oles@ovh.net
06-30-2011, 03:44 PM
Hello,
It's been a while since the "what's new" at OVH has been updated so it's now time to give some info
Go get a cup of coffee before continuing ...
Datacentres
-----------
The new RBX4 datacentre was put into production and almost one floor is already filled with servers in the 2011 range. That happened quite quickly and there is no problem even when the weather was hot like the beginning of the week. The system regulates itself and with the pressure difference on the height of the building, and the electric fans use little power (which has replaced the air conditioning).
It's pretty magical. Worth seeing for yourself and so we try to share that experience here or there by small groups before really opened the doors of the cube.
On RBX1 we had some problems related to air conditioning of a room, room 6 (in 2006 was again used air conditioning for 30% of the heat). The origin of the problem is related to insulation against noise as we asked around the building at the end of 2010. There were two design errors. And this has been fixed.
We started work on containers that are leaving for Strasbourg. Preparations for this new datacentre is advancing at a good pace and in a few weeks, we will cut our network between Paris and Frankfurt to insert our datacentre into the loop. This will also allow us to directly connect Strasbourg to Zurich and Milan . It will be closer to our customers.
Network
------
We are currently deploying the network in the U.S., Asia and Brazil. The goal is to be present at the peering points in Roubaix as well as Hong Kong or Chicago or Dallas or Atlanta or Miami. Why? To direct traffic generated by your infrastructure in the most efficient way based on our network. At the same time we're building a "C2DN" network i.e. "Cloud Content Delivery Network." With C2DN, we can deliver the VMs to you instantly around the world, so you can migrate a VM from Paris to Dallas or Hong Kong to Madrid in minutes (powered by VMware vCloud). We can also provide you with the NSC through five datacentres with 10 POPs in the U.S. and POP 3 with a datacentre in Asia. Let's not forget Brazil. We will also have a capacity over the Atlantic.
In Paris we will move our Telehouse 1 router to Telehouse 2 and then all together to Th2 in a small room. The goal is to connect 100 NRA of Paris and the crown of Paris on TH2 and GSW. How? We validated DWDM transport equipment between the NRA and our routers that has allowed us to connect each NRA with as much capacity as we need today and in the future. 2G? 4G? 10G? 20G? 80G per NRA? Yes! We can!
Also, we do not have to put intelligence into the NRA, as a router or switch. So a simple network. With intelligence on POPs where we can make both the IPv4, IPv6 and private networks between two ADSL with a central point out from behind a firewall. The DSLAMs are directly connected to our routers at the heart of Paris, Lyon, Marseille, Lille and Bordeaux. The goal is to provide a xDSL service with real bandwidth guarantees , which does not row at night and can watch TV in unicast, enabling the use of all the SaaS world.
We wish to provide the connection for our customers to use services that will be created in the months and years to come. Basically, OVH respect the neutrality of the Internet and we provide the bandwidth it takes.
It is neither very complicated nor very costly.
Cloud Computing
---------------
Apart from providing C2DN we bump on the new features like (privateCloud) and the start of pCI (public Cloud) will be available in just 10 POPs worldwide. Who says pCI said the VPS.
That said, given our experience with dedicated servers, it is thought that despite the cloud we also have a role to play in terms of dedicated server resources. So we're working on the project to build a datacentre in the USA / CA. Looking and thinking about it. For example, next week we're visiting factory to buy in Montreal, Canada. But also an empty datacentre in San Antonio in Texas. Why these places? The base energy is cheaper than in France. So we're continuing to think about this now.
We have changed the VPS service so that you can enjoy a 99.99% SLA. It's a bit more expensive but offers 50G by default and has a better SLA. Thanks to VMotion, DRS etc..technologies that are behind the "99.99% SLA" In fact, we can now move the VPS from a "hot" host to another and from one system
storage to another. All this without a break in service. That's what we started doing to upgrade the entire VPS infrastructure. Unfortunately we were a little late because one of filers that was being used during the beta crashed yesterday. The fault is on our side: the beta infrastructure did not have HA as a disk shelf. After the beta it had been a bit too late to start migrating the infrastructures of the final customers who had renewed their VPS. All this because we did not generate a break in service for customers by 99.9%. We wanted to wait for 99.99% and for all ... we have waited too long. The service is again in place for all VPS, there was no loss of data, but we were very wrong on this one there and the duration of the outage. We've now upgraded the data recovery service so that if it happens on a NAS-HA it will be put on another one in a few minutes at most, even if the NAS-HA has two independent SCSI channels and therefore it will always be the worst one that works ...
