Dominique Delisle, programme director for content & services transformation at Orange-France Telecom, on alternative methods of content delivery via mobile networks, the challenges of caching dynamic content, and how cooperative caching might be used in mobile environments.
What are the principal differences between content delivery via mobile networks and wired ones?
With mobile networks, access conditions change continuously and very quickly, while on fixed networks that access conditions do not change so quickly, and generally only due to network congestion.
Due to the design of mobile network architecture and the use of tunnels for supporting mobile data connections, as well as the necessity for a CDN to be efficient in order to serve a sufficient number of end-customers, CDN servers may have to be kept higher in the mobile network infrastructure than with the fixed network.
CDNs are still designed to leverage on DNS and permanent IP address during content delivery, which is another reason to keep CDNs higher in the mobile network infrastructure.
Delay is also a limiting factor in mobile networks, and harms the user’s perceived quality of experience (QoS). Solutions close to so-called Web/application acceleration will be even more important in mobile networks than they are in fixed networks.
In theory, it could be easier to play on different network QoS in order to privilege the delivery of some content, giving an opportunity for network monetisation through CDN for mobile.
However, this approach may be killed off through the introduction of content delivery “optimisation” tools (boosters) currently introduced by a number of ISPs, leaving limited space for tiered services.
Content delivery solutions other than CDNs may be considered in mobile networks to help popular content be delivered: for example, broadcasts (especially on LTE networks) that could be used in conjunction with the large amounts of memory available on smartphones and tablets for delivering news or very popular content, or of course live streaming.
This is still to be evaluated further, not only in economic terms but also in terms of ecosystem (standards for terminals etc).
What challenges are there in dynamic content caching, and how can they be met?
By definition, dynamic content is by itself not cacheable. Some tools aim to limit/identify the actual part of the content which is dynamic from the static part, and process the two transparently.
Another solution might be to let content caching evolve into application caching, allowing the content provider to get some cloudlets to be executed in the CDN servers, in order to manage locally dynamic content.
Potentially, in order to help content provider, this could mean some form of standardisation, so that content providers may design their cloudlets independently of the CDN technology used.
How can cooperative caching be used in a mobile environment to improve network performance?
When asking a network to handle data blindly, it may just apply a one-size-fit-all strategy for all types of content, while constraints raised on content may be different according to the type of content.
One way to get round that is to allow the network to know more about what type of content is to be handled, and the caching server may be a good point where the interaction between the two levels may occur.
For that, it is helpful to use CDNs operated by ISPs (fixed and mobile), while having a cooperative agreement between the content provider and ISP to optimise the caching and thus get the best from the network.