oles@ovh.net
09-14-2008, 01:38 AM
Hello,
Here is the current status of deliveries (2008-09-14 01h00).
Servers to discover and to play with:
- Kimsufi L OK -> 1 hour
- Kimsufi XL OK -> 1 hour
- Kimsufi 2XL OK -> 1 hour
- Kimsufi 3XL OK -> 1 hour
- Kimsufi 4XL 24H delay
Pro, business and critical servers:
- miniSP OK -> 1 hour
- SP bestof OK -> 1 hour
- SP Max fr,es OK -> 1 hour
- SP Max de OK -> 1 hour
- SP Max uk,pl OK -> 1 hour
- EG Bestof OK -> 1 hour
- EG Max OK -> 1 hour
- EG Large OK -> 1 hour
- MG Bestof 6 days
- MG Max 5 days
- MG Large 5 days
- miniHG OK -> 1 hour
- HG Bestof OK -> 1 hour
- HG BiTurbo OK -> 1 hour
- HG Large OK -> 1 hour
Everything is going well. We sold just over 2,000 dedicated servers in 2 weeks. Deliveries are within 1 hour. As expected, the kimsufi 4xl (which becomes a new best-seller of Ovh) is out of stock. From Wednesday at the maximum the delay will be caught up. Backlogs on the MG began this week. Something unforeseen happened but everything is fixed now. We still have a small delay that will be caught up in a few days. Generally speaking, our objective to deliver to customers in 1h is reached.
However, we had some servers delivered to hackers with a very old nic handle (3 years) and a server deployed with a payment manually accepted (by me). To sum up, we can not tell the difference between a customer and a hacker. It's a lottery and the security of our network is becoming the lottery (servers of hackers have been closed because they were scanning the network, this is how we found them). We decided to change the settings for the validation of payments by post. Now, for a new server, you must validate the code sent by mail. Only the 1st time. We do not deploy any more new servers without the validation by mail. This is the only way to deliver servers without jeopardising the security of the network. This validation is at our expense and is free. You just need to give us the correct address, wait for 24H/48H for the letter, then validate the code you received: the pending order will automatically be validated and you get the server up to 1 hour after validation. Customers validated by mail, will no longer be limited in the number of servers per month. They can order 40 servers and receive them in 1 hour. Still no commitment.
Some customers may have applications for validation by letter when renewing their server. Generally we validate these payments manually, and therefore with a lag. Once the customer validates the letter, everything is automatic in the future.
All these measures are binding on everyone. But with this new generation of hackers and the quantity of stolen credit cards and hacked PCs, we can no longer distinguish an honest customer from a hacker. Initially these measures may seem excessive, and might make us lose a certain number of revenue in cases where the customer gets angry or does not accept such treatment. At our level, we lost 150'000 euros in 2 months. The choice between "accept all customers without any verification and lose 70'000 euros per month in non-payment" or "lose some customers and no longer have non-payments" is easy.
Let 's get back to the subject: Roubaix datacentre 1 (which was started in August 2006) will be full in 2 days! With Roubaix 1 full, we expect to have 40,000 servers leased. We have a few days of work left to exceed the figure. Roubaix 1 is our 2nd full datacentre. The 1st is in Paris 19. It started in 2004 and was also full in 2 years. In 2 years time, will we sell 3 times more than in the previous datacentre and sell this time 100,000 servers in 2 years? The datacentre has been scheduled for ... wait and see ... In any case Ovh will be much more agressive on the market shortly ...
The routing in Roubaix 2 will be started on Monday with a new Cisco router VSS 1440 [url]http://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps9336/product_solution_overview0900aecd806fa5d0.html [/ url] which will manage approximately 40,000 servers. It pings already in the lab 213.251.130.6. On Monday, we move it physically. We connect 10G transit of Teleglobe directly on the chassis (we move the transit between Paris and Roubaix on Infinera). Thus, the backbone is very short, direct and in 10Gbps. The packet of the server arrives on the chassis and goes directly to the Internet via intermediate routers, no intermediary 10Gbps ports , no redundancy routers. The router is already double and manages failures itself. We will be able to deliver as many servers as customers want while ensuring the flow at 1Gbps (and 2Gbps to burst) per customer because everything goes on a single router that has all the necessary redundancy features.
In Roubaix 2 we built 4 hosting rooms to begin with, of 40000 servers in total. To sum up, everything is ready. Starting this week. In total 120 jobs will be created with full-Roubaix 2. Indeed, we assure the management of our datacentres and our own technicians are on duty 24/7.
