Cloud servers

Combell's network

For an optimal availability and excellent performances, Combell's infrastructure is entirely redundantly connected to two prominent transit providers (COLT and Tiscali). As a customer, in case of a peak load, if necessary, you can use the entire capacity of these lines (burst). The existing connections can be rapidly enhanced at any time with more capacity or specific additional carriers, which can be easily accessed in our data centres.

Each network component, up to the access switches, is redundant up to the third level. So, even if a transit line, an edge router and a core router fail simultaneously, the network remains fully functional. The entire network is equipped with high quality and durable hardware of Cisco and Juniper and was designed in close collaboration with certified consultants.

Combell also has its own optical fibre connection with a capacity of 2 x 10 Gbps to its second own data centre space, which is located about 5 km farther (LCL). This distance ensures extremely fast response times between both sites. If, despite the double access, both ISPs fail simultaneously in the InterXion data centre, the Internet traffic will still be able to find a route to your servers via the second data centre and through Combell's own (redundant) optical fibre link.

Combell offers the Multi-data centre option. With this option, your server and a copy of your server run in 2 physically separate data centres. If one of the data centres fails, the server in the other data centre instantly takes over.

Multi-data centre in depth

The concept is quite simple in itself. Combell has a double connection between both data centres. This loop has been set up redundantly and the route has been chosen in order to offer an alternative route, e.g. in case of damages caused by road works.

In each data centre, servers are placed, which are identical copies of each other. Should the server in one of the data centres not be available, the copy of the server would still be available.

Other Multi-data centre solutions use a DNS system, but with this system, changes are performed very slowly (up to several hours). Combell chose the Border Gateway Protocol (BGP), thanks to which changes are immediately active. Via Combell's own 20 Gbps optical fibre connection, synchronous replication is possible between both sites.

Thanks to a unique approach based on the Border Gateway Protocol (BGP), Combell offers a multi-data centre setup, that acts as a large logical network for the public. This offers huge benefits such as a much faster rerouting (compared with changes made via DNS) in case of failure or intentionally modified configurations. BGP also offers more granular routing parameters than its technical alternatives.

Features and benefits of BGP

Thanks to the Border Gateway Protocol (BGP), Combell's two data centres, which are linked via the company's own optical fibre network, are united in a single logical network.

BGP can manage millions of active routes. This is essential for an optimal access to applications that must be available to the whole Internet. Therefore, BGP is a real standard at the transit providers. Combell manages its network itself as far as routing is concerned and has an excellent knowledge of BGP and many other routing protocols.

BGP searches the fastest link at any time and per individual route. BGP bases its routing on configurable network rules. The routers constantly indicate which underlying systems are available via their route. When a route is suddenly not available anymore, or when a faster route appears, the neighbouring routers automatically choose the other route.

Alternatives such as DNS-based rerouting go together with very short TTL times (time-to-live), which causes heavier loads on your DNS servers. Moreover, you have no guarantee at all that all the other DNS servers will observe these TTL times. Changes are thus not automatically active in the whole world. BGP provides a solution to this problem and guarantees that all the changes are instantly active, everywhere in the world.

Combell's multi-data centre approach allows very high availability guarantees. Even if a terrible disaster should destroy an entire data centre, your application would remain fully functional.

Both sites are connected through an active-active setup to our transit providers, so that the performances are doubled in a multi-data centre setup. Another benefit of this architecture is e.g. off-site backup.