Vess Bakalov’s excellent commentary on the challenges facing IT infrastructure monitoring products succinctly documents the multitude of concerns that separate many legacy monitoring architectures from those that have a keen eye on the ongoing evolution of large enterprise networks: click here for full article.
At Entuity we’ve long focused on handling the increasing numbers of devices and their sub-components found on typical corporate networks. Simply implementing a traditional distributed polling architecture where all the raw data samples are funneled back to a central database only improves scalability to the next level. To scale beyond there’s a need to distribute more than just the polling … the real time analysis and storage of samples also needs to be distributed. Only then can you really properly contend with the challenges surrounding truly large scale monitoring.
Corporate networks usually comprise a mix of equipment manufactured by multiple different vendors which often limits the scope of software monitoring tools provided by their own vendors. Long gone are the days when just monitoring the state and behavior of ports on devices provides an adequate coverage. As Vess rightly points out the various typical blind spots in many monitored environments need many other aspects of device behavior covered especially those related to environmental and power conditions.
Life used to be much simpler for monitoring tools when a good SNMP/ICMP polling engine was adequate for most data gathering needs but these days the proliferation of APIs without accompanying SNMP agent equivalents means the monitoring system needs to be way more multilingual. Relying on traditional threshold based alert generation also only addresses a subset of problem detection requirements. Threshold settings rely on an appreciation of what constitutes normal behavior and, especially where latencies and response times are being analyzed can vary dramatically in a global or geographically distributed network.
All told, Vess has hit the nail on the head by highlighting these issues. We’ve long shared the same opinions and are ready to help those who have to plan for the future of IT deployments meet their monitoring goals.