Since February, there has been an internal routing problem at a
collect operator. Instead of moving directly to our links dedicated
to the collect, the traffic moved through our peering. The peering being on Paris,
all the traffic moves through Paris even if you had chosen an LNS at
Bordeaux, Lyon or Marseille.
The routing has been rectified and you can now go directly through the chosen
LNS and you are faster on the OVH network.
Update(s):
Date: 2011-12-15 13:43:23 UTC Everything is OK from OVH and sfr's side.
The switch is done.
Date: 2011-12-09 01:01:56 UTC Paris done.
All PPP sessions are now on the new architecture.
Tomorrow we will check with SFR that everything is alright from both sides.
Date: 2011-12-09 01:01:46 UTC Lyon done.
Date: 2011-12-09 01:01:29 UTC Bordeaux done.
Date: 2011-12-09 01:01:12 UTC LNS Marseille done.
Date: 2011-12-09 01:00:57 UTC LNS Roubaix done.
Date: 2011-12-09 01:00:40 UTC The first accounts on roubaix are cut.
The cuts takes only a few seconds.
Date: 2011-12-09 00:59:30 UTC The migration from SFR's side was done.
Now we are going through the concentration BAS located in Paris. The connections are directly established from the DSLAM if it is configured on Level 3 or by BAS as close as possible if the DSLAM are on Level 2.
The old sessions are still connected by the BAS in Paris. For the next reconnection, the new path will be taken.
Tonight we will resetet all PPP connections in order to switch to the new architecture.
Date: 2011-12-09 00:52:06 UTC We modified the network characteristics of the links between OVH and SFR in order to make our tests.
The LNS Paris / Roubaix / Marseille / Bordeaux / Lyon are ready.
Date: 2011-12-09 00:50:31 UTC We will make some tests on the links of tests in order to validate the architecture.
Date: 2011-11-03 12:53:10 UTC There are problems persisting on the routing between the collect operator and OVH. The incident is being resolved between our teams.
Date: 2011-05-31 18:43:41 UTC Some problems persist on the routing between the collecting operator
and OVH. The incident is being solved between our teams.