Carsten Ludwig, Market Manager DC, Reichle & De-Massari AG
Modular data centres provide easy scalability, lower costs, more efficiency and significantly shorter time to market. What do you need to keep in mind when designing, specifying, building, testing and operating a modular DC?
A Modular data centre is, essentially, an integrated, predesigned set of modules selected and configured with possible expansion in mind, as well as the possibility of downscaling or altering functionality as and when circumstances require this. The data centre is fully functional from the outset, but can be scaled as the client’s business develops or as their IT needs and processes change. Anyone who needs to get up and running fast, or who works in an especially fast-moving market, could benefit from this design approach.
A ‘building block’ approach can make it possible to quickly and independently develop and organise the most efficient DC architectures, networks and services possible. Each element can be freely configured: from software to physical infrastructure, backhaul, interconnects, networks, and meet-me rooms. A modular solution can be precisely tailored to customer needs and topology. The modular design approach offers enhanced flexibility to accommodate growth or changing requirements while providing an opportunity to spread investment costs over a longer time. Costs remain fixed until utilisation increases and depreciation can be managed in stages.
All components of a prefabricated, pre-developed solution are highly standardised. A data centre can be up and running in weeks and the commissioning phase can be very short. This makes expansion easy, but can also make changing system functionality or meeting new client requirements harder. A flexible phased design allows greater freedom to accommodate specific client requirements and future changes in equipment, TIER levels and so on.
Designing and specifying a system
It is important to take a number of questions into consideration, related to the intended usage and type of DC:
- What type of applications and functionalities do we need to support?
- What are the minimum and projected requirements related to bandwidth, storage and traffic?
- How much time until the DC needs to be up and running?
- Is the DC located close to customer? Is it an edge facility?
- Will it be used by a single customer or multiple customers?
- Is this a greenfield or brownfield development?
- Which containment measures and factors need to be taken into account?
- Which considerations are important for the ‘non compute’ aspects, such as power, cooling, connectivity, DCIM, and housing?
- How sustainable is the complete solution?
It’s also vital to have a clear roadmap including, but not limited to, the following steps:
- Installation planning
- Acceptance testing
- Operations
- Troubleshooting
- Maintenance
Depending on the relevant business and application areas your data centre needs to serve, the answer to these questions and considerations will be very different. This considerably adds to the decision-making complexity…
A factor that significantly contributes to complexity is the fact that not all DCs are created equal – not just with regard to scale and equipment choices, but also in relation to the vertical market they serve. The type of vertical is an important factor when it comes to customisation that allows individual applications to run. For example, for a Hyper or CoLo DC, infrastructure underpins their entire business model, whereas for Financials, being able to support (near-zero latency) financial products and link to stock exchanges is key. Telcos need to support an ever wider range of requirements with their DC infrastructure, from enabling IoT and 5G to providing data transport and communications. Governments need to provide DC support for everything from administration and security processes to highly demanding Smart City applications and, increasingly, autonomous traffic. In healthcare, requirements can range from safeguarding and exchanging patient data to providing a foundation for digitally supported treatment, even remote surgery. Each of these applications introduces a very different set of priorities and requirements. There’s no ‘one size fits all solution’, and that’s a perfect opportunity for modular solutions.
However, you will need to specify your required solution in great detail and take many overlapping and interacting factors and considerations into account. As DC design involves numerous and ever-increasing types of expertise, covering ‘computing’ as well as ‘non-compute’ components that all affect each other, a multidisciplinary approach is highly advisable.
‘Rainbow team’
When specifying and building a modular solution, dealing with multiple suppliers is practically unavoidable. It makes sense to have one single representative manage all of the different parties and ensure consistency, compatibility and interoperability of processes and products.
It’s also important that every single discipline involved in creating the data centre works very closely together and exchanges information in a highly structured, standardised manner. Each individual competence area is currently highly specialised and rapidly developing. Customisation for example, has become too complicated be carried out by anyone but experts who understand all of the dependencies and variables.
By assembling a ‘rainbow team’ of experts from all relevant disciplines, and appointing one single point of contact for client interfacing, a third party can take care of optimising the modular solution, regardless of the number of variations and product types, and the resulting complexity. The members of the rainbow team can come from inside one company, but could also come from different third parties – the main requirement is that each member complements the other members and possesses up-to-date knowledge in their field, and can quickly develop a ‘helicopter view’ of a given project, incorporating knowledge from other fields and finding synergies.
For example, by taking a ‘rainbow team’ approach, computing and non-compute elements can be made to work together to save energy and water for example, and score high on KPIs such as efficiency, effectiveness, and sustainability. Time to fault finding and repair can be vastly shortened and an overall higher quality level achieved, which supports competitive positioning. This can all be realised in a completely technology – and vendor – agnostic way.
We safely conclude that a modular approach can cut down the costs normally associated with building and certifying a green DC while delivering long-term cost savings on expansion, operation and maintenance. This does require some very elaborate and often potentially complex decision-making. All in all, a ‘rainbow team’ approach with a single point of contact provides greater knowledge, focus and flexibility for selecting exactly the right solution for each customer.
