CDN variations
The term CDN is not well-defined. It covers many solutions and applications to geo-optimize delivery.
Single purpose CDNs
Some CDNs and CDN solutions are single purpose. We often see this in IPTV cases where the entire CDN solution is geared towards a single application and a single operator, where the entire workflow chain is integrated and proprietary.
If the operator has just one business case (the need to move large IPTV volumes on their network), then a single purpose CDN could do the job.
But most telcos have a wider vision on how to use CDNs: host and deliver on-net services like IPTV, host and deliver services for content owners (streaming hosting and OTT hosting and distribution), and network offloading.
To support these various business cases, a single-purpose CDN solution is too limited because it doesn’t support multi-customer, multi-services, multi-technology. Single-purpose CDN solutions are almost always to integrated into the network layer or delivery layer, or even integrated with specific workflows such as portals, encoders and set top boxes. Read this for more details about multi-purpose CDN architectures and how they are a better fit because of the better abstraction between infrastructure, delivery and content workflows.
Content offloading
We see an increase in content owners / distributors asking telcos to deploy a black box (edge) in their network core. Google and Akamai for instance. The proclaimed win-win is that the popular content doesn’t have to move across the internet, carriers and exchanges: better performance, lower costs for both parties. But there are some catches:
1) The content owner / distributor who supplies the box, gets free traffic. All traffic costs are 100% paid for by the telco. The true motivation for the edge box supplier is because they foresee that peering agreements will be re-negotiated by the telcos in the near future.
CDNs and companies like Google (Youtube) generate huge amounts of traffic. Normally, they would have to pay carriers and ISPs to dump their traffic into their networks. But by getting the ISP status themselves, these companies could negotiate free peering contracts. Normally, free peering is agreed based upon the expectation that both parties generate an equal amount of traffic for each others networks. But as we all know, CDNs and large content companies generate a lot of traffic, but hardly move traffic for the ISP. Free peers are abused. By having a box inside an ISP’s network, the volume monsters escape a possible re-negotiation for traffic costs.
2) An edge pop isn’t enough. By placing one or multiple edge boxes in their network core, the telco isn’t really helped. The problem isn’t in the performance or capacity of the exchanges and peers. The problem is in the limitation of the ISP’s metro network. The backbones and metro rings are heavily overbooked and just can’t handle huge amounts of traffic. Actually, the last mile isn’t the problem either. If an ISP really wants to scale and optimize traffic and save costs, they need a CDN with edge servers deeply deployed into the networks arteries.
3) Telcos will never want to lose control over the traffic in their network. By allowing one company to setup an edge box, they lose control. And if one company comes, many will come. ISP’s will never allow third parties to deploy CDN nodes even deeper into their network. The trend we see is that telcos are now deploying CDNs into their own network, and are offering these resources (free or at a fair rate) towards content owners.
The benefit of a multi-purpose on-net CDN is that telcos stay in control of delivery and traffic, can optimize traffic, can offer an efficient shared environment towards multiple content owners. The benefit for the content owner is that they don’t have to deploy boxes: they can just load their content on the telco CDN and enjoy much better performance and capacity (due to deeper distribution).
Caching or streaming CDN
Most internet based CDNs are optimized for web acceleration. And streaming is an add-on. When you start an e-commerce business in the USA, you need a CDN to offer high performance throughout the states and beyond.
But in Europe, most e-commerce initiatives are locally oriented. There is no EU-market. European e-commerce companies tend to work with regional hosting providers to host their online services. They don’t need a CDN.
Except when they use streaming media. Because most hosting providers have no technology, experience and scale to offer professional streaming services. (Yet, we are licensing our CDN technology to an increasing number of hosting providers, they are moving in).
This is why EU CDNs primarily focus on streaming services, and global CDNs primarily make money (if they make money) from web caching services.
This explains why global CDNs can’t really get EU based customers. Their EU presence is to offload traffic into Europe for their US based (web caching) customers. For streaming, their EU presence isn’t strong enough. The EU streaming market is dominated by local and regional CDN’s. They offer better localized services, better customer support, various services to address the local market needs and in many cases, better performance. (Read this and this and this article for background information).
No comments yet.