Data centres are one of life’s necessities but as global demand for compute power continues to rise exponentially, driven as much (or more) by cloud and web-based applications as the traditional needs for secure data storage, many data centres, even comparatively new ones of less than 10 years, are struggling to keep up.
The sheer scale of the secure physical spaces required, let alone the growing power and energy implications, are presenting data centre users and professionals with unprecedented challenges.
Jones Lang LaSalle’s last Data Centre Barometer survey among European stakeholders, including corporate users, carriers/integrators/web hosters and colocation operators and investors, found a significant swing towards outsourced solutions with around 75 per cent in favour compared to just 50 per cent 18 months earlier.
These findings, among others, combined with the continuing solid performance and growth of local and global independent operators, show the market for outsourcing some or all of the data centre estate to third parties to be in good health.
As a speedier, nimbler and commercially more viable alternative to DIY new build or legacy upgrade projects, data centre outsourcing clearly presents some compelling arguments and benefits.
Magnified by the continuing uncertain economic climate, the whole concept and complexity of planning, designing, building, not to mention the investment and capital expenditure involved, is making private ownership increasingly untenable.
On paper depreciating the build cost of a data centre over 15 or 20 years may appear to give a lower TCO. The reality is that an in-house facility is either sized to cope for 10 years of growth in which case it will be massively under-utilised for many years or it is sized for just the immediate requirement in which case it will be too small to enjoy the economies of scale of a larger data centre.
Juggling cost-benefits, space availability, investment ROI, asset depreciation, obsolescence, and making decisions about infrastructure for ensuring optimum operational efficiency, are all additional drivers behind growing appetites for outsourced pain relief. And with IT planning horizons now reduced from the traditional 2 years to 3 to 6 months the use of on-tap collocation space is often the only solution available.
Nevertheless, deciding to take the leap of faith and outsource is only the start of the journey. For corporate/public sector colocation occupiers and managed hosted services providers, there are still important factors to consider before selecting the most appropriate independent data centre facility.
Power, Cooling and Energy Management
Power, cooling and energy management are critical to overall resilience and uptime. Demand for power and cooling within data centres will continue with the steady increase in processor power producing the effect of increasing heat and thermal rise. More usage of high density server applications has also resulted in more intensive use of a given footprint resulting in a rise in energy consumption per square foot.
As the sole responsibility of the third party in an outsourced environment, look carefully when checking into a prospective candidate’s energy credentials.
Clearly energy cost is a central issue, and with cooling averaging 40 to 45 per cent of total operating costs for many data centres, the sustained rise in electricity charges has spurred the search for new alternatives and solutions. These include water and liquid cooling as well as nano-cooling technologies and others on the horizon that could ultimately require certain older facilities needing complete re-engineering.
Location and security
The perceived convenience of choosing facilities in close proximity to corporate headquarters and the potential, albeit diminishing, concerns over distance on latency, led to most first generation data centres being located in or around large metropolitan areas. While after 9/11 many US organisations realised it was time to move facilities well out of town, the UK still suffers heavily from this legacy especially in the London area where newer generation facilities have and still are being built.
From a physical security perspective this presents a higher risk potential for attack or breach. Greater consideration must be given to how mission critical it is to have the data centre in the metro area and compare with operators offering more naturally secure rural locations well away from city centres, large populations, airports and the inherent security risks these present.
To look for:
- Ideally the site will be safe from flooding, flight paths, ground contamination and have good road access. It will be away from congested, urban areas with sufficient acreage to implement high security measures. It will have good fibre connectivity and most importantly access to a lot of power
- Buildings should be as invisible as possible with landscaped areas that ensure visitors and their vehicles enter via manned and gated (preferably air-locked) security checkpoints before passing into parking areas sited well away from the building
- Check ingress points are limited to a single main entrance and a loading bay. Keep windows to an absolute minimum, ideally these should feature bomb-proof glass. These should be reinforced by anti-ram defences placed at strategic points around the building, especially entrance points
- Prison grade perimeter fencing, CCTV and speed intrusion prevention systems (IPS) are pre-requisites
- Double or triple skin reinforced concrete walls and earthquake proof foundations are preferable
Communications Links and Latency
Decreasing costs of fibre combined with remote diagnostics are offering users real choice and potential costs savings by outsourcing to facilities located much further away. The aforementioned Jones Lang LaSalle report found around 50 per cent of corporates would consider locations in alternative countries in Europe.
To look for:
- Check the data centre is carrier neutral to ensure as much choice and flexibility as possible for regional, national and international communications
- Latency: Modern facilities with direct on-site fibre interconnects will offer superior latency; NGD Europe now offers sub 2ms transit performance between South Wales and City of London, a distance of over 150 miles
Business Continuity/Disaster Recovery
Ensure the operator has effective business continuity planning in place that can accommodate disaster recovery plans as well as orchestrate effective day-to-day communications remotely if needs be.
To look for:
- All equipment should be fully fail-over capable and endure multiple concurrent equipment failures to ensure the facility can continue to function unhindered
- Fire protection and suppression: NGD Europe deploys a dual stage system incorporating detection and suppression which uses an IG55 suppressant. The use of an inert gas is effective at combating fires and, unlike fluids or foam, there is no clean up. Power need not be interrupted so equipment within the centre remains unaffected
- Ensure data halls and plant rooms are fitted with VESDA, (Very Early Smoke Detection Apparatus). and appropriate fire suppressants
- All systems must be monitored 24 x 365
Space and scalability
Last but not least, consider the future: According to the recent AFCOM data centre association’s ‘The State of the Data Center’ global survey of data centre managers, facilities have been expanding in size; with over 44 per cent now occupying more floor space than three years ago. Another 49.4 per cent are in the process of expanding with only 16.4 per cent having downsized.
To look for:
- Potential for accommodating expansion in the medium to long term from organic growth and/or by consolidation of further parts of the estate
- Flexibility to meet and house changing compute requirements such as Grid, Cloud and Container