Since its launch in 1982, Channel 4 has seen major changes in the broadcast industry, and its London base had evolved from traditional television production and broadcast operation to a role more like a corporate multimedia business headquarters. Channel 4 was planning to upgrade its Disaster Recovery (DR) and Business Continuity Planning (BCP) facility—and that called for a rethink of its legacy network.
The plan to locate disaster recovery in a new secure facility presented the double challenge of moving simultaneously to a new data center and a new network—just the sort of complexity that makes it difficult to predict performance and identify the source of faults arising. Instead, Channel 4’s Bruce Rawstorne, head of operations and senior service manager for Channel 4, decided to work on the existing system and redesign the network architecture as if the main and DR functions were already separated: “Let’s make our mistakes and learn from them in advance, rather than while we are in the midst of adapting to a new WAN-linked data center.”
What Bruce found was a legacy system “like a MINI fitted with a Ferrari engine.” The network already boasted Extreme Networks high power Black Diamond switches that could serve an operation many times larger. The 40GB backbone linking the two sites was also over-provisioned with dark fibre. As Bruce explains: “This was the first company I’d worked for where the network was not the issue. Normally, when anything goes wrong everyone begins by blaming the network—they’ll do anything rather than admit that the fault lies in their own systems. This did not apply at Channel 4.”
This system was no doubt its legacy from the days before transmission and broadcast was outsourced, and provision had been made for supporting
massive media file transfers rather than more normal business applications.
What was now being planned did include mirroring production storage clusters between the two sites, but this synchronous transfer required only the transfer of incremental updates, rather than full files, once the mirrors had been initially aligned.
The amount of spare capacity meant that IT operations between the two data centers had grown organically into a campus-style network without the usual discipline of limited resources. Such a system would prove costly and inefficient if migrated to a remote data center in its current state. What was needed was a slimming operation towards a leaner, less wasteful network that could more readily migrate to a remote data center with outsourced connection.
As a BCP system it was necessary to not only mirror the production files between the two sites—so that the loss of one site would mean a replica was immediately available—but also to make the network fully resilient so that it would still operate if one or other other site was shut down or destroyed. While latency was not a problem in the old network, in the future deployment the two data centers could be up to 100km apart and connected by a 20GB virtual private line with a number of DWDM channels instead of all that dark fibre. So latency could also become an issue for certain applications.
The legacy network was based on Extreme Networks equipment provided by Metropolitan Networks, but it was necessary to get comparative quotations for other vendors’ solutions. “We looked at things like capability, throughput, how the company’s technology was developing and price,” explains Bruce Rawstorne. “The cost of the Extreme Networks solution was considerably more attractive, it offered sufficient spare capacity and the end of support date gave us the required five years – so we went for that.”
“While our support and maintenance contract was with Metropolitan, Extreme Networks was very good when we needed their help,” adds Bruce. “We got them to validate our detailed design, and they were very responsive when more complex problems arose.”
A period of intensive testing followed. “There’s quite a bit of important traffic between the two, so we needed to be sure it was working properly before the move when, in theory, the only difference should be the longer link between the sites,” says Bruce.
At Channel 4, network capacity was never really the issue. Bruce Rawstorne had inherited a very powerful system but it was no longer appropriate to the new business strategy: “We are trying to mature Channel 4’s network and ingress to be more like a fully functional media company with serious enterprise capabilities for future growth.”
As a result of this upgrade to industry “best practice,” Channel 4 is now better placed for any future.