Data Centre Bandwidth

  • Printer Friendly

A simple explanation of IP transit is that it is a service to deliver IP traffic between your servers and the rest of the Internet. Put plainly, IP transit provides you with an Internet connection.

Typically, a service provider offering IP transit will offer it in two forms, the first being full transit and the second being partial transit. Full transit assures the delivery of IP traffic to and from practically every Internet location, whereas partial transit usually only includes the provider's most inexpensive routes.

The amount you will need depends on how much data a typical user will transfer from or to the server, as well as how many users you expect to have. For example, two hundred people reading news articles would require significantly less capacity than ten people viewing high-definition video.

At such time as you begin comparing service providers offering IP transit, there are a number of important factors you should be considering:

Device reliability

All providers able to provide a reliable service will be able to commit to a level of availability in their service-level agreement with you. The better providers will be able to offer you four-nines availabilty (99.99%, which is less than 1 hour per year) for single-port connections, or five-nines availability (99.999%, which is less than 6 minutes per year) using hot-standby routing protocol with two ports connected to different routers.

Network redundancy

Even using a design featuring a resilient hardware setup, occurrences outside the provider's control could potentially cause an interruption to service. In the example of utilities workers cutting through a cable in the street, a provider not using multiple physical routes out of each location could find itself offline.

Aim to select a provider whose network features multiple diverse paths in to each data centre, and who implements redundancy in their link layer and network layer using protocols which are both quick to converge and to detect link failures, such as the Rapid Spanning Tree Protocol and Open Shortest Path First routing protocol.

Network capacity

A well designed and implemented network will feature sufficient spare capacity to handle current traffic levels as well as short-to-medium-term growth, with provisions in place to expand capacity in the longer term.

Ensure your provider's inter-site connections and connections to upstream providers and peers have been capacity planned to ensure negligible packet loss now and in the future. A provider that is able to offer assurances of utilisation levels at which they have planned to expand their capacity should be chosen to ensure your peace of mind regarding the quality of the delivered service.


Many applications on the Internet are sensitive to latency; that is, the time taken for an IP packet to move from its source to its destination. Applications such as voice, thin-client, and gaming are typical examples of these.

A provider is able to reduce the latency to destinations by entering in to peering agreements with other networks, allowing traffic to be sent between their and the other party's network directly and with the minimum of latency.

Good service providers will work actively to maximise the number of networks with which they are peered. Check that your chosen provider does not rely on a small number of routes out of their network. In addition to the latency consideration, having many routes also allows a greater choice for rerouting in the case of a connection going offline, thereby improving resilience.

Questions? Let us help

To speak to one of our consultants about what our Partner Programme can do for you, or for any queries, please fill out our Enquiry Form, Call 08000 098 028, Request Callback, or Request a Partner Pack.

It couldn't be easier to apply for the C4L Partner Programme
  • Printer Friendly