Modern content delivery networks (CDNs) are multi-purpose solutions that can help you address some of the most prominent IT challenges. But even as they evolve, CDNs retain their core functionality—improving connectivity to provide a faster and more reliable user experience.
In prior chapters we talked about CDNs’ caching and front-end optimization (FEO) mechanism, and their acceleration benefits. In this chapter we discuss how CDNs boost network level performance by:
Anycast routing enables CDNs to bring your content closer to your users. It’s best explained by way of comparison with unicast, its more basic alternative.
Things are simple using unicast—each network node has a unique address; all requests to one are forwarded to its address. But with anycast, a single address is advertised by multiple nodes, making each a valid candidate to receive any one request.
With multiple connection options, the request is expected to be routed to its nearest recipient. The premise being that the shortest path will be identified and prioritized by devices that govern the flow of traffic within the network.
CDNs are massive networks performing anycast on a worldwide scale. By advertising identical IP ranges in numerous points of presence (PoPs) located globally, a CDN provides local Internet service providers (ISPs) with multiple access points from which they can choose the one having the shortest route. The shorter the route, the better the connection time for website visitors.
Using anycast also has the side effect of improving availability by providing multiple backup choices for any endpoint. If one data center goes offline, it’s simply skipped in favor of the next best option.
In computer networking, hop count measures the distance between two hosts, reflecting the number of times a request changes hands between points A and B.
For example, a London-based user request to a New York-based server may be routed through a local ISP that forwards it to another ISP on the US east coast. Here the hop count is two.
In an anycasted network, keeping track of hop counts allows routing devices to identify nearby nodes. In combination with other optimization techniques, this helps determine optimal connection paths within that network.
A regional anycast is an improvement on the basic model. It’s best suited for worldwide networks, including global CDNs.
With regional anycast, a network is divided into virtual clusters; each corresponds to a specific geographic area. Identical IP ranges are advertised only on nodes within the region, not on the rest of the network.
Cluster topology offers control over routing choices of local ISPs. By limiting the number of options, local ISPs must route traffic to a nearby node, even if compelled to do otherwise by suboptimal configurations or spur of the moment decisions.
Today, most commercial CDNs rely on regional anycast to further accelerate their content delivery.
Internet service providers are classified into three tiers based on their connection and payment agreements.
Tier 1 providers are a select group of carriers representing the Internet backbone. A typical tier 1 ISP operates its own networks and doesn’t pay for bandwidth usage. It also maintains interconnected (peering) relationships with other tier 1 ISPs.
Second and third tier providers are smaller carriers who purchase IP transit from a tier 1 ISP and resell it to their users. A typical tier 2 provider is a local carrier with some peering arrangements. Tier 3 providers constitute even smaller ISPs, or dedicated networks (e.g., one belonging to a university).
Of the three groups, tier 1 providers offer the best network performance by far. They have more presence, better peering arrangements, and direct control over traffic flow.
In addition, many tier 1 networks have global coverage. This allows effective end-to-end routing of both transnational and intercontinental traffic.
Commercial CDNs use their funds and bargaining power to purchase transit directly from tier 1 providers. As a CDN subscriber, your website visitors benefit from that arrangement. They reach your website directly via the Internet backbone, with minimal hops and very low risk of packet loss.
Much focus is given to CDN caching and FEO features, but it’s direct tier 1 network access that often provides the largest performance gains. It can revolutionize your website page load speeds and response times, especially if you’re catering to a global audience.
Peering is a process in which two or more networks exchange traffic, usually to reduce transit costs while improving stability and performance for users. With peering, visitors can move directly from network A to network B with minimal delays.
As a rule, intercontinental networks, as well as networks located in hard-to-reach regions, are considered the most lucrative peering partners. They’re able to broker the best peering arrangements and further extend their coverage.
Almost exclusively, tier 1 providers peer with other tier 1 providers for two practical reasons:
“The Tier 1 Club” is the result. Here top-tier ISPs engage in donut peering—exchanging transit solely between themselves for the benefit of their direct users.
When using a CDN service that purchases tier 1-provided transit, you also benefit from the peering arrangements forged between these top-tier carriers.