Hello,
In January and February, we carry on the work on the network. 3 goals are targeted:
- to increase the network capacity locaaly (in relation to our subsidaries)
- to improve the quality of the bandwidth (reduce latency network)
- to make the network reliable and increase the redundancy (and to avoid breakdowns in the event of an incident)
Therefore, we are going to work in January/February on POP of Frankfurt (DE), Amsterdam (NL), Lyon (FR), Marseille (FR) and Milano (IT)
Here is the detail of the work:
Work on the Brussels / Frankfurt (via Cologne) has started. In about 8 weeks, we will have a link of 100Gbps between our datacentre Roubaix City and Frankfurt. This will enable us to improve ping between the datacentre and our customers in DE, CH, AT, CZ and IT. Indeed, the traffic currently goes from Paris Roubaix then to Frankfurt, and will go from Roubaix to Brussels to Frankfurt. We should win 5ms/10ms for DE / CH / IT and 20ms/30ms for AT.
- We will strengthen POP (Point Of Presence) in Frankfurt with creation of an "important" local network (as in Paris). We will have 2 routers have instead of only 1 (in Interxion 4 and Fibernet), this will enable us to better manage links Paris/Frankfurt and Roubaix / Frankfurt. Overall, we will have 80Gbps capacity interconnected on 2 routers in Frankfurt.by the end of February.
- A 10G private peering will be established with TPSA (the telephone provider in Poland) in Frankfurt. Initially planned in Amsterdam for end of January, we have decided to delay the 1-month implementation of the peering to benefit from our network of Frankfurt. The objective is to move this peering directly in Warsaw to go faster and given the amount of bandwidth that we use to TPSA, we already do that in Frankfurt then we will discuss quietly in 2009 with the real technical data on what we will do in the future.
- We are increasing the capacity of the network Roubaix / Brussels to 200GBP allowing us to add new 10Gbps links between our datacentres in Roubaix and Amsterdam. The aim is to interconnect the routers of Roubaix 2 directly on the router in Amsterdam instead of passing traffic through the datacentre Roubaix 1. Also, we wil be able to interconnect new transit links (in Amsterdam) directly on the routers of Roubaix 2 datacentre (and avoid to use traffic through routers). It will improve the quality of bandwidth while continuing to ensure high network capacity. It also aims to prepare the start of our subsidiary in NL where we wish to propose our services with high quality network.
- Beginning of February, we will be present in Lyon and Marseille in datacentres of Neuf Telecom (LDCom). The aim is threefold:
- Improving the redundancy of the network towards Madrid and Milan - Seeking regional peering national operators
- Proposing ChtIX offers in Lyon and Marseille
- We are currently looking to how to improve the quality of our network for our IT customers. This will happen with a local transit provider in Milano to control the flow of our traffic locally (instead of "we do not know where it goes" between Paris / Frankfurt / Zurich / Milan). We are quite disappointed with the state of MIX peering. Indeed, many large IT networks are interconnected on MIX but with reliable capacity network and without the possibility of change in 10G. We have peerings with such networks but in Amsterdam and London. So instead of using the traffic locally, we continue to sell traffic to some IT network via peering in Amsterdam and London, which makes no sense at all. We continue to discuss with officials of these networks and motivate them to increase the capacity of their network locally and with us in Milan showing them that peering with their network costs them cheaper in Milan than in Amsterdam. It's slow.
- We will set up a POP in Euratechnologies in Lille / Lomme. The goal is to provide ChtIX services and telephony on the Euratechnologies pole and support businesses in North of France on the Net. This is a "pilot" project for Ovh that will be tested to know whether to set up this type of POP (where there are between 50 and 100 companies) are to be used in the future for the development of ChtIX and the telephony at Ovh or if it is a "fake good" idea. Set up: around March / April.
Incidents:
- We had an interesting incident on the router of Roubaix 2 that we have resolved with 2 small cuts of 4-8 minutes. The system works then pretty well even if it is a very very young product. We will have to wait for a new version of IOS software (SXI1), which should be released in 2-3 months to fix the bugs that we had. Meanwhile, we know why and how and we should spend this time with few breaks.
Find out more:
http://translate.google.com/translat...hl=EN&ie=UTF-8
- For 48 hours, we have stability problems with Amsterdam / Warsaw link. We are seeking the origin of the problem
Learn more:
http://translate.google.com/translat...hl=EN&ie=UTF-8
Future already decided:
- In May, we will start work on Roubaix / London network. The goal is to have 100Gbps capacity to London and transport transit from London to our datacentres (directly instead of using this traffic through existing routers).
Regards,
Octave
