Cote’s post on People over Process here , some of the calls to action on the new Open Management Consortium blog and my Q&A with peers in the industry over the past year said it’s time to see what other people think on this topic.
I suffered through a painful server agent forklift project that lasted almost two years (and appears to have finally failed and be in the midst of a vendor/legal contract fight) and I’ve always had “pent up” feelings (I apologize for venting…) on the whole server agent and server monitoring space. I had to deal with no less than four different competing tools – two from vendors and two open source. The mission from the top was one standard platform for server monitoring, the reality in the systems admin silos was they wanted their tools and didnt’ want to have anything to do with what was perceived as the “NOC’s monitoring tool”. List price for some of these base monitoring agents would end up costing more than some of the servers that we used in that environment. None of those vendors had a pricing model for addressing that issue or multi-core CPU’s either so the costs in the long run were certainly uknown for basic server hardware and operating system monitoring. My sense was that something had to give in the future in what customer’s were going to be willing to pay for.
I’ve been thinking (hoping) now for some time that in the near future base server HW/OS monitoring would be as easy as SNMP monitoring on the network side of the house. It’d become a commoditized component in the overall NMS/ESM architecture. I’ve heard from server manufacturers about embedding these functions in mainboards, add in cards or ASICs. Every release of an operating system builds on the management and monitoring capabilities of the previous version. Open source software solutions for monitoring foundational server and operating system components continue to mature. The KVM vendors are sure to invent monitoring capabilities for their product sets that can monitor this stuff via out-of-band connections to the server. Microsoft’s WMI continues to improve. WS* standards are developing to allow new ways to monitor servers. Heck, I bet Cisco or some of the network layer guys could event get into this somehow by watching the wire or something via a suped up switch connection in the future.
Recent vendor acquisition activities over the past year or so have these vendors making decisions on how to address product line overlap. Some vendors have already changed from agent based solutions in favor of agentless based approaches. I can think of at least two vendor situations where multiple server monitoring agent products exist post acquisition. I think there’s a great opportunity for vendors to think about basic server hardware and operating system monitoring now and what it’ll look like the next few years. Cote’s recommendation about teaming with this OMC group could be a perfect fit here for the future contribution of product and technology that may be passed over at some point in the future. Should this OMC group align with OASIS, DMTF or other standards body and work towards a free or very low cost standard base agent for the common platforms? Something that supports an extensible architecture where vendors could plug in their more value added agents for applications, etc.? This could help with the “multiple agents on a box” problem as well.
I don’t know all the economics on the product side by any means. I don’t know anything about revenue streams, pipeline, etc. There are certainly architectural and customer deployment challenges that would have to be addressed, etc. I do think that it’d be nice to see the monitoring of a server’s hardware and operating system get simpler and get standardized. I think the place the vendor of the future focuses on is in the monitoring of applications (ARM), services, transactions, SOA/bus publishers, consumers, batch/cron jobs, etc. on the server. They would be focusing on business services and applications, IT processes (ITSM/ITIL and other IT best practices) coming to life via these new rich agents.
Does anyone else share similar thoughts? Can base server HW/OS monitoring be as easy as monitoring the network? Will it be a commodity in the near future? What role will the incumbents, open source software, server hardware, operting system and KVM vendors play?
I’ve sat on actually posting this for some time. I’ve seen more and more of our roadmap plans this week around server and application monitoring. We’re doing the right thing for our clients. Our focus on our composite application management (CAM) products will be stellar and really enable both the management and the monitoring necessary for services of the future. We’ll make sure our customers get the best of both worlds in terms of server monitoring as their environments and needs demand. I hope to see IBM Tivoli take a leadership position in this area of open source base server hardware and operating system monitoring and management sometime in the future just as we do in so many other standards development areas.