How to properly scale your company's fibre

Etienne DETRIE
22 Nov 2022

As a B2B telecom operator, we are regularly asked by SMEs, ETIs and very large companies to deploy fibre to their offices.

But before deploying, you need to know what to deploy! Professional fibre with a guaranteed symmetrical speed? Without a doubt. Mutualised fibre technologies do not offer sufficient guarantees in respect to the connectivity challenges facing businesses.

This raises the question: What speeds do you really need to connect all your employees? How do you scale your fibre? 

If throughput is underestimated, bandwidth will be saturated and this will have a negative impact on productivity and employee satisfaction. If the throughput is overestimated, the risk is purely financial, perhaps you are paying too much for something you don't really need. But overall, we cannot stress enough the need to be cautious and to allow for the continued growth of usage.

The key to a successful scaling is to match the expected connection peaks. If you have a daily connection peak between 5pm and 5.15pm, you need to ensure that the speeds are adequate to guarantee the quality of service during this peak.

We will give some examples by considering two approaches: the mathematical approach and the empirical approach, which are not necessarily opposed to each other.

The mathematical approach

Each Internet user consumes bandwidth. The mathematical approach consists of listing all the expected uses in a company, evaluating the bandwidth required for each of these uses, and then making assumptions to determine the activity peaks (simultaneous usage rates) of these uses.

Flow rate required for primary business use :

Usage Mbps
Web browsing 1
Voice calls 0,5
Video calls 3
Audio streaming 0,5
Video streaming 5
4K streaming 24

 

Let's take an office with 100 employees. The very conservative approach would be to say that employees should be allowed to make a video call all at the same time, and thus provide a total throughput of 300 Mbps in the office. 

To be more refined, one could certainly fine-tune the hypothesis. Out of 100 employees, here is a scenario that might be more likely:

  • 30 people outside the office (teleworking, meetings)
  • 30 people using video calls simultaneously
  • 10 people streaming video simultaneously
  • 25 people doing simple internet browsing.
  • 5 people offline at any one time.

The required throughput would then be 165 Mbps, which can be rounded up to 200 Mbps.

The empirical approach

The empirical approach is our preferred approach when recommending a design to a client.

Our customer portfolio includes a very large number of companies of all sizes, from different industries, to whom we provide our managed connectivity solutions. So we have a whole repository of data consumption data that allows us to predict internet capacity requirements.

Example 1 - Office of 500 people ( a digital solutions provider)

  • We provide 2 fibres of 500 Mbps (i.e. a total dimensioning of 1 Gbps).
  • We are seeing traffic peaks of around 340 Mbps. 
  • We are therefore a priori in a situation of over-sizing, but this has been assumed by the client, who wanted the other line to be able to fully cover the employees' traffic in the event of a failure of one fibre line.

Example 2 - Office of 1700 people (Media - Advertising)

  • We provide 10 Gbps fibre
  • It appears that the peak flow consumption rarely exceeds 2 Gbps.

I think that by mixing these two approaches and analysing your current usage and traffic, you will easily find the necessary throughput for your business. Obviously, it would seem that the minimum size required is 1 Mbps per employee.