0

My small office has a 30Mbit/s internet connection, on which two or three people work. We could get twice the bandwidth at a substantial cost.

I imagine a bandwidth upgrade would slightly improve latency for all internet-related tasks and thus provide a small increase in productivity.

I'd like to estimate the productivity gain that could be achieved by this upgrade, based on per-second traffic data. Are there any rules-of-thumb that can be used for such a calculation?

For example, one could count how often the bandwidth maxes out. If the bandwidth maxes out for 10 minutes every day, then doubling the bandwidth may reduce waiting times by 5 minutes (in the best case). Obviously the relationship between latency and bandwidth is more complex than that. Is there any way to do this estimation more accurately?

New contributor
LTR is a new contributor to this site. Take care in asking for clarification, commenting, and answering. Check out our Code of Conduct.
  • 1
    There are tools for this. NetFlow (IPFIX) is built into most enterprise-grade devices, and there are many different applications to analyze the data collected. We use that to keep an eye on sites and order bandwidth upgrades as necessary. – Ron Maupin Feb 11 at 15:13
2

I've learned a rule of thumb by a Cisco CCIE and he said, that if a link is utilized more than 50% on average over a period of a workday, you should consider upgrading the connection.

Here's an official way of calculating Link Utilization and Efficiency:

Link utilization is simply the average traffic over a particular link expressed as a percentage of the total link capacity. Link efficiency is a less commonly used term that is defined as the ratio of the time taken to transmit a frame (or frames) of data to the total time it takes to transmit and acknowledge the frame or frames:

Utilization (U) or Efficiency=(Time taken to transmit frame)/(Total transmission time)

The efficiency of a link with stop-and-wait ARQ can be determined as follows: If the time taken to transmit a frame or block of data is tf, the propagation delay for both frame and acknowledgement is td, the time taken to transmit an acknowledgement is ta and the total processing time is tp, then:

U = (tf) / (tf + ta + tp + 2td)

In many situations the acknowledgement transmission time and processing times can be ignored, giving:

U = (tf) / (tf + 2td) = (1) / (1 + 2a) ; where a = td / tf

Example:

A point-to-point satellite transmission link connecting two computers uses a stop-and waits ARQ strategy and has the following characteristics:

Data transmission rate = 64 kbps
Frame size, n = 2048 bytes
Information bytes per frame, k = 2043 bytes
Propagation delay, td = 180 ms
Acknowledgement size, ta = 10 bytes
Round-trip processing delay, tp = 50 ms

Determine the throughput and link efficiency.

Solution:

Frame transmission time tf = (2048 x 8) / (64000) = 0.256 sec.
Acknowledgement transmission time ta = (10 x 8) / (64000) = 1.25 msec.
Total time to transmit frame and receive an acknowledgement is:
tf + ta + tp + 2td = 0.256 + 0.00125 + 0.05 + 0.36 = 0.66725 sec.
Throughput k = (2043 x 8) / (0.66725) = 24.494 Kbps.

Note that the resulting throughput is considerably less than the transmission rate of 64 kbps.

The link efficiency (Utilization) can now be calculated, neglecting ta and tp, as follows:

a = (td) / (tf) = (0.18) / (0.256) = 0.7
U = (1) / (1 + 2a) = (1) / (1 + 1.4) = 41.67 %.

Source: https://uomustansiriyah.edu.iq/media/lectures/5/5_2017_03_18!09_07_27_PM.pdf

  • An internet connection is not usually stop-and-wait ARQ, so this calculation is not useful here. I appreciate the "50%" rule of thumb! – LTR Feb 11 at 11:40

Your Answer

LTR is a new contributor. Be nice, and check out our Code of Conduct.

By clicking "Post Your Answer", you acknowledge that you have read our updated terms of service, privacy policy and cookie policy, and that your continued use of the website is subject to these policies.

Not the answer you're looking for? Browse other questions tagged or ask your own question.