oles@ovh.net
03-06-2011, 09:50 AM
Hello,
We have completed work in London. The POP has been transferred to another datacentre. We have set up the second router. Linx 224 is past 4x10G and the Linx 226 remains at 2x10G. We still have transits to do. The collection of the VoIP with BT is through the old bay that we could not be transferred.
A small photo during the work:
http://demo.ovh.net/fr/5b7cb9181cd90...b8282f51d3d3d/
In the middle we have the Infinera transport with 140Gbps of capacity from Roubaix and from Amsterdam. Then the 2 sides of the 2 routers: ldn-1-6k and ldn-5-6.
The network schema:
http://weathermap.ovh.net/london
and global:
http://weathermap.ovh.net/backbone
On the MRTG graphs we now see that a peer that we have on Linx, manages the QoS brutally as with a chainsaw. Between 10am/6pm and 1am, it must make a QoS by limiting all its peers (or OVH only) at a fixed value to prevent the links from its network from saturation ... like that before, it saturates its network and therefore the problem does not exist! ...
The MRTG graph:
http://demo.ovh.net/fr/22f03f3cbc478...f1d3ce3f7038b/
The next step: increasing transport capacity of Amsterdam/Warsaw from 2x10G to 4x10G and Frankfurt/Warsaw from 2x10G to 4x10G. We will also remove 1x10G Prague/Warsaw which is useless. The 8x10G to Warsaw will be tied on 4 routers: 2 to Amsterdam and 2 to Frankfurt and with traffic management with load balancing, from Roubaix, we will exploit all of these links.
The step in parallel: increased capacity to Italy with an upgrade from 2x10G to 4x10G to increase capacity and peering with Interbusiness (AS3269) to from 1x10G to 3x10g.
And more: the USA.
Towards the end of the year we'll be able to increase the capacity of our 2 routers in Roubaix (ASR) from 1.2Tbps to 3.84Tbps ... and thus triple the outgoing bandwidth of Roubaix.
All the best,
Octave
We have completed work in London. The POP has been transferred to another datacentre. We have set up the second router. Linx 224 is past 4x10G and the Linx 226 remains at 2x10G. We still have transits to do. The collection of the VoIP with BT is through the old bay that we could not be transferred.
A small photo during the work:
http://demo.ovh.net/fr/5b7cb9181cd90...b8282f51d3d3d/
In the middle we have the Infinera transport with 140Gbps of capacity from Roubaix and from Amsterdam. Then the 2 sides of the 2 routers: ldn-1-6k and ldn-5-6.
The network schema:
http://weathermap.ovh.net/london
and global:
http://weathermap.ovh.net/backbone
On the MRTG graphs we now see that a peer that we have on Linx, manages the QoS brutally as with a chainsaw. Between 10am/6pm and 1am, it must make a QoS by limiting all its peers (or OVH only) at a fixed value to prevent the links from its network from saturation ... like that before, it saturates its network and therefore the problem does not exist! ...
The MRTG graph:
http://demo.ovh.net/fr/22f03f3cbc478...f1d3ce3f7038b/
The next step: increasing transport capacity of Amsterdam/Warsaw from 2x10G to 4x10G and Frankfurt/Warsaw from 2x10G to 4x10G. We will also remove 1x10G Prague/Warsaw which is useless. The 8x10G to Warsaw will be tied on 4 routers: 2 to Amsterdam and 2 to Frankfurt and with traffic management with load balancing, from Roubaix, we will exploit all of these links.
The step in parallel: increased capacity to Italy with an upgrade from 2x10G to 4x10G to increase capacity and peering with Interbusiness (AS3269) to from 1x10G to 3x10g.
And more: the USA.
Towards the end of the year we'll be able to increase the capacity of our 2 routers in Roubaix (ASR) from 1.2Tbps to 3.84Tbps ... and thus triple the outgoing bandwidth of Roubaix.
All the best,
Octave