In short, following the changes to the offers, we're finalising the infrastructure changes by synchronising the offers we're providing.
ADSL / SDSL
---------
We are in the process of unbundling to Marseille. Lyon, Bordeaux and Paris are in the pipe and everything should be finished by the end of September (I am very optimistic). We will at that time, our own 170 NRA through several loops and therefore completely redundant with a capacity of almost unlimited bandwidth for each NRA.
And maybe we can provide you with Netflix before anyone else! In any case our network will be open to everyone and will ensure a guaranteed bandwidth for each subscriber. Much like the dedicated servers.
Partnership
-----------
After months of discussing internal and external, that's it, we know what we will offer, for which service and why. We will guide these partnerships on services such as telephony and Internet access on three levels: white label sites/products, agents, integrator. Start up is scheduled this week. Don't believe it? Me neither We'll see
Support
-------
You can start thinking of OVH having two activities. On the one hand to web hosting, domain names, telephony, email and internet access. And the other datacentre resources that can be used through services such as dedicated server, the privateCloud, the publicCloud or VPS. When you look in the corner there, one concludes that OVH should have two websites and two supports for the objectives, customers, and services are completely different. In any case we understand why one and the same support can not be adapted to these two activities: on one hand we have customers who use wordpress and other customers who do hdparm on the disks. This is not the same job.
We continue to think that way internally.
Starting USA / CA
----------------
Within a few days, we will start testing with USA / CA customers . It is likely that the starting point will challenge some of the pricing strategy at OVH France and in Europe. w8 & c
All the best
Octave
It's been a while since the "what's new" at OVH has been updated so it's now time to give some info
Go get a cup of coffee before continuing ...
Datacentres
-----------
The new RBX4 datacentre was put into production and almost one floor is already filled with servers in the 2011 range. That happened quite quickly and there is no problem even when the weather was hot like the beginning of the week. The system regulates itself and with the pressure difference on the height of the building, and the electric fans use little power (which has replaced the air conditioning).
It's pretty magical. Worth seeing for yourself and so we try to share that experience here or there by small groups before really opened the doors of the cube.
On RBX1 we had some problems related to air conditioning of a room, room 6 (in 2006 was again used air conditioning for 30% of the heat). The origin of the problem is related to insulation against noise as we asked around the building at the end of 2010. There were two design errors. And this has been fixed.
We started work on containers that are leaving for Strasbourg. Preparations for this new datacentre is advancing at a good pace and in a few weeks, we will cut our network between Paris and Frankfurt to insert our datacentre into the loop. This will also allow us to directly connect Strasbourg to Zurich and Milan . It will be closer to our customers.
Network
------
We are currently deploying the network in the U.S., Asia and Brazil. The goal is to be present at the peering points in Roubaix as well as Hong Kong or Chicago or Dallas or Atlanta or Miami. Why? To direct traffic generated by your infrastructure in the most efficient way based on our network. At the same time we're building a "C2DN" network i.e. "Cloud Content Delivery Network." With C2DN, we can deliver the VMs to you instantly around the world, so you can migrate a VM from Paris to Dallas or Hong Kong to Madrid in minutes (powered by VMware vCloud). We can also provide you with the NSC through five datacentres with 10 POPs in the U.S. and POP 3 with a datacentre in Asia. Let's not forget Brazil. We will also have a capacity over the Atlantic.
