IT automation and automated network management are becoming essential technologies for increasing the value of IT and ensuring application performance in increasingly complex networked environments. SearchNetworking.com caught up with Rick Sturm and Jim Frey of Enterprise Management Associates Inc. (EMA) to discuss the current state and importance of automation technologies, which will be a major focus at this year's Interop conference in Las Vegas.
Sturm is the CEO of EMA and the chair of Interop's IT Automation track. Frey, research director at EMA, frequently speaks about network management and application performance at industry events. He will present a panel session, "Automation for Identifying and Troubleshooting Performance Problems," on Tuesday, May 19, at Interop.
IT process automation (ITPA) has become a major focus at Interop over the past couple of years. Why?
Rick Sturm: IT automation is the way that you get consistent behavior every time in every situation. And it's that consistency that then drives consistent levels of service and ensures that you're meeting the commitments you've made in service level agreements to the clients.
Automation is the way that IT is able to get to "faster, better, cheaper" in providing its services to its clients. Every vendor on the planet is trying in some way to achieve that.
Automation is not something that anybody has decided to build a marketing campaign around, yet it is the wave of the future. It's the only way that IT can continue to increase its value to the organization. If they fail to continue to increase their value, to become a strategic part of the business, then they're just a commodity and might as well be outsourced.
Jim Frey: I think that automation technologies are fast becoming a requirement. They're becoming necessary in order to also keep up with two key factors:
One is the continued, additive, aggravating factors of more interesting diverse and abstractive technology. By this I mean virtualization in the computing layer, things like Web 2.0, SOA, mashups, movement of virtual machines. The fact is that all the piece parts of the whole application delivery architecture -- or infrastructure, if you will -- are getting more complex, not simpler. So keeping up with the complexity is something that really demands more innovative and automated management tools.
Throw the other big factor in there, which is pressure to keep up with all of that change and yet not expand staff. So more and more, I think the knowledge that can be captured and leveraged through automation management technologies is going to be an essential resource that operations organizations are going to have at the ready.
In the past, automating basic processes was a given in IT, while many of the more complicated functions of IT -- and network management in particular -- couldn't be automated. Has this changed and, if so, how?
Sturm: It has changed and continues to change, but the changes are gradual and not something anybody's building a big marketing campaign around.
But if you look at IBM, for example -- they have this autonomic initiative. And it's a relatively well-kept secret. I don't mean that they're trying to keep it secret, but it's sort of the stealth marketing approach. IBM has poured huge amounts of money into automation. Their objective is to drive down the cost of ownership for their systems, with their tools, so that there are more IT dollars freed up to spend on new initiatives.
Without automation, you run into a brick wall, and you're spending all of your dollars just running in place.
So then the economic downturn is driving more innovations in automation?
Sturm: It drives people to step up and say, "OK, I understand that there may be a risk in doing this, but the potential savings is so great, we can't ignore it." Just as, this morning, I was reading an email about VoIP from some salesman. And I'm sitting here thinking, I know quality is not the same as landline. But maybe the time is soon to be here when we'll need to be looking at it, because it does offer significant savings.
Likewise with automation. People kind of tried to hold back the water a bit and say, "No, no, no -- this is too risky; we can't do it; nobody understands my job." In fact, as you lay off people, the remaining people have a lot more to do and they need the help of automation. It finally wears them down so they aren't quite the luddites that they may have been in the past. Some people in management are risk-averse, and this helps them to get over that hurdle.
In the panel session Automation for Identifying and Troubleshooting Performance Problems, Jim Frey is presenting on "state-of-the-art technologies and practices for automating identification and analysis of application performance issues using data from multiple viewpoints." Are these state-of-the-art technologies futuristic, or are they available today? Are they being adopted in the enterprise?
Sturm: They're real today; they are being adopted.
Frey: There are a number of technologies in this endeavor, and a lot of it has to do with things we've known well for a long time, which are root-cause analysis types of tools. But it's a slightly different challenge to apply those to performance issues where, in the past, most of that technology has been applied to fault isolation from a failure perspective.
The same sort of challenge exists in correlating performance data. So this is part of the open discussion I'm going to have with a couple of technologists about where we are with applying correlation techniques.
Another area is in intelligent analytics and intelligent processing of performance data to watch for early indications of problems. The reason, by the way, that I have a representative from Netcordia on the board is I would also like to introduce into this conversation a slightly earlier piece of the lifecycle, where the potential source of the problems typically is -- it very often has to do with change and change management. So the better job we do of automating not only the process of making changes, but also the process of understanding the impact of changes, the more likely we will be able to get a better grip on handling performance issues in a more timely manner.
So I think that this is something I hope to bring into the conversation that's not always part of the conversation, which is looking at more than just the typical firefighting tools that the NOC engineer would have -- but what else should be considered, and what other automation technologies are relevant to this challenge of optimizing performance?
Which enterprise networking functions are especially good candidates for automation, and how can network managers use ITPA to reduce costs?
Sturm: There's a lot of automation out there today -- something as mundane as auto-discovery. In the past, you had to go out and ping everything to try to figure out what was really attached to your network. Today, it's just a given that almost any network management tool is going to be able to auto-discover your network. Change and configure areas -- not just in the network, but in systems as well -- lend themselves to automation. It's easy to automate setting up a new user. It's easy to automate resetting a port or bouncing a line. You know, not huge things, but when you take them all together, it's a large portion of time spent on maintaining a network.
Which types of vendors are coming out with advanced tools for automation?
Frey: Everybody's adding bits and pieces, and they're all coming at it from different angles. I chose the panelists in particular to try to give a mixed viewpoint. We have a gentleman from Netcordia who's really mostly focused on the network management role, and his specific expertise is in that change management area we talked about. The representative from HP has a broader viewpoint. He actually is network-based as well, but HP of course has a much broader set of solutions that look not just at the network, so that's another complementary viewpoint. The third person I have is an actual front-line practitioner who is trying to use these tools to try to deal with performance across the network.
This audience and this particular panel, it's a network-centric panel, but the topic is application performance because that's what, in the end, is really important.