By Chris James, Director of EMEA Marketing
There is much discussion about how the software provided by data centre hardware vendors can monitor system and application performance effectively with in-band agents. The short answer is it can’t. Agents or ‘in-hardware’ monitoring add load to the data stream and consume CPU resources – therefore they can’t monitor constantly at line speed as the resulting impact would affect the application performance response times. These solutions can only be used sporadically, roughly every five minutes, providing an averaged result over time rather than the ‘peaks and troughs’ of latency that is actually going on. The only way to monitor consistently at line speed is with an out of band, agent-less solution.
Equally important is the ability to monitor from the VM’s running on the physical servers, right down to the LUNs in the storage arrays, and everything in between from a ‘single pane of glass’. If you don’t have the whole picture then you can’t manage the data centre effectively.
In addition, traditional monitoring tools only work if the network is up. An out of band solution using analytical probes keeps looking and listening even if the network is down, unlike agent based in-band approaches.
But, that doesn’t mean hardware related monitoring tools are useless. They provide a good insight into individual hardware elements. If you have a problem or want to fine tune performance they can be used as part of the solution. They also frequently provide a standard API so that monitoring information can be gathered and integrated with other elements to give a more comprehensive view of the entire data centre performance.
The Virtual Instruments Infrastructure Performance Monitoring platform is a passive, non-performance impacting solution that operates ‘out-of-band’. It generates a copy of all data that is traversing the SAN by utilizing a small portion of the light within the fibre-optic cable and sends it to the Virtual Instruments SAN Performance Probe. Server, switch, network and storage data is then consolidated and correlated to diagnose application performance across the entire data centre, at line-speed, without any further interruption or downtime. This deep insight approach de-risks applications and ensures all elements within the data centre are performing optimally and therefore cost effectively.