In Paris we will move our Telehouse 1 router to Telehouse 2 and then all together to Th2 in a small room. The goal is to connect 100 NRA of Paris and the crown of Paris on TH2 and GSW. How? We validated DWDM transport equipment between the NRA and our routers that has allowed us to connect each NRA with as much capacity as we need today and in the future. 2G? 4G? 10G? 20G? 80G per NRA? Yes! We can!
Also, we do not have to put intelligence into the NRA, as a router or switch. So a simple network. With intelligence on POPs where we can make both the IPv4, IPv6 and private networks between two ADSL with a central point out from behind a firewall. The DSLAMs are directly connected to our routers at the heart of Paris, Lyon, Marseille, Lille and Bordeaux. The goal is to provide a xDSL service with real bandwidth guarantees , which does not row at night and can watch TV in unicast, enabling the use of all the SaaS world.
We wish to provide the connection for our customers to use services that will be created in the months and years to come. Basically, OVH respect the neutrality of the Internet and we provide the bandwidth it takes.
It is neither very complicated nor very costly.
Cloud Computing
---------------
Apart from providing C2DN we bump on the new features like (privateCloud) and the start of pCI (public Cloud) will be available in just 10 POPs worldwide. Who says pCI said the VPS.
That said, given our experience with dedicated servers, it is thought that despite the cloud we also have a role to play in terms of dedicated server resources. So we're working on the project to build a datacentre in the USA / CA. Looking and thinking about it. For example, next week we're visiting factory to buy in Montreal, Canada. But also an empty datacentre in San Antonio in Texas. Why these places? The base energy is cheaper than in France. So we're continuing to think about this now.
We have changed the VPS service so that you can enjoy a 99.99% SLA. It's a bit more expensive but offers 50G by default and has a better SLA. Thanks to VMotion, DRS etc..technologies that are behind the "99.99% SLA" In fact, we can now move the VPS from a "hot" host to another and from one system
storage to another. All this without a break in service. That's what we started doing to upgrade the entire VPS infrastructure. Unfortunately we were a little late because one of filers that was being used during the beta crashed yesterday. The fault is on our side: the beta infrastructure did not have HA as a disk shelf. After the beta it had been a bit too late to start migrating the infrastructures of the final customers who had renewed their VPS. All this because we did not generate a break in service for customers by 99.9%. We wanted to wait for 99.99% and for all ... we have waited too long. The service is again in place for all VPS, there was no loss of data, but we were very wrong on this one there and the duration of the outage. We've now upgraded the data recovery service so that if it happens on a NAS-HA it will be put on another one in a few minutes at most, even if the NAS-HA has two independent SCSI channels and therefore it will always be the worst one that works ...
In short, following the changes to the offers, we're finalising the infrastructure changes by synchronising the offers we're providing.
ADSL / SDSL
---------
We are in the process of unbundling to Marseille. Lyon, Bordeaux and Paris are in the pipe and everything should be finished by the end of September (I am very optimistic). We will at that time, our own 170 NRA through several loops and therefore completely redundant with a capacity of almost unlimited bandwidth for each NRA.
And maybe we can provide you with Netflix before anyone else! In any case our network will be open to everyone and will ensure a guaranteed bandwidth for each subscriber. Much like the dedicated servers.
Partnership
-----------
After months of discussing internal and external, that's it, we know what we will offer, for which service and why. We will guide these partnerships on services such as telephony and Internet access on three levels: white label sites/products, agents, integrator. Start up is scheduled this week. Don't believe it? Me neither We'll see
Support
-------
You can start thinking of OVH having two activities. On the one hand to web hosting, domain names, telephony, email and internet access. And the other datacentre resources that can be used through services such as dedicated server, the privateCloud, the publicCloud or VPS. When you look in the corner there, one concludes that OVH should have two websites and two supports for the objectives, customers, and services are completely different. In any case we understand why one and the same support can not be adapted to these two activities: on one hand we have customers who use wordpress and other customers who do hdparm on the disks. This is not the same job.
We continue to think that way internally.
Starting USA / CA
----------------
Within a few days, we will start testing with USA / CA customers . It is likely that the starting point will challenge some of the pricing strategy at OVH France and in Europe. w8 & c
All the best
Octave