Intelligent Enterprise

Everyone loves SaaS, or so it seems. However, the increased use of SaaS-delivered applications creates headaches for IT operations folks. What's core to the problem is the fact that enterprises don't directly control their SaaS infrastructure, and thus, things such as outages and performance issues are largely out of their control. Indeed, the best they can do is to send e-mail messages or make phone calls. There is no server to kick, no host to reboot.

The dilemma is that while IT operations wants to continue to control all applications, including SaaS, there is little they can actually do to resolve issues. Or, is that completely true? There are ways that IT operations can manage SaaS, and do so proactively.

The best way to manage SaaS is to deal with the resource as any other remote application. The difference is in degrees for granularity. While you can manage traditional internal applications at the network, database, application, and interface levels, with SaaS you're limited to the network and the applications, typically only understanding whether it's working, or not (outages).

SaaS providers understand this issue and do open up points of management where IT operations can check on the status of their SaaS-delivered IT resources. You can monitor availability, performance and even the provider's ability to live up to Service Level Agreements (SLAs). It's just a matter of pointing your management infrastructure at the SaaS provider, and thus having a holistic look at all critical systems, internal and external. It's also a matter of frustration that the modern data center where you could see rows of very cool looking boxes, are now becoming more virtual in nature. If you love to hug your server, then you may not like the future of IT.

There is an ongoing passive-aggressive battle between existing IT operations and the emerging notion of SaaS. It's really a matter of responsibility and control, versus economics and utility. With a bit of work, both existing IT operations and SaaS-delivered applications can coexist nicely. I recommend that you do the following:

• Create a management strategy that includes SaaS. Consider the points of management that the SaaS providers are offering now, and will offer in the future.

• Test the operations through drills. What do you do during an outage, performance issues, or other issues that arise in the real world?

• Those who lead IT operations should make sure that everyone understands the strategy and the value of SaaS. Some in IT operations are still not fans of SaaS.

David S. Linthicum is a managing partner with Zapthink, a consulting and advisory organization dedicated to SOA planning, implementation, training, mentoring and strategy. He is a well-known application integration and SOA expert who has authored 10 books on related topics. Write him at david@linthicumgroup.com