Project · Network
Network core across two data centres
Core routing stretched across two sites: the existing Catalyst switches stay, the new server room gets a Cisco Nexus core.
Topology
Two sites, one core
The network ran across two data centres at once. We stretched the core routing between them: the working Cisco Catalyst switches stayed in the old site, while the new server room got a core built on Cisco Nexus.
Hardware that worked stayed where it was. The new environment could go live without replacing the entire core in one move - usually the most expensive and riskiest part of a change like this.
The link
Layer two over fibre
A fibre link ties the two sites together, and we carried layer 2 across it. Dozens of servers spread over both locations therefore run on one network rather than two separate ones.
That topology has a price: the core on both sites has to be designed and operated as a single whole, because a change on one side affects the other. So the order of work and the conditions for rolling back are agreed up front, not mid-change.
Technology
What it runs on
A Cisco-based core, tied together by a fibre link between the sites.
Cisco Catalyst
The existing core in the old site - still in production.
Cisco Nexus
The new core layer in the new server room.
Fibre link
Layer 2 carried between the two sites.