The Penn State University Milton S. Hershey Medical Center uses OpenNMS to manage networks at its medical center, medical school, children’s hospital and numerous clinics. They switched to OpenNMS from HP Openview Network Node Manager (NNM) in 2008 and have realized significant gains in network operations while also reducing costs in both time and direct expenses. The following are comments by Dale Meyerhoffer, Senior Network Analyst, on his experiences with OpenNMS.
The Switch from Openview NNM to OpenNMS
Before OpenNMS, the Hershey Medical Center used HP Openview NNM for its network management. By 2008, we had reached its optimal capability but were still missing many things we needed such as graphs, notifications, and other tools to easily alert us to issues. We were frustrated by the lack of these things, as well as several simple reports that we thought should be available.
We made a list of all the things we wanted from an NMS that might replace Openview NNM. First, we looked at commercial software, then we looked at Zenoss and other open source solutions. OpenNMS (v 1.3.9 at the time) seemed to fit most of what we wanted to do. We didn’t yet know how sophisticated the programming was beneath the surface, but the default screens and menus gave us information on the complicated things that we needed, and we were impressed by the simplicity of its design, so we decided to give it a try.
After nearly a year of testing OpenNMS on the side, we reached a decision point when our Openview NNM installation began calling for a $75k upgrade. That potential expense tipped the scales in favor of OpenNMS, and we committed to the conversion. In six hours, we had OpenNMS running with all the functionality that it had taken us three years to develop with Openview NNM. This was a rude but happy awakening.
How do the finances add up? Here’s the comparison we made: $40 to $50k to install Openview NNM plus consulting time covered us for the first three years. To get the next update, we would have to spend $75k (including a new server). In contrast, the OpenNMS software was free, and we added a $5k “getting started” consult to make sure the process went smoothly. The OpenNMS support contract price was comparable to that of Openview NNM. From a cash standpoint, OpenNMS was the clear winner. But more importantly, OpenNMS does a better job of helping us manage our network, and requires less of our time to do it. Openview NNM took up 90% of my time; OpenNMS requires 10-15 hours, combined, from me and one colleague. Lower investment plus higher return is a great solution.
Comparing Openview NNM and OpenNMS – Functionality
With Openview NNM, it was very difficult to produce graphs, and there were limits on how many we could do. To add insult to injury, every time we did an upgrade, the graphics broke and it took us two months to recover. Graphs are easy in OpenNMS.
Metrics with Openview NNM could take 12 or 16 hours to produce. With OpenNMS, it takes only a couple of clicks and two minutes. Clearly, we gained a lot of time there.
With Openview NNM, it was difficult to customize alerts. With OpenNMS, we customize alerts all the time. We save time and headaches by customizing whether an alert goes to email vs. pager, based on severity or any other criterion. That saves us a lot of time and headache. Of course, we also benefit from OpenNMS’s constant improvement and evolution. OpenNMS is adding new functionality all the time, and we just have to tweak it for our own use. In one instance, we had plans to manage investigative events in a particular way, but Jeff Gelbach at OpenNMS convinced us that alarming and notifications would take away our problem and some of our administrative time – a double win.
OpenNMS Platform – Integration and Customization
With the switch to OpenNMS, we’ve reduced operational costs. We’ve started to find ways to save money by running scripts in the background that are tied to the database, which is something we could never do with Openview NNM. We’ve come up with solutions so that OpenNMS now tells not only that a switch is down, but what rooms are down, and what departments are affected. We’re tying this to a department phone directory so that we know who to call before they call us.
We also use OpenNMS to manage UPS’s – a simple but important task. We’re using OpenNMS to see which UPS’s are too hot and won’t last the five years that would otherwise be expected. We’re being proactive.
Working on the OpenNMS platform is just great. Most of the time, we can make changes all within the XML. It’s very easy to copy a section, paste it elsewhere and make the adjustments we need. We could never even think about doing that before.
Support from the OpenNMS Group
Support from the OpenNMS Group has been outstanding and they consistently give us better support than we’re paying for. Sometimes they giving us fast responses to the support tickets we file outside our service hours at 4 in the morning or 9 at night. Other times, they give us quick but very effective consulting advice even though we only pay for support and not consulting. During a recent challenge, Tarus Balog worked with me until 11 at night, saying, “I’m not giving up on you and I’m not getting off the phone until you tell me this is done right.” As it turns out, the problem was caused by an error on my part, not a problem with OpenNMS, but the point is that the OpenNMS team is committed to making sure we’re taken care of.
In related instance, Jeff Gelbach inferred from our logs that we had a problem with the database, not with the network. He walked our team through the solution and the network has been working like a dream since then. The level at which they understand our system is not typical for most companies. They are small enough to be intimate, and intelligent enough to tie things together where other companies would be blind. I have every confidence this will remain true even if the OpenNMS Group grows. That’s just they way they run the business.
What the CTO Should Know about OpenNMS
Who is a good client for OpenNMS? Anybody who has services running that they want running all the time with software that’s 100 percent free. That is phenomenal in and of itself.
Big commercial providers do not give you more than open source! That’s a myth that needs to go away. OpenNMS performance – both the platform and the Group – is the standard that our medical center is beginning to expect from all of our vendors. The OpenNMS platform is powerful, flexible and easy to use. We can do things with OpenNMS that we could never have done with Openview NNM, or could never have done so quickly. Support with OpenNMS is excellent, while support with Openview NNM was bad in the beginning and then got worse, to the point that we were depending on third party vendors who knew Openview NNM. The solution was far from optimal.
The learning curve for OpenNMS is very inexpensive. It will cost you nothing but a little bit of hardware to try it, and the installation process is simple. The key ingredient to OpenNMS’s success is the quality of the thinking that goes into it. The way they’ve designed it is very intelligent, and that’s why you can start to navigate with any application with a few minutes. You won’t find that in many of the big commercial packages that require much more configuration and effort.
Before we fully committed to OpenNMS, I was naturally worried that they didn’t have enough people to provide good support. David Hustace eliminated most of my worries when he asked, “in the year you’ve been running OpenNMS on the side, how many problems have you had?” To my surprise, I realized that the answer was “none.” David pointed out that 80% of support for bad packages is needed because the product isn’t designed well. With OpenNMS, you only need support when you change something, and you can plan for that. Since then, we’ve observed that the OpenNMS Group provides exceptionally fast and thorough support on the rare occasions that we need it. The support comes from engineers who developed the platform and know what they are doing, and that makes the support immediately effective. We are very satisfied, and happy we’re not getting multiple layers of ineffective support that is typical of large commercial vendors.
If you have a large enterprise and you’re dealing with many different types of equipment and things that aren’t the easiest to build around, it’s great to have that support. I can say to OpenNMS, “I’ve got a brand new device that you’ve never seen before – can you help me with the midbuild?” In forty-five minutes, they’ll send something back and say “here you go”.
We really win with the ability to customize OpenNMS. Any large enterprise is going to want custom things that no commercial or open source vendor will give you out of the box. OpenNMS in its current release covers 80% of what we’re hoping for in the forseeable future. We’re going to build that last 20% to do exactly what we want – for example, perhaps tying the network database with another database so that we can make a visual picture of every closet, what’s in it, and where it’s located in our buildings. When we’re done, I’ll present our change to OpenNMS and say “we don’t know if any of your other users need this, but here you go.” That will advance the project, and OpenNMS will keep getting better at no charge to everyone who uses it. That’s the open source advantage. And unlike the commercial source we mentioned earlier, access to the upgrade won’t cost $75k.