Gradwell have installed Firebricks in the core DSL network to monitor all ADSL and FTTC circuits 24/7. They allow us to build up a historic picture of the line so that when an issue is reported our support team can cross-reference the fault report with what we can see on a graphical representation of the circuit.
There is a huge advantage to this over standard broadband diagnostics. Diagnostics that are run in response to a customer fault report take a snapshot of the circuit at that moment in time. Occasionally regular diagnostics can then be ran to build up a picture of the line, but this is very reactive and doesn't take into account issues that were encountered prior to the fault report being submitted.
Instead Gradwell support are able to start diagnostic testing and view the history of the circuit to aid the resolution of the fault.
Every circuit is measured every second for:
- Online/offline status
- Packet loss
- Sync speed
- Download usage
- Upload usage
Although usage is shown, Gradwell teams are unable to view where that traffic is going or has come from, so rest assured that no personal browsing data is exposed. The circuit stats are generated from an LCP echo that is sent to the router and has no effect on general browsing speed or VoIP calls (the size of the packet is around 53 bytes). The LCP echo also has no effect on any download allowance that you have on your broadband product.
An example graph is below. There are several scales that make up a graph. First of all, time of day is shown along the bottom. There is also a scale for speed and latency time on the y axis. To avoid the graph being too tall the speeds are plotted using a logarithmic scale, which needs to be taken into account when reading the graph.
- The black line is the sync speed of the circuit. From the sync speed this appears to be an FTTC circuit
- The green and red dots are traffic and show general use of the circuit. Green is download, red is upload.
- Latency is shown in blue, this is a pretty low latency line
- Red is packet loss and comes from the top of the graph downward - the longer the line the longer the packet loss appeared on the circuit
- Purple is offline time and works to the same rule as packet loss. The longer and thicker the purple line, the longer the circuit was offline
Here's how our support team might use this graph. If a customer was reporting issues we could view this historic view of the line and see that the customer has been using it normally. However, the circuit has started to go offline for small periods, culminating in a large number of disconnects in the morning. This line would start to be unstable and unusable for the customer, which probably prompted the support call.
If you look really closely you'll see that the black sync speed line also drops at around 11:50am. This is probably due to the large number of disconnects, and the equipment at the exchange has recognised that the line is frequently needing to resync, so has attempted to drop the sync speed of the line to see if it becomes more stable.
Using this data, we could start to see if there are any trends to the disconnections. For example, maybe they always happen during rainfall, and there is water ingress in the network. This would aid our fault report to Openreach.