Dan Sahar, co-founder of US video delivery specialist Qwilt, on deploying transparent caching solutions, the advantages of a hierarchical topology, and the marginalisation of P2P caching.
What are the main requirements for deploying transparent caching systems with large operators?
It is a well agreed upon paradigm that the best solution for video is to serve the video bits over the shortest distance from the end user – both QoE and economically wise.
For the larger operators this typically means deploying transparent caching solutions at the subscriber edge in a distributed fashion, going beyond the traditional legacy caching solutions that were used offset only transit costs, often times a small part of the operator costs structure.
Deploying at the subscriber edge places stricter requirements on any solution – while some legacy solutions were more geared towards data centre environments where space and reliability are of lesser importance.
A subscriber edge deployment calls for high efficiency on all aspects of the solution, e.g. small footprint/high performance, five-nines reliability and dead-simple insertion and management principles.
As with any large scale network solution, cost is always one of the fundamental criteria. When comparing the costs of different caching solutions, network operators need to take into account all of the extra cost features needed to effectively cache and deliver a high volume of video content in their unique network environments.
Caching products that may look like the most cost-effective solutions based on list price for the software or appliance package, but will likely end up costing much more when all of the required components, necessary licenses, and add-on modules or services are acquired.
Carriers should carefully identify and quantify any hidden deployment and network reconfiguration costs and possible upgrade requirements to existing infrastructure.
What are the advantages of hierarchical topology (core and edge) when deploying in complex networks?
Video caching is a game of statistics: the more subscribers a cache sees, the greater the efficiencies that can be achieved. Thus a a hierarchical solution may make sense in cases where transit prices are high and there is a need to offset them with centralised caching nodes at the core.
They can then be coupled with edge nodes that can optimise the bulk of a typical operator capital spending which lies in the backbone infrastructure.
How can such a deployment achieve high capacity and scalability?
In order to achieve high capacity, the approach that we took at Qwilt is to create a highly optimised software architecture for the task – instead of bolting together disparate generic pieces, high capacity and scalability can be achieved by combining all required functionality into a single, integrated platform that requires a fraction of rack space and power consumption of competing products.
This all-in-one software approach can deliver 5x the performance per rack unit of first generation transparent caching solutions, resulting in a smaller footprint, higher scalability and reliability, and lower total cost of ownership (TCO) than any alternative solution.
How can it benefit from P2P support?
The feedback that we at Qwilt have been receiving from multiple operator engagements worldwide is that for them, video is a critical service, one where operators can have a positive effect on the user experience and THE place to invest is the best solution for video.
P2P caching is very different from video caching – it’s non time sensitive, does not impact consumer experience – is not business critical. P2P caching is relevant at best in the short term only whereas many operators in particular in North America and Europe is not a functionality they are willing to consider due to legality issues.
As true video usage increases, P2P traffic will become a very small percentage of the traffic and will actually become a real distraction from a business perspective for the current short term market, and a real pain to have to continue to support the legacy customers in the long term.