Managing Critical Mechanical and Electrical System Replacement in Data Centres: A Business Challenge

Over the past 20 to 25 years the construction of new data centres has been very much a focus for this sector to align with the exponential growth and demand for data storage and processing as well as cloud-based services, cryptocurrencies and the ever-expanding number of people and devices that are connecting to online streaming services such as Netflix and Spotify, which have millions of concurrent users.

‍

By today’s standards those aging and obsolete data centres that were constructed over this same period and still in operation, are now facing the challenge, at best, of having to replace critical pieces of plant and equipment, at worst, are in need of major refurbishment; both of which present both a financial burden and logistical challenge for the owner operators of these data centres.

‍

Data centres as a complete building and building services entity are typically considered to have an economic and reliability lifespan of between 15 and 25 years, despite many elements of the building services and building fabric having a potentially longer life span. For example, steel structural frame and cladding system could have a potential lifespan of up to 60 years.This contrasts with the IT equipment located within the data halls which it typically updated every 3 to 4 years to align with technological advancements.

‍

As technology changes and enhances and targets to become more sustainable are imposed, so do the demands on data centres to achieve higher power densities, but at a lower energy usage which in turn imposes pressures on outdated and less efficient power and cooling infrastructure. With the current advent of the Internet of Things, Artificial Intelligence, 5G technology and the like, which demands greater power densities and more innovative methods of cooling modern IT equipment such as liquid cooling, significant financial investment to keep pace with this is unavoidable.

‍

Therefore, critical mechanical and electrical plant and associated systems will need to be replaced to either align with the development of IT technology demands or where the main plant naturally reaches its economical end of life stage, or sometimes for both reasons combined.

‍

Data centre owner operators who have invested millions of the applicable currency into new build facilities over the past 20 to 25 years should have injected further financial investment over the following operational years for planned and reactive maintenance purposes and should have also had the foresight to build up significant cash reserves or seek additional equity or debt finance to anticipate the funding of Capex for major plant asset replacements at least once, if not twice, over the life span of their data centre portfolio. As with any occupied and operational building, a data centre requires constant and consistent maintenance and upgrading which is very costly and requires expertise both in-house and externally to be involved in this. As the data centre facility ages over the years, its infrastructure, plant and equipment are no longer considered as being “state of the art” and more often than not, existing legacy installations cannot be replaced like for like as they are no longer supported or maintained by the suppliers that they were procured from originally. Therefore, the funds need to be set aside to pay for both maintenance and major upgrading works when the need arises.

‍

Focussing on the upgrade / replacement of end-of-life plant assets, upgrading in a live data centre environment is fraught with challenges to overcome. Swapping-out new for old plant has to be meticulously planned for this to be seamless and with minimal disruption as possible to maintain continuous active plant and equipment up-time. Given that the duration of such projects can take many months and in some cases a number of years, the planning, sequencing and co-ordination of these works is not an insignificant activity and requires the full cooperation between the operations teams and key stakeholders. In addition, more often than not due to the lack of up-to-date operation and maintenance manuals, a significant input is required initially to as certain the as-built condition of the main plant infrastructure and distribution installations and to calculate the existing power utilisation levels before any proposals for upgrade / replacement works can be conceptualised.

‍

Powering down to facilitate the plant swap-out, even forTier III certified data centres having two concurrently maintainable power and cooling paths, will still involve replacing the plant one at a time to mitigate any risk to the operations, hence the reason why the duration of these projects, as referred to above, is so long. Any interruption to the requirement for continuous critical power and cooling systems creates a significant risk to business as usual operations. Working on one path, whilst the facility operates from the other must be carefully co-ordinated so that if the load sitting on the other path in operation drops, there is a swift back out plan to bring the services back online.

‍

Plant swap out works may necessitate the provision of temporary power and / or cooling installations, but even in these situations, the works involved may still present a high risk to the normal operation of the data centre.

‍

Data centre owner operators, in particular for co-location facilities, are facing a particular conundrum when encountering the need for the replacement of end-of-life plant assets. Traditionally their business case justifications for large tranches of funds for Capex works within existing data centres are predicated upon expanding power and cooling capacity to create sellable IT power capacity to their customers, which in turn generates revenue for the owner operators. However, where there is only a requirement for end-of-life plant replacement to maintain business-as-usual status, the business case for funding becomes a challenge, particularly for data centre owner operators who, up to now, have not encountered these circumstances where significant Capex expenditure does not result in additional revenue generation. Therefore, the justification of a business case for this expenditure becomes a challenge if it is only to maintain current SLAs with its existing customers.

‍

All is not lost at this stage and there are other possible justifications that provide positive outcomes that could just convince the CFO to sign off the Capex funding required.

‍

The business case justification for end-of-life plant replacement could be supported by one or more of the following:

‍

• ^{The outcome of achieving efficiency improvements(operational and energy based) through the procurement of the latest innovative technologies that significantly contribute to a lower annual average PUE andWUE1, which also benefits reduced ongoing Opex costs.}

^‍

• ^{Opportunities to release, subject to some infrastructure reconfiguration, previously locked or trapped power and / or cooling capacity to data halls that were previously unusable but for the need for end-of-life plant replacement enabling this to be remedied – particularly if this also aligns with customer churn or migrations that then enable re-evaluation of power utilisation levels and power densities to facilitate higher customer rack densities - which in turn could generate additional revenue.}

^‍

•  ^{The satisfaction of maintaining the SLAs agreed with customers without fear of a major downtime, penalty costs and crucially reputational damage, due to critical plant failures.}

^‍

•  ^{Contribution to reducing Scope 1 carbon emissions as part of overall carbon reduction targets that data centres are being challenged to achieve in response to the sustainability agenda.}

‍

The above list of points assumes that data centre owner operators have very limited control or influence over the environment or IT equipment located within the data halls, which generally does apply to co-location facilities. Should this not be the case, particularly where the Hyper or MegaScale owner operators are concerned, they would also have the flexibility to expand their opportunities for a business case justification by proposing to expand data hall ‘whitespace’ internal temperature and humidity range which will improve the energy efficiency and lower the power usage of the main plant and equipment.It would also minimise the power usage of IT equipment by selecting energy efficient technical devices located within the racks, which in turn lowers the power draw that they require to operate.

‍

As more and more aging data centre facilities become eligible for upgrading, end of life plant replacement or wholesale refurbishment, this opens up a whole new work stream for both the data centre owner operators and the consultants and contractors who already operate in the data centre sector. However, this type of work presents a completely different challenge that requires a different mind and skill set to enable projects of this nature to be delivered successfully.

‍

Evaluation Consultants have developed the knowledge and experience in recent years working with data centre owner operators on projects where end of life plant replacement has become necessary for their existing facilities. We have gained an insight into the complexities and challenges that are encountered, together with the intricacies of meticulously planning, sequencing and mobilising the swap outs as well as robustly costing the financial implications of such an undertaking.

‍

If you are planning to undertake a programme of upgrading or end-of-life plant replacement of your main plant and equipment at your data centre, or even a wholesale refurbishment project, please get in contact with us as we can support your project with our capable and experienced project managers and cost consultants.

^{Note 1} - The energy efficiency (not sustainability) of a data centre is commonly measured on a scale of power usage effectiveness (PUE), which equals the total power entering a data centre divided by the total energy used to run the equipment within it.So perfect efficiency would receive a PUE score of 1.0. The same principle applies to water usage effectiveness (WUE) for water usage.

‍

‍

‍

‍

‍