Of Open Agents and Wheel Reinvention: The BMC Performance Manager Agent

One of the best call to action sessions last week at barcampESM was Erik Dahl’s “can we get an open agent?” talk. The discussion has continued over at the newly revived OMC.

The issue reminds me of what I tell people all the time in conversation on this topic but have somehow neglected to post here: BMC’s Performance Manager agent would be prime picking for open sourcing and attempting to solve this open agent problem. As a disclaimer, I actually worked on that agent on and off, so you can call me biased.

Here’s the case:

• It’s mature code vs. new code that still needs the kinks worked out.
• It’s battled tested to work at scale, at least, when you cluster enough of them together.
• It’s built on a plugin module that allows all sorts of extensibility by dynamically deploying new plugins.
• The agent alone is not the most valuable thing to BMC, the plugins that get deployed to it and the “platform”/server at the center of the BMC Performance Manager is where the real value is. Thus, open sourcing the agent itself wouldn’t harm BMC’s revenue and, indeed, if the venture was successful, open sourcing the agent would increase revenue if more people where writing “plugins” for it and looking for a central server to hook the agent up to.

What’s Good for BMC

Now, on the BMC side, they need something more to do when it comes to open source than just providing CMDB adaptors. Nothing wrong there – breaking the closed source ice up at BMC was and is a big job – but we’re looking for the big bang of BMC getting into open source. The problem is, it’s hard to find something in their portfolio that meets the two criteria of open sourcing previously closed source software:

1. You don’t want to trash your current revenues or make The Shareholder thing you’re letting value out the door.
2. And at the same time, you don’t want to just open source dump and put crap out there.

As mentioned above, while I think the BPM agent is a good piece of code, it’s just part of a larger value proposition. The full BMC Performance Manager platform and the integration with BMC’s BSM stack is what’s valuable. BMC customers are buying the views into their IT environment that the overall IT management suite makes possible, not the agent itself. As I recall, BMC doesn’t really charge for the agent itself, they charge for how many things you monitor and manage with the agent.

And to repeat myself, getting more people to use, care about, and write “plugins” for the BPM agent would only increase that overall value for BMC.

What’s Good for the IT Management Community

For the open source community, the agent would bring a ready made, in production agent along with the protocol and “schema” for telling the agent what to do and how to format the data it gathers about the health of IT assets it monitors and manages.

Operationally, having a code base to start with is a hell of a lot easier than starting from zero and arguing every little thing to build up the 1.0 version. Starting with a blank slate in a multi-party group is usually a bad idea because everyone wants to make their dream code-base, which ends up in a bunch of great ideas put together all funky rather than just a bunch of OK ideas put together well.

Also, having BMC (or one of the other Big 4) behind an effort to provide The Open Agent for the industry would add a tremendous amount of credibility and a ready made install base to start from.

What’s Good for Customers

Finally, for customers it’d bring the expanded base of “plugins” and integrations that we’d assume would come from successful open source IT management agent. But it’d also bring a community managed and owned base at the low-levels. Duplications and proprietary software at the lower levels of the IT management stack don’t help customers. Companies change their priorities with the software, people come and go, and most importantly customers often have different priorities as far as upgrades than vendors: they want to keep using the same thing with only bug fixes.

Now, open source is no solution for the last problem: the tension between developers wanting everyone to run the newest release vs. customers wanting to keep chained to the devil they know are extremely polar. But, in my experience, both sides of the fence tend to work things out slightly better in an open source world

Growing the Pie

So, there’s the proposition: open source the BPM agent. From where I sit, it seems like a pretty big win for all three parties involves – the IT management community, BMC, and BMC’s customers.

As a final analogy, SNMP makes possible a massive amount of value in the IT management world. If SNMP disappeared today, the whole industry would collapse. By SNMP being open in it’s own pre-open source, open source way, it makes all of that value possible. Finally having an open agent at the same level of ubiquity as SNMP would increase that overall value-pie, and it seems like a fast moving BMC could be the agent of change to fatten up that pie, and, to mix dessert-metaphors, eat it too.

Disclaimer: BMC is a client, as is Zenoss

Technorati Tags: bmc, bpm, zenoss