“Video is putting huge strains on the Internet – or at least that is the received wisdom.”
The success of video streaming sites, such as those provided by YouTube, Hulu and the BBC’s iPlayer, are clearly encouraging a new kind of user behaviour, but what kind of strains do these services put on the Internet? And are they any different from those that faced ISPs during the shift from dial-up to broadband, when networks seemed to grow quite efficiently to deal with the increased traffic?
The nature of the loads being placed on networks today is clearly new – the essentially random nature of traditional Web surfing is very different from the extended duration (often 10–20 minutes) of viewing a video stream. This has network dimensioning – and hence cost –implications for ISPs. It can also be argued that local access is no longer the primary bottleneck, instead it is backhaul where the problems currently lie. This may be true in some markets, but not necessarily all. One reason why we see increasing use of ‘traffic shaping’ is to manage backhaul costs, but video streams cannot be downgraded in the same way non-streamed files can.
Many European broadband markets are very competitive, with very low retail prices, and this puts greater cost pressure on ISPs. Local loop unbundlers are better placed than those using wholesale broadband access (WBA) or bitstream to deal with these pressures, especially in markets such as France where multiple backhaul networks are available. However, rural areas will almost certainly be reliant on WBA in the foreseeable future, so cost pressures are not going away. Another indication of how ISPs are planning for future traffic growth is the re-emergence of long-term (ten years or more) fibre contracts for the core network, evidenced by the deals signed by The Carphone Warehouse and Tiscali with Geo in the UK last year. It is not inconceivable that similar deals will be seen in backhaul in the future.
From a commercial perspective, we may see new wholesale pricing models emerging with a structure that is more supportive of the underlying growth in traffic. This could be a model that combines low unit price with high volume, and uses the economies of scale offered by wavelengths over fibre. Or there may be more sophisticated wholesale models with geographical, time-of-day and type-of-traffic components. How this will translate into the retail market is not straightforward, although local loop unbundling and WBA have already pre-conditioned the market to expect different pricing models for different locations.
From a technical perspective, assuring quality of service is one potential solution. It makes most sense when networks are heavily congested, as some backhaul networks are currently. However, there are technical and commercial issues with end-to-end quality of service, complicated by the fact that content is being delivered across multiple networks, from local access through to interconnection (peering and transit).
A shorter-term solution is a device-led strategy that uses the continuing decline in the cost of hard disk space as a (very) local cache. Interestingly, this kind of solution may serve to delay next-generation access (NGA) deployments.
Content delivery networks (CDNs) are another possible solution, where popular content is pushed onto servers located closer to the users. One issue is whether the content can get close enough to the users to address the backhaul problem – this would require co-ordination between ISPs, CDN providers and, potentially, content providers.
The strategy of fixed operators and ISPs without mobile interests should be to drive as much video traffic onto their networks as possible. This will serve to reduce the risk of line loss to mobile, particularly important as mobile broadband prices are falling. For some operators, such as BT, this creates a dilemma over how best to encourage the use of video sites whilst also encouraging the use of its own video-on-demand (VoD) service, BT Vision. There is a complex balance between the wholesale and retail operations, as well as understanding how free services like YouTube impinge on paid-for services like VoD.
For fixed operators and ISPs with mobile interests their strategy should be one that makes the consumption of content as easy as possible on the ‘three screens’. France Telecom’s Chief Executive Officer, Didier Lombard, made an announcement at MIPTV this year about a new triple-play offer, including some exclusive content, distributed via DSL and satellite; this is a good example of an appropriate strategy being put into practice.
Much of today’s video content is being provided quite successfully over DSL networks. Clearly NGA will support video at a greater resolution (ultra-high-definition video is under development in Japan) and multiple streams per home. But more importantly, NGA raises complex questions about wholesale access and backhaul products. As Analysys Mason has shown in other work for Opta and ComReg, next-generation WBA is likely to be more important than it was in first-generation broadband. Now is the time to design these products to enable the explosive growth in traffic to continue.
For more information please contact Matt Yardley, Partner, Analysys Mason. matt.yardley@analysysmason.com
