud discussions, and understandably so. With more data on the books than ever before, companies large and small are hesitant to move their confidential information off-premise. On that topic, Linthicum gave advice that some might find surprising:
I think people have a tendency to overestimate the amount of security they actually need. … Be very realistic, because security is going to cost you money. If you're protecting innocuous data that's not necessarily going to hurt you if it gets compromised somehow, then a lower level of security will be fine.
When it comes to storing and managing large amounts of information, Linthicum said that big data systems outshine their counterparts, object databases and XML databases. Big data systems such as Hadoop are a way to manage the complexity of today's explosion of unstructured data, he explained:
They don't require any structure to occur at all. You can have massive amounts of information in a file -- something we wouldn't have called a database years ago -- and have the capability of mapping that into a structure at the time of analysis so [you] can make sense of it.
With cloud computing, the enterprise can now leverage Hadoop because of accessibility to commodity processing, he continued. Today, according to Linthicum, it is poised to be the only answer to the enormous business problems of big data and integrating data in the cloud.
Follow us on Twitter at @SearchSOA.