In an XCHANGE Magazine article on Thursday, Level 3’s Lisa Guillaume mentioned an intriguing new aspect of its CDN business. Here is the quote:
We’re trialing a service called Express Routes to help broadband service providers manage their costs in metro and local networks,” explained Lisa Guillaume, … The routes are custom-built and connect the operator’s CDN nodes directly to equipment such as DSLAMs in access networks, she added.
What does that mean exactly? Well, I’ll give it a shot based on what information we have, maybe Level 3 (LVLT) will correct me if I’m too far off. It seems that they want to build dedicated pipes from their CDN nodes to the far edge of the access networks. In terms of raw assets, Pure CDNs are just strategically placed piles of servers with small leased pipes between them to update the caches and IP transit or peering connections to reach users. The Level 3 concept adds dedicated transport spokes from the centralized node to edge devices, bypassing public internet routes to the edge. Thus they can centralize the servers yet still distribute the data locally without touching the public internet, the latency introduced by greater centralization will be reduced.
Such spokes would benefit both the CDN by bypassing congestion in other networks they have no control over, and also the access network by helping manage that congestion and make their traffic more predictable and easier to manage. Streaming of HD quality movies should have more reliable quality over such a bypass route than otherwise. One could think of these dedicated spokes as analogous to Akamai’s tens of thousands edge servers – let them have space and they reduce the ISP’s IP transit costs. Except these aren’t servers, they are direct pipes to servers farther away. The CDN saves money through centralization, yet mitigates the effects of distance between storage and the edge. Another way an ISP might think of this is as an offer by Level 3 to take over part of their aggregation network.
I don’t pretend to know how the economics might work. It’s not clear who pays, i.e. how much of the cost might be subsidized by the benefit to Level 3 or whether it would be more like metro ethernet or IP transit in structure. But it is a very interesting design change that is probably only available to a network operator like Level 3 because they already have the fiber and are in control of the costs. Of course, there is a threshold problem. They have to convince many ISPs to use the service in order for it to give them a real delivery advantage, and the ISPs aren’t going to care until Level 3’s CDN moves enough data to move the needle. However, it is clear that Level 3 is trying hard to turn their claimed advantage of owning the network into reality.
If you haven't already, please take our Reader Survey! Just 3 questions to help us better understand who is reading Telecom Ramblings so we can serve you better!
Categories: Content Distribution · Internet Backbones · Metro fiber
Join Our LinkedIn Group
lvlt has talked about hooking up with fiber the cable headends and the switching centers of the rbocs for awhile now and it appears this article is outlining one reason they want to do it. i found it interesting she mentioned dslams which makes me think of an rboc and not a cable company. how likely do you think it is that uverse or fios would want to use this service? the rbocs obviously have plenty of metro fiber which is what this “system” appears to replace. thanks for your work.
Level 3 has lots of metro fiber, but not everywhere. If this service were to take off, you might definitely see the RBOC contribute part of the metro fiber to make it happen when they have it . As for whether they would want to be involved at all… we just don’t know enough of the economics.
But off the top of my head, look at it this way, AT&T and VZ are tier-1 IP networks, they don’t pay any IP transit to anyone and they peer with Level 3. So if either were to agree to do this with Level 3, it would surely be like a peering express lane for CDN traffic whose benefit would have to come from a superior aggregation network design somehow – price, performance, or some combination.
the part that is also interesting here is how lvlt could control the traffic on this service from an nfl stadium to the dslam. if its a live broadcast that is just a huge advantage over anybody else, if its a youtube video it wouldnt be as big a deal. i think as we also start seeing more and more tv shows come over the local loop there has to be congestion issues in the metro at some point. i wonder if zayo is involved in this in areas outside of lvlt’s metro footprint
This approach works well when there are few large and concentrated CDN users to be connected by the “custom-built routes”.
Looks like Virgin Media is one of the customers checking this out: http://www.lightreading.com/document.asp?doc_id=168631&site=cdn