Feature Request: Network Health Score Indicator

[Ronit0104123]

I was thinking that we could add a Network Health Indicator(of the user) for the homepage. This would provide users with a colour coded status (Green, Yellow, Red) based on their IP’s network stability. The score would be calculated using latency, packet loss, BGP stability, and outages, fetched from RIPE Atlas API.

How it works:

Fetch network data for the user's IP.
Assign a health score based on latency, packet loss, etc.
Display a status bar adjacent to the "Your Connection" section.
This would make IHR more user-friendly by giving instant insights into network health without requiring users to analyse raw data.

Let me know your thoughts!

[romain-fontugne]

Hi,

This is a good idea, the main difficulty is really how to calculate such indicator and make sure it makes sense.

[Ronit0104123]

We could look at the most important parameters that effect the network, like latency, packet loss, BGP flaps(for routing instability),jitter, outage.

Network Health Score=100−(W×Latency)−(W×PacketLoss×10)−(W×BGPFlaps×10)−(W×Outages×10)−(W×Jitter),
where W will be the weight we give for each parameter( which will be different for each), as some parameters impact the network more, like packet loss impacts the network the most, so we could give it the highest weight.

Then after we get the score, for different range of scores we can represent it with either red, yellow or green.

Looking forward to feedback on parameter selection and weight distribution.

[dpgiakatos]

Hi,

What is the time frame for each parameter?

[Ronit0104123]

I was thinking that it would be better if we find the parameters for ASN to which the user is connected, also what parameters for the network health score should i look for to get from IYP.

[dpgiakatos]

Hello, I apologize if my previous question was unclear. I believe the time frame for the proposed parameters is somehow important. For example, the latency parameter can have different values depending on the measurement time frame. A short time frame may capture temporary spikes, while a longer one can provide a more accurate average.

Also, IYP is updated weekly (unless this has changed), so we cannot get real time measurements.

With that in mind, how do you plan to find this time frame? Do you have references indicating that this score is being used in measurements?

As @romain-fontugne mentioned, "the main difficulty is really how to calculate such an indicator and ensure it makes sense." So, I believe we need to validate that the proposed metric makes sense across various scenarios before implement it to the website.

[Ronit0104123]

I think the time frame for the Network Health Score should depend on the specific parameters being measured. So i was thinking it would be better to use different time frames for different parameters.

At first i was thinking of having a weekly score, to get more accurate data, but that might feel too static to the user, so i suggest these time frames.

Time Frames for Parameters:
Latency and Jitter - Hourly / Daily ( we could look for the median daily latency, and the daily average jitter )
Packet Loss - Daily (to avoid capturing short-term spikes that might not be meaningful)
BGP Flaps - Weekly (since BGP updates tend to change less frequently)

For data extraction, RIPE api can be used to fetch data for specific time frame for each parameter being considered.

Would this approach align with your expectations for the network health score? Let me know if you have any suggestions.

[dpgiakatos]

I understand your logic, but I would like to ask again:

Do you have references (research papers) indicating that this score is being used in Internet measurements?

If you have references (research papers), please attach them here. Otherwise, if this metric is created by you, you will need to validate that it makes sense through various scenarios and use cases.