Network operators are tasked with providing a foundation network that can deliver a variety of applications to their users at any time. For the most part, the network is in the background, humming away while users enjoy the applications. However; once in a while, the application will slow down, or hiccup, and the first suspect is usually the network.
Sometimes this is not without just cause; the network may be composed of various administrative domains in various states of work, and many things that are outside the domain of control of any network operator (bad fiber, rain clouds, bulldozers, floods etc.). There is the common experience of the ping test passing, but still something wrong within the network. Or the other case, where the ping fails intermittently, but there is no clue about the location of the problem.
Offline packet capture analysis tools (check out Embedded Packet Capture!) and probe solutions are either too cumbersome, non-real time or too expensive to deploy everywhere problems could potentially occur. Network operators have simply lacked the tools to be able to properly diagnose application performance issues in a network wide way.
The medianet performance monitor feature is meant to provide greater confidence within the network by its ability to analyze voice, video and data traffic and report on loss, latency and jitter. Performance monitor operates at the protocol level (basically at the RTP and TCP level) by analyzing timestamps and sequence numbers to determine the health of the flows traversing the switch or router. The reports can guide the operator towards the location of the problem (between N2 and N3 in Figure 1), problem ownership identification, and an accelerated resolution. Performance monitor allows for per-application thresholds and actions (syslog and SNMP traps) so, it may even alert the network operator to an issue before the users call!
The analysis reports are available via a MIB, NetFlow, a new embedded tracing tool called mediatrace (more on mediatrace in another post) and of course the CLI. In fact, there are already NetFlow analysis tools making use of the enhanced reporting, and more on the way.
Performance-monitor was released with 15.1(3) T initially on the cisco ISR, but will make its way across the cisco enterprise portfolio in the coming year.
The medianet performance monitor is part of the medianet media monitoring tool chest. The features in media monitoring are designed to accelerate application deployment and create confidence in the network for application and network operators as well as the end users.
More to come!