Now that we have Riba Roja deployment operational, it is time to test the service and tweak it for optimal performance. It is quite impressive to have a GbE connection to the Internet but unfortunately the speed that you get in these circumstances is not limited by your pipe but by the server you connect and the hops in between source and destination.
Let me explain.
We tested the service connecting my laptop to a FTTH port with 10M symmetric service and we tried to connect to several destinations using speedtest servers. I expect this is what users will do at first when they get the service, so it is important to check that everything works fine.
The results are as follows: direct connection to Barcelona and Lleida worked perfectly, 10M symmetric without any problem. Well, to be precise around 12M because we are following ONO policy of offering more than the contracted speed. However, unlike ONO, we have fiber all the way to the end user.
In Spain, no problem. 10M symmetric are achievable. Problems start when trying to connect to servers located in the US or Japan. Connecting to US, which will be quite common, showed 3M symmetric and to Japan 3M down / 2M up. This means that when downloading big files, unless there is an intermediate proxy, the limitation will not be in the access, but in the core network.
This is an effect that will increasingly happen as fiber becomes more popular in the access. I remember that while I was preparing my PhD I asked to a fiber guru what would happen with the aggregation networks when everybody would have 100M symmetric because then, the required bandwidth in the core would be enormous. His answer was quite diffuse.
The reality is that we hope (if we deploy FTTH in more places) to move the bottleneck from the access to the core network and this means that big telcos should start thinking about upgrading their core infrastructures.
Maybe this is why they do not want to go FTTH. If they do, they would not only have to invest in access but also in the core.