The Performance Zone

Flow to Packet: Network Performance Management from A Service Provider's Perspective

With increasingly converged and complex networked services, service providers are under a lot of pressure to deliver quality service and customer assurance. Key service enablers are playing ever more crucial roles and need to provide quick response for a good customer experience with the service. AAA servers, need to authenticate cell phone users each time the phone is turned on,  DHCP servers need to dynamically assign IP addresses, DNS servers need to perform name to IP address translation. Any one of these service enablers being a weak link will negatively impact the service they are meant to serve.

Since there is no room for guesswork here, service providers place instrumentation in front of key service enablers such AAA , DNS, and DHCP servers as well as the application servers hosting the  service (such as WAP - mobile WEB service on a cell phone). The instrumentation provides real-time end-to-end visibility into services delivered, the customers using them, and the network infrastructure delivering them.

With insight into critical KPIs and proactive automated problem detection and diagnostics, service providers can significantly reduce MTTR. Without such assurance, services could go south in a hurry. In other words, deeper service visibility delivers higher service quality control. This ability to see the whole picture on the flow-level and at the same time dive deep into targeted packet analysis whenever needed allows service providers to assure services and increase customer satisfaction.

"Flow to packet" vs "Find the needle in the haystack"

In an earlier entry, I discussed the importance of using a top-down approach to network and application performance management in order to reduce MTTR. I promised to pick up this theme again and here I am. On one hand, you want to have visibility into the flows instead of looking at all packets if you want to reduce MTTR. On the other hand, you need the capability of diving deeper into the packet data whenever necessary to pinpoint and fix performance problems. In other words, you need a solution that allows you to combine comprehensive real-time and historical monitoring with continuous packet recording and detailed packet level data mining. I will share with you a few customer stories in some key vertical markets in upcoming posts.