Intelligent Data Centres Issue 10 | Page 22

INDUSTRY INTELLIGENCE POWERED BY THE DCA The increasing deployment of IoT devices will also drive a degree of migration to the ‘Edge’ but this is by no means the only driver. (and expect) instant results or low latency connection to services no matter where they are located. The ‘Edge’ in this sense is computing facilitated by smaller satellite data centres often themselves supported by one or more larger core data centres which might aggregate data and perform the deep analytics that businesses now expect and increasingly require. The increasing deployment of IoT devices will also drive a degree of migration to the ‘Edge’ but this is by no means the only driver. The Edge as a concept will continue to grow, shaped by the demand for proximal processing and data storage including data sovereignty and regulatory compliance as well as latency, etc. some have predicted. Core data centres will still be required for core functions including the storage and processing of large volumes of data accumulated from Edge sites. This is being offered in different ways by hyperscale, large enterprise, colocation and managed service providers. Where a data centre is designed and built with a high degree of site resilience and correctly maintained it should be recognised that the risk to uptime is very low and a permanent team of engineers based on site is not necessary. What are the challenges? The typical Edge data centre will be unstaffed and follow a dark site operating model typically with a smaller footprint than traditional core data centres. Whatever the definition or type of Edge being deployed, there is one aspect that is somewhat being overlooked. We increasingly rely upon, and expect, high reliability of data centre site operations and high (24×365), availability of the services hosted within those sites. This high level of service is often delivered by sites with multiple levels of equipment redundancy, a very comprehensive preventative maintenance plan and dedicated data centre engineers based on site in shifts 24×7. The high (possibly unreasonable) expectations around uptime and availability cannot automatically apply to smaller, remote dark sites which are the standard Edge model. How do we achieve this expected level of reliability without staff on site running traditional planned preventative management operations? Mark Acton – Critical Support Director, Future-tech SCi Ltd and a member of the DCA Advisory Board For many businesses there is an increasing expectation and need to have processing power and data held locally to wherever their customers happen to be. This is potentially a major consideration in delivering increased service quality and ultimately increasing both customer satisfaction and providing new or significantly enhanced products. In this sense we could reference the terms ‘competitive Edge’ or ‘gaining an Edge’. This does not mean the death of the traditional core data centre though, as 22 Issue 10 engineering teams and maintenance services. In the UK the typical support model is to employ a team of onsite engineers that work shift patterns at a large cost to the business. Since Edge sites are typically remote and unmanned there needs to be a new industry approach to providing support for these in order to deliver the levels of reliability and availability that customers now expect. This can be achieved if all applications and data hosted within a site have multi-site resilience such as site failover or genuine synchronous replication, however this is not the case in all Edge sites. This therefore raises a challenge. How do we maintain the continued availability of Edge sites when they are remote and unmanned? One of the biggest overheads for data centre operators is its ‘on premise’ A better way to operate In this scenario, without full time engineers on site, new operating models and predictive maintenance techniques need to be applied to ensure operational reliability in addition to the standard preventative maintenance plans. Edge sites should be seen as typically ‘dim’ or ‘dark sites’ where engineering support is performed remotely under contract by a third-party operator. This also has the great advantage of significantly reducing OPEX costs. Companies like Future-tech, who are already operating remote services for data centres on behalf of multiple customers, are therefore particularly well suited to those wishing to move to a ‘dark site’ operating model in order to reduce the extremely high costs of onsite personnel, yet maintain site reliability and availability. We need to make more use of predictive maintenance, effective remote monitoring tools, intelligent management systems using Machine Learning and analytics, as well as engineers and engineering teams with multiple skillsets including IT related skills. Routine site visits do need to be performed, yes, but with greater purpose and range of planned activities, upgrades or installations, depth of knowledge and insight rather than mere adherence to a rigid PPM Schedule. www.intelligentdatacentres.com