Brocade and Viavi have produced a joint solution designed to enable mobile operators to capture subscriber data analytics for thousands of users in near real-time to resolve quality of experience issues.
The product partnership was developed at the request of a tier-one mobile operator in the U.S., according to Daniel Williams, principal director for product marketing for data center routing, network visibility and analytics at Brocade. The mobile operator was already using solutions from both Brocade and Viavi in its network, Williams said, but was seeking a better way to ensure it was meeting service-level agreements for high-value enterprise customers. Identifying VIP customers within the network and ensuring good quality of experience for them is no easy task, Williams noted.
“When you think about it, in a mobile network with tens of millions or hundreds of millions of users, the number of VIPs will be a fraction of that – but still on the order of thousands,” he said. “That’s a big challenge for anyone, particularly when you think about the amount of data that goes across these networks.”
Typically, Williams said, operators can opt to capture all customer data, all the time – presenting huge issues for the cost of storage and compute power – and then pull the data from storage for analytics. But that approach can’t be done in real-time, he said. Williams added the mobile network operator customer had explored other real-time subscriber data analytics solutions, but due to network speeds and the large amounts of data involved, other offerings could only identify and capture data from a limited number of subscribers, from a few hundred to as few as a dozen subscribers. The joint solution is said to support up to 6,000 subscribers simultaneously per solution instance, a scale Williams said is industry leading.
Subscriber data capture in about a millisecond
The joint solution leverages Viavi’s XSightTargeted Subscriber Search, which takes in filtered data from Brocade’s Packet Broker and Session Director so only the relevant data is sent to XSight. Williams said a very fast API allows the Packet Broker to essentially be reprogrammed in real time so Viavi’s analytics software can identify particular users once they come onto the network, and data can begin to be captured for analytics in “under a millisecond” to rapidly identify quality of service issues for a defined set of users.
“That’s really important and critical in the mobile network,” Williams said, noting there is significant upfront user plane and control plane traffic for mobile devices. “If you can’t do it in one millisecond … there would be that lag that on the surface doesn’t sound like a lot:a few seconds or maybe a few hundred seconds. But you lose that critical initial information.”
Williams said the solution was specifically developed to help with better analytics visibility into LTE networks for VIP users, although it also captures 3G data. But it could be expanded to Wi-Fi and it could also be scaled up to enable new, premium services for end users, he added. Brocade and Viavi estimate a seven-month payback period for the solution, Williams said, because it cuts down the time for end-to-end customer traces from hours to minutes and can significantly reduce overall capital expense and operating expense by allowing operators to conduct analytics faster and also reduce their overall storage and compute needs through more targeted data collection and analytics.
The solution is a combination of software and hardware, Williams noted, and has not been fully virtualized because it was designed to be deployed in current telecom networks. The mobile network customer for whom the solution was designed wanted to be able to get subscriber traces in five minutes or less. Williams said the solution achieves traces in three minutes and provides a wealth of control and user plane data including physical characteristics of the network where the subscriber device is located; latency, delay and jitter in that part of the network; the number of ISP hops; and application-level data to identify if issues are related to network conditions.
