When are probe-based congestion management solutions ideal?
February 16, 2011 - Openet
Yesterday’s story about in-network probe usage (and a follow-up on a similar topic today) generated some email responses from readers questioning the use of network probes as the foundation for congestion management. There are a number of PCRF (Policy Charging and Rules Function) vendors that have deployed congestion management solutions that interface with network probes to mitigate radio access network congestion, and a handful of operators today are employing this approach.
Some industry insiders contend, however, that there are major challenges to the probe-based approach, starting with cost, said Gary Rieschick, senior product manager with Openet, which has skin in this game as a maker of policy servers and charging platforms. “Implementing hundreds of probes to a small set of PCRFs can be operationally complex to implement and support,” Rieschick said.
He also warned of signaling congestion: “When you have hundreds of probes sending triggers to the PCRF and the network traffic is bouncing above and below a threshold such as ninety percent, you get excessive triggers or signaling to the PCRF…this is essentially a signaling storm creating by the probe-based solution.”
The key point is that subscriber satisfaction invariably suffers if operators wait to manage congestion after it occurs. “Operators should proactively manage the eighty percent of congestion that can be predicted and proactively enforce stricter policies, and if required, use probes to detect anomalies (the other twenty percent of congestion),” Rieschick said.
Rieschick even questioned whether a business case for using probes to monitor that final 20% is justifiable. “If eighty percent of your congestion is managed in a highly cost efficient manner, is there a business case for the other twenty percent that’s much more expensive? [That’s] something operators have to decide, but [it’s] fodder for debate,” he said.