DCN September 2017 - Page 23

Colocation & Outsourcing In an ideal world, businesses should use diverse communication links to wherever their infrastructure is located, with both links capable of carrying the full load at peak time. The network should switch between these links seamlessly with users unaware a failure has even occurred. In this scenario, neither customers nor employees would notice any service impact in the event of a failure. In practice, fully redundant links are often expensive, and businesses need to understand both the risks their infrastructure carries, the potential costs of downtime and the potential costs of the diverse or redundant network solutions. Armed with this information management can then make an informed choice of risk vs cost. Risk and options assessment At the core of any organisation’s network design is a set of requirements and a risk assessment. This should define an organisation’s requirements for uptime and connectivity, by site and application and detail the impact associated with loss of services. From these requirements a set of options can be built, and their associated costs determined. Each option should detail the risk involved or eliminated, and crucially, any assumptions made. This activity requires a significant amount of due diligence and pragmatism, for instance, it is very common for carriers to lease fibres or buy services from each other. Significant work is required to ensure fibre routes that are intended to be diverse are as resilient as they appear. It may be necessary for an organisation to sub contract this work to a third party. Similarly, looking at the performance and uptime record of a service provider, especially when things go wrong is critical. There are many stories of floods or power issues which can take many hours to fix. One update received from a well known carrier indicated ‘Engineers are in the vicinity of the fibre break awaiting the fire brigade to declare the area safe’. That fault took well over 30 hours to fix, but due to good network design resulted in zero downtime or business interruption. Other network providers offer a 100% network uptime guarantee. This is attractive on the surface, but in practice, may be a marketing ploy, with the provider factoring in the cost of paying out on an SLA when their performance falls short. Cost of downtime In order to decide on an appropriate resilience option, the true cost of downtime must be understood. The basis of this may be simple, for instance loss of trade for an e-commerce site is easily quantifiable. More difficult are intangible costs such as loss of reputation for a website or service being down, or an inability to provide a full service to customers. The impact of reputation damage can last a great deal longer than any network outage. Ideally the cost of downtime of each critical application should be calculated, as different applications will have different requirements in terms of uptime and performance in the event of failures. Calculating this cost by application may allow cheaper, lower specification backup connections to be used, in the knowledge that only critical traffic will be carried in the event of failure. Cost of solutions ‘In an ideal world, businesses should use diverse communication links to wherever their infrastructure is located,’ Once the potential options have been established, they must be priced and ranked as to how they address the business need. For instance, one carrier may provide sub second failover, whilst another may provide a better SLA guarantee in the event of a fibre cut. Selecting a single communications provider may create a single point of failure from a supplier perspective, but it also ensures there is a one point of contact for support and management, which will increase speed of response in the event of failures and avoids the finger pointing which can occur in a multi supplier scenario. Selecting a single Wide Area Network provider is the most common route as an organisation can leverage the expertise of the selected supplier, who should take responsibility for delivering to agreed service levels. In practice, combining all the above trade-offs – probability of failure, time to fix cost of the solution and budget should lead to the most appropriate solution. The cost of a fully redundant solution can be a hard sell to management, but best practice dictates this cost should be factored in to the initial colocation or cloud project as a starting point. The same management who wish to keep costs low tend to be even less accepting of downtime when it occurs – even when identified as an ‘acceptable risk’. In summary, centralising data storage and processing offers a great many advantages to organisations both in terms of operations and cost savings. Any project to move services away from a site increases reliability on the network infrastructure. To deliver the full range of benefits cloud and data centre services offer the stability and reliability of communications infrastructure is key. September 2017 | 23