Regards,
Octave
Here is the current status of deliveries (2008-09-14 01h00).
Servers to discover and to play with:
- Kimsufi L OK -> 1 hour
- Kimsufi XL OK -> 1 hour
- Kimsufi 2XL OK -> 1 hour
- Kimsufi 3XL OK -> 1 hour
- Kimsufi 4XL 24H delay
Pro, business and critical servers:
- miniSP OK -> 1 hour
- SP bestof OK -> 1 hour
- SP Max fr,es OK -> 1 hour
- SP Max de OK -> 1 hour
- SP Max uk,pl OK -> 1 hour
- EG Bestof OK -> 1 hour
- EG Max OK -> 1 hour
- EG Large OK -> 1 hour
- MG Bestof 6 days
- MG Max 5 days
- MG Large 5 days
- miniHG OK -> 1 hour
- HG Bestof OK -> 1 hour
- HG BiTurbo OK -> 1 hour
- HG Large OK -> 1 hour
Everything is going well. We sold just over 2,000 dedicated servers in 2 weeks. Deliveries are within 1 hour. As expected, the kimsufi 4xl (which becomes a new best-seller of Ovh) is out of stock. From Wednesday at the maximum the delay will be caught up. Backlogs on the MG began this week. Something unforeseen happened but everything is fixed now. We still have a small delay that will be caught up in a few days. Generally speaking, our objective to deliver to customers in 1h is reached.
However, we had some servers delivered to hackers with a very old nic handle (3 years) and a server deployed with a payment manually accepted (by me). To sum up, we can not tell the difference between a customer and a hacker. It's a lottery and the security of our network is becoming the lottery (servers of hackers have been closed because they were scanning the network, this is how we found them). We decided to change the settings for the validation of payments by post. Now, for a new server, you must validate the code sent by mail. Only the 1st time. We do not deploy any more new servers without the validation by mail. This is the only way to deliver servers without jeopardising the security of the network. This validation is at our expense and is free. You just need to give us the correct address, wait for 24H/48H for the letter, then validate the code you received: the pending order will automatically be validated and you get the server up to 1 hour after validation. Customers validated by mail, will no longer be limited in the number of servers per month. They can order 40 servers and receive them in 1 hour. Still no commitment.
Some customers may have applications for validation by letter when renewing their server. Generally we validate these payments manually, and therefore with a lag. Once the customer validates the letter, everything is automatic in the future.
All these measures are binding on everyone. But with this new generation of hackers and the quantity of stolen credit cards and hacked PCs, we can no longer distinguish an honest customer from a hacker. Initially these measures may seem excessive, and might make us lose a certain number of revenue in cases where the customer gets angry or does not accept such treatment. At our level, we lost 150'000 euros in 2 months. The choice between "accept all customers without any verification and lose 70'000 euros per month in non-payment" or "lose some customers and no longer have non-payments" is easy.
Let 's get back to the subject: Roubaix datacentre 1 (which was started in August 2006) will be full in 2 days! With Roubaix 1 full, we expect to have 40,000 servers leased. We have a few days of work left to exceed the figure. Roubaix 1 is our 2nd full datacentre. The 1st is in Paris 19. It started in 2004 and was also full in 2 years. In 2 years time, will we sell 3 times more than in the previous datacentre and sell this time 100,000 servers in 2 years? The datacentre has been scheduled for ... wait and see ... In any case Ovh will be much more agressive on the market shortly ...
The routing in Roubaix 2 will be started on Monday with a new Cisco router VSS 1440 [url]http://www.cisco.com/en/US/prod/collateral/switches/ps5718/ps9336/product_solution_overview0900aecd806fa5d0.html [/ url] which will manage approximately 40,000 servers. It pings already in the lab 213.251.130.6. On Monday, we move it physically. We connect 10G transit of Teleglobe directly on the chassis (we move the transit between Paris and Roubaix on Infinera). Thus, the backbone is very short, direct and in 10Gbps. The packet of the server arrives on the chassis and goes directly to the Internet via intermediate routers, no intermediary 10Gbps ports , no redundancy routers. The router is already double and manages failures itself. We will be able to deliver as many servers as customers want while ensuring the flow at 1Gbps (and 2Gbps to burst) per customer because everything goes on a single router that has all the necessary redundancy features.
In Roubaix 2 we built 4 hosting rooms to begin with, of 40000 servers in total. To sum up, everything is ready. Starting this week. In total 120 jobs will be created with full-Roubaix 2. Indeed, we assure the management of our datacentres and our own technicians are on duty 24/7.
Regards,
Octave