[oles@ovh.net]

Hello,

As stated earlier, we are finalizing the
work on the network. This week we finished 90%
of the heavy work. Now all routing in Roubaix
is centralized in 2 large ASR9010 routers with a
total capacity of 1.2 Tbps. Between Roubaix and
"the rest of the world" we have 620 Gbps capacity.

This week, we exceeded 380 Gbps outbound
Internet traffic generated by the 90,000 servers that OVH hosts.
Something new: we have about 50 Gbps of internal traffic
between your servers and it increases every month. This shows
that our customers build the infrastructure and
clusters based on multiple servers in different
data centers and all these servers communicate with each other.

All these works on the network has also enabled to
balance better the links between Roubaix and London/Amsterdam
/Frankfurt/Paris. We've added a router in
Amsterdam and one in Frankfurt. We have 2x(4x10G) on Amsix
and 2x(4x10G) on Decix. We increased the capacity
of Sfinx in Paris to 2x10G. The connection to Milano
Warsaw and Madrid is more balanced but we must also
do some work still. Bottom line: we
gained approximately 100Gbps capacity between Roubaix and
"the rest of the world" just by centralizing routing.

The network:
http://weathermap.ovh.net/backbone

Now we still have some refining but also
capacity increases.

Apart from Roubaix, in Europe we still have to:
- Increase capacity through peering with Orange
though the regional peering in Lille, Lyon and Bordeaux.
We have 2x(2x10G) in Paris and talking about 3x(1x10G)
and more.
- In London, we need to transfer the main POP
from Global Switch to Telehouse. We'll add
a 2nd router and increase capacity to LINX
and the USA.
- Warsaw capacity increase of PLIX to 4x10 Gbps
- Increase the capacity between Amsterdam/Warsaw/
Frankfurt/Warsaw to go to 2x(4x10G)
- Milano peering capacity increase with
the incumbent operator in Italy (Interbusiness) to
2x10G see 3x10g
- Increase capacity between Frankfurt/Milano
2x(2x10G)

In the U.S., we are beginning to sign the POPs with transits
and peerings, but here we go again easily.
We will expand to Philadelphia and we
started recruitment with the training of the staff.

In Roubaix it remains to establish the internal network
routers for new services in beta, which should be
out soon. The famous privateCloud. As VPS,
the Exchange and the publiCloud (ex miniCloud/coreCloud)
based on the pCC, the routing of pCC is particularly
important: it guarantees 100% availability.

To better differentiate service, we will also
establish a separate network for "kimsufi" servers
and also innovate on the "ovh" servers. The goal is to
ensure new things ... that our customers
begin to claim on a regular basis. Basically,
we will better meet their needs. This work will
take a few weeks and should be completed
to start the fourth datacentre Roubaix OVH
Valley: RBX-4.

Regards,
Octave