Q&A: Roland Mestric, Nokia
What new service opportunities does cloud technology offer to service providers?
We all know the cloud offers cost reduction and faster time-to-market, but it also enables service providers to offer a wealth of new video services to their subscribers. For example cloud DVR allows them to deliver time-shifted TV services on any device. The cloud breaks the limitations of the home-based DVR in terms of tuners and capacity, since end users can now record as many programs as they want, even simultaneously if they like.
Does this translate into additional revenue for service providers?
Potentially, yes. These new cloud-based services can be monetised. For example, the extra DVR capacity above one terabyte can bring in a few euros from the end user.
The shift to on-demand video services, and unicast streaming in particular, also opens new revenue opportunities. With unicast the service provider can not only personalise the content delivered to each user but also can personalise ads delivered as well. The service providers can make ads more relevant to their subscribers, for example, ensuring they don’t get Christmas promotions during summertime if they are watching a content they have recorded six months earlier.
This shift to unicast you mention, isn’t it creating challenges for service providers?
Definitely. The shift to unicast translates into more traffic. Fortunately we have some solutions to limit its impact, by caching content closer to the consumers. Content delivery networks (CDN) are widely used now in combination with cloud DVR.
We know video traffic can go up dramatically at certain times of the day if it’s unicast. How can service providers handle this?
We’ve all heard about failures of OTT services to deliver popular live sport events. One solution is to dimension the network to handle those traffic peaks. This is very costly and is a waste of capacity since 99% of the time the capacity needed is hundred times below the peak capacity requirements.
Another solution is to leverage the cloud technology by deploying an elastic CDN on virtual machines, which allows to dynamically spin up and down the capacity as needed.
Do you see other challenges remaining, for cloud DVR in particular?
Scalability is the biggest challenge. Don’t expect all the subscribers to intensively use the new services from day one, but ultimately the usage can be massive. This is what happens: some subscribers will try the new services. Once they see the value and get used to it, they use it more. They spread the word and new subscribers get on board. The ramp up can take time but it is inevitable and steady. In the end, it means that the system needs to scale in a cost-effective way, from a few hundred terabytes of storage capacity to tens of petabytes.
The failure rate of the hard disk is known to be an issue with home-based DVR. Does cloud DVR solve this?
Many of us have had the annoying experience of losing our recordings due to hard disk failure. Availability of the content is a key promise of cloud DVR. You can easily duplicate the content into the cloud. The challenge is to do it in an economical way, with performance: consumers don’t want to wait a day or two before their content is recovered after a failure in the server.
Is content rights still slowing the adoption of cloud DVR technology?
Looking at the economical equation only, the private copy model has always been a blocking factor for cloud DVR. However, things are moving and many service providers today have obtained the rights with some content providers to deploy a shared copy model, which improves the business model.
But you need to balance this against the benefits of cloud DVR for the consumer, the service provider and the content provider. Today we have reached a point where service providers see the payback of deploying cloud DVR despite the remaining content rights issues.
How do service providers deploy cloud DVR and is there an impact on its characteristics?
A cloud DVR platform has to adapt to the service provider needs, now and in the future. We are currently deploying cloud DVR with a mix of private and shared copies, but the platform has the ability to migrate into a fully-shared copy model in the future.
We also have customers who start with a subset of services, for example live TV with pause and restart, and are planning to introduce catch-up and network PVR services later.
The end user service directly impacts the system characteristics: live TV will require a high throughput while video on demand will require more storage. We can dimension these two characteristics independently from each other and evolve them smoothly over time as the demand evolves.