Dedicated server and Co-location data billing models

An important component of all dedicated server and co-location services is the data/bandwidth allocation that comes as part of the arrangement as well as any costs above and beyond what is included.

Before I go into a lot of detail lets take a step back and cover off why this is a necessary component of your service.

When you provision an Internet service there is a requirement to provide network services which will provide an Internet feed to and from the physical server hardware. Provisioning these network services have numerous associated costs a which need to be covered. In order to meet this requirement telecommunication providers typically charge for the amount of data which crosses through their network. Obviously, covering the costs for these services are part and parcel of doing business in this industry.

Part of the challenge we find with providing these services is there really isn't any single way of charging for these services. Every provider has their own preferred billing model which makes comparing apples with apples quite difficult. Further to this, depending on your application and intended usage patterns of your dedicated server there may be billing models that work out much better than others - whilst providing a similar level of service.

In general data billing is done on one of two models:

1. Variable rate data

Under this arrangement you are connected to many high speed internet connections and you should never be speed limited in any way. With Anchor services, you are connected to the internet via 3 x 100Mbps links which are always kept well under full utilisation. This means that you have a guarantee of fast, always available, uncongested internet connectivity. This method of billing is really good if your traffic is likely to have peaks and troughs. Given we have more than sufficient capacity, you will never suffer from slow network speeds. Data accounting is typically done on a user-pays basis where by you pay on a per MB or GB basis dependent on volume.

Under variable data your usage will typically be calculated using one of the following three methods:

a) Inbound only

Under this billing model your usage is calculated based on the volume of inbound data (traffic from the internet to your server). This is the more traditional method of data accounting and is how the majority of telcos supply data. This also seems to work really well for most hosting services as the vast majority of traffic is outbound meaning that you're only paying for the smaller component of data.

Anchor Co-location services are typically billed on an inbound only basis.

By nature of this billing model outbound data is typically free, however, many providers will apply a inbound:outbound ratio which may be maintained. This is done to ensure that the network links don't not get saturated by excess outbound traffic which is not being billed for. It is important to keep this ratio in mind as any outbound usage outside of the agreement ratio will typically result in an adjustment on the actual usage to determine the effective inbound usage to match the ratio.

To give an example, if your agreed inbound to outbound ratio is 1:5 and you do a total of 5GB of outbound data then you will need to pay for a minimum of 1GB (irrespective of if your actual usage is less).

We tend to find that most servers on the Anchor network will typically have an inbound to outbound ratio of around 1:5 - 1:7. Depending on the use of the server. Obviously this can vary quite significantly depending on what your server is being used for, the table below gives some example ratios, based on server use:

  • Serving

    Approximate inbound:outbound ratio

    General web pages

    1:7

    Mail server

    1:2

    Mail server (outbound MX)

    1:9

    VoIP

    1:1

    Streaming Media (Music/video)

    1:15 or greater

    cPanel server (web/mail)

    1:5

Note: These values should only be used as a guide and are by no way a guarantee of usage.

At Anchor we provide this billing model with a inbound to outbound ratio of 1:10 which is quite high. In line with industry standards the more inbound data used the less the per MB cost runs out at.

b) Total data

Under this model data is calculated on the aggregate total of inbound and outbound data.

This seems to have been a billing model which has been adopted largely by the dedicated server market in recent times. From a consumers point of view this billing model is much more simplistic to understand. There is no need to give consideration towards usage ratios and a really good place to start if you are unsure what your traffic patterns are going to be like.

With Anchor dedicated servers we throw in a allocation of 25GB data. We tend to find that this suits the vast majority of our dedicated server customers primarily because few consistently go over this limit. Additionally, if the usage is below this level then consistent usage trends are often an unknown.

c) Outbound data

This billing model has relatively small usage in the industry as has only been included in this document for the sake of completeness.

Under this data model all outbound data (from your server to the Internet) is measured with inbound data free. Outbound data is typically going to be many times larger than the inbound data component.

Whilst Anchor can provide variable rate data on this billing model we generally avoid doing so and as such is not included by default on any services.

Fixed rate data

Under this arrangement a fixed amount of bandwidth, (Eg, 1Mbps) is provided and there is no per MB charge at all. Under this arrangement you can push as much data across the link in either direction. The limiting factor is going to be the amount of bandwidth you chose to purchase. With Anchor bandwidth can be purchased in conjunction with both dedicated and co-location services and is provisioned in 1Mbps increments.

Our article on data transfer speed calculations should give some indicative values of how much data is theoretically possible.

To discuss this further we'll look at a 1Mbps bandwidth. The theoretical maximum at 100% utilisation gives a total data transfer of ~330GB/month. It is important to note, that to get this volume of data across the link in a month it would need to be running at 100% of capacity for the entire month. This is quite unlikely, and for the point of this example, we would expect that real world examples would be more in the vicinity of 160GB total/month, based on 50% utilisation for the entire month.

This data model works really well if you know your data transfer is definitely going to be dispersed evenly over a 24 hour period without significant peaks or troughs in traffic volume.

Examples of services that will find fixed rate data more cost effective include: VoIP services, streaming audio/video services or perhaps gaming servers.

A good rule of thumb is if the application is going to have a fixed number of users consuming a fixed volume of data over a fixed period of time, then fixed rate data is going to be work out more cost effective.

The draw back is of course, if your data is going to have peaks and troughs then you need to make sure that you purchase enough bandwidth to take into account your peaks traffic periods. Failure to provision sufficient bandwidth under this arrangement is likely to result in service degradation seen as packet loss during peak periods.

You may note that the approach is a bit opposite to the variable rate arrangement.

Under variable rate data, you are guaranteed to get fast performance all the time, you get a base data allocation you will continue to get fast network performance and excess data charges will apply. Inversely, under variable rate data you get a set allocation of bandwidth, a set cost each month with the trade off being performance if you exhaust your data allocation.

If you would like to get pricing for fixed rate bandwidth or discuss if this is a suitable option; Please contact Anchor.

Other Billing Models

95% Percentile or burstable billing

This is a billing model which attempts to give you the best of both worlds and is used widely internationally when there are significantly large volumes of concurrent users and available bandwidth.

Essentially, you purchase a fixed rate link, commonly referred to as the committed information rate (CIR), which is an amount of bandwidth that you are guaranteed to get without any additional charges being applied.

However, unlike the fixed rate bandwidth arrangement you can burst above the CIR if additional bandwidth is required. The 95% percentile aspect comes into play with the assumption that you will remain within your CIR for 95% of the billing period. If, however, you are over the CIR for in excess of 5% of the billing period, then you need to purchase a CIR which would take into account the maximum usage for the entire month.

This is a much more complex method of data accounting and can easily lead to confusion.

For this reason 95% percentile/burstable billing is not something that Anchor sells at this point in time.