Lately the hype around “virtual blades” has been picking up. For instance this article here at InfoWorld is just one of the many articles that describes using ‘small disk-less servers without any storage’.
This whole paradigm shift towards using more “embedded” clusters of low servers gets even more interesting when you consider the number of projects out there that are thinking of using low power mobile processors to build ‘clouds’. For instance, this article talks about using Netbook chips to create a “Low-Power Cloud out of Wimpy Nodes”.
The moral of the story is, that as memory and network bandwidth capabilities and cost decrease exponentially, the proliferation of these very wonderful and creative solutions to complex data center problems will continue to emerge. It’s not a surprise that many of these capabilities arise out of the need to address the special challenges of massive virtualization (think VDI), and massive density to reduce power consumption and physical space requirements.
The question that should be in your mind is, with all of these great new paradigms emerging for hardware, how am I going to monitor it all?
The answer definitely doesn’t lie in locking yourself into a vendor that only has an interest in monitoring their own stack effectively (legacy, crusty, old guard vendors).
Clearly from the trends described above, having the ability to place probes across any hardware/virtual/software stack will continue to be important in these situations. Of even great importance is the ability to flexibly create, re-use, swap probes as the technology platform changes and still retain the context of how each of these components affect the performance and availability of the application services that power our enterprise.
The other answer that looks tempting, but that is equally dangerous, is the urge to select a new niche vendor for these new tool sets and stacks. The problem with this is that again you end up with a “stained glassed window” of point tools that force you to do “screen level integration”. As well, many of these solutions will become obsolete as these new server paradigms change rapidly. (Does anyone use a Leostream connection broker for VDI anymore for instance?)
Let’s avoid all of this, especially in a time when efficiency and “tools rationalization” is top of mind.
Any monitoring solution that expects to be able to survive this massive change must have an extensible plug-in architecture to allow for the rapid incorporation of new metrics and allow for those metrics to be used in the context of the monitoring logic (no bolt-ons). Wouldn’t it be a bonus if the plug-in architecture was easy enough for your own staff to use to extend the platform without having to wait for your vendor?
Come join us on a webinar and see what a truly adaptable monitoring solution might be able to do for you.
ShareThis
Subscribe to this post's comments
