Thelen
07-03-2013, 03:52
Yep agreed there. If only software supported v6 only. Most of the VM software doesn't yet :/
Hello,
What's New at OVH?
Lot of projects are being finished
and go into production. To see more clearly,
a small recap what we occupy our days.
Datacentres:
------------
DC RBX5, capacity 14,000 servers, will soon
be full, probably sometime in March.
It provides a RBX6 end of April. Then the following
in 2015. We have indeed bought a lot of
new buildings in Roubaix Valley and can be
build RBX7 RBX8 RBX9 RBX10 .. but in two years.
This is the time it takes to bring two EDF
arrivals 20KV a new address.
This is to avoid saturating completely RBX,
we started the 2nd DC SBG: SBG4.
It was put into production there 3 weeks
and discharge already RBX5 for customers
, PL, IT. In parallel, the DC is SBG2
under construction and is expected
for Q3'13. In the meantime, the ends
site security and fiber optic
increasing the electrical power
through the work of the GC port
Strasbourg. SBG2 will actually mount
power by the end of 2013 the network shortest
to IT that will build on our dark fiber.
To take the following RBX, we
GRA expected to Gravelines. DC will start GRA1
current March. It is currently
make the GC to move the fiber optic
and connect to two networks:
Paris / London (bypassing Roubaix) and
Gravelines / Brussels. DC aura 3
outputs to 3 directions: London, Brussels
and Paris, and will be connected to BHS and RBX
live through loops secure.
As you understand, the month of March and
April will be busy in the event. It is
all to continue to deliver you in 1H, but
between RBX5 becomes full, which started SBG4,
GRA1 RBX6 and that will start, it is not
can not we have a few days, max 2
weeks of uncertainty, due to lack of
our place in DC. There is a risk, even if
it manages. So if you have plans for
March / April, please take a little
in advance.
The continuing American we sell
more servers from our
DC BHS. Because success is exponential, we
have started the construction of the 2nd round
BHS2 whose capacity is 10000 servers. The
Following BHS3/BHS4 be built according to the model
RBX5 of that room to start in 2000
servers in just 2 weeks. BHS3/BHS4 will
also supplied as 400V in Europe, instead
from 480V to BHS1/BHS2. Indeed, we have succeeded
to standardize and validate the tensions European
in North America
We started the search for
build 3 DC in the USA: It can not be # 1
"World" without being connected to the No. 1 market where
growths are most important. It is a
project that should take us less than 2 years.
NETWORK:
-------
We have completed the capacity expansion
USA on our network. Currently we push
20% of our traffic from Europe through the
new network is already very well connected inter
operators with American and South American
Asian. DC on our BHS: we have a
slightly less than the output capacity of 500Gbps and
actual traffic exceeded 30Gbps It just goes.
In Europe, we have completed updates
our network along distance between RBX / NDA / AMS.
We changed technology (always
Infinera) and we use "100G coherent" for
our transport circuits between DC and RBX
main points of presence in Europe. Week
Next, we take the work on AMS / BRU / RBX
to complete the first loop and secure optically
all circuits. This week we will
update the Cisco ASR9 of RBX, Paris and SBG.
This will allow us to add capacity in our
network in Paris but also fix the bugs
netflow that prevent us from properly handle
attacks.
Speaking of attacks, we test solutions
Arbor Network, which can detect an attack
the return on boxes and mitigation of
neutralize. We also test for some
Tilera months for filtering time attack
real. 2 give good results. It is
trying to decide which solution we will retain
to implement a comprehensive solution for each
DC. The goal is to treat attacks and 200Gbps
more.
To return to the network, after the loop
RBX / NDA / AMS / BRU / RBX, we will work on
loop RBX / FRA / SBG / PAR / RBX with the aim of securing
all circuits. Why secure circuits?
When failure of optical fiber (an excavator which
hard fiber), the circuit 10G between 2 routers
is cut. Unless it is protected in optical
another way. Thus, when the fiber is
failure, the circuit is UP because the equipment has
automatically used another optical path. It
There is therefore more break in service. This type
network is easier and cheaper to build
in 100G coherent.
Should sign the optical fibers between SBG and
Milan, IT in a few weeks and then put in place
this network in September. In parallel, we look for
a DC location in Germany ..
CDN:
----
BETA is going pretty well. We fixed bugs and
sees that already manages more than 1.5Gbps of traffic
our CDN. Thank you to the BETA testers for feedback
that allow us to adjust the settings.
We think the first start offering commercial CDN
Dedicated in a few days for the number 9.99e/mois
unlimited domain with an IP Anycast.
In parallel, we have planned to start the DNS service
Based on the same infrastructure. DNS DNS Low Latency aka LL
1e/an/domaine cost him.
VPS:
----
Offer great start. We are very happy because we
think it is a good tip for a VM offers 100K / year.
By cons, we have some concerns internally when
First deliveries. It lacked the resources
filers and we noticed a bug in the vCloud enough
late .. short, we made some mistakes.
VPS Cloud are new and correctly delivered
eventually catch up with the passive. End of the first week
VPS Cloud will be all set.
The manager should be out in a few days. At this time
then we get out the offer of GAMMA to make FINAL
So no bug.
pCI 2013
--------
Private version of VPS with LB, VPN, FW, IO HD
guarantees, multi site and time is the pCI 2013.
We think out the BETA end of March.
Hubic:
------
The infrastructure is stable and it's nice. We
still testing different clients on iPhone,
iPad, Android, Windows and WepApp. There is still
few days of work before making them public.
As we said, we want to offer you a
works. Short.
Customers who have trusted us and have paid a
subscription, will have a 2nd year free. All
this will happen at the same time.
Mutu & Geocache
---------------
Work on shared hosting now well forward. The main
problem is due to the fact that the tools we
developed to monitor the activity of some sites
have not taken into account the actual use of
CPU / RAM each site. It was thus found
with a lot of big sites who took 40% of capacity
CPU / RAM below. As 7000 comes to servers
HG, 40% is still 3000 HG servers. The last
work consists in replacing scripts
"Police" (the famous okillerd) to share equitably
resources among all clients. Is used
BigData and MapReduce to collect statistics
resource use on clusters. This we
allows you to see problems more quickly.
We think the problems are still felt on
approximately 5% of sites. Work continues to arrive
0% then update offers Mutu 2013 ..
To accelerate the mutu, we will integrate a CDN default
called internally GeoCaa 75% of our investments
this period. The rest is done by self,
through actuators that do not require the
dividends at the end of each year. This we
will support our growth in France and
in Europe thanks to your commands.
I'll stop there. It's very long to read and
heavy to digest.
In short, it works hard every day
Regards
Octave