Managing capacity in a world of relentless data generation

By Nigel Stevens, Managing Director for IO in the United Kingdom.

  • Monday, 24th August 2015 Posted 9 years ago in by Phil Alsop

It is an understatement to say that the digital universe – the amount of information processed online – is growing big, and fast. We now generate as much data in two days as was generated between the dawn of civilisation and 2003. There is no indication of this wave abating: The research firm IDC has suggested that by 2020, the amount of information processed online will have grown tenfold from its size in 2013 – from 4.4 zettabytes to 44 zettabytes.
With this data explosion comes increased demand for data centre capacity – demand that many data centres are not equipped to accommodate. According to the analyst firm Forrester, “Many enterprises are faced with data centres that are obsolete or radically under-powered.”


What is the best way for IT leaders to respond – to meet their organisations’ capacity needs today, and into the future? The short answer is with a data centre solution that matches infrastructure density to application requirements and can scale to meet changing demands. Complemented by data centre infrastructure management (DCIM) software that enables capacity planning.


Matching infrastructure density to application requirements
Some applications, such as high-performance computing applications like those used in financial services, for example, require high power density. Other applications, data storage, for example, are well suited to low power densities. Many organisations respond to that kind of mixed demand by building their data centres to accommodate the highest density needs. But that means they’ve over-provisioned for some applications, which is wasteful.
As my colleague Patrick Flynn, Group Leader of Applied Intelligence at IO, explains, “If you’ve over-provisioned your data centre, then you’re underutilising capacity. And that’s incredibly inefficient. After all, energy overhead is fairly fixed, so if you can produce more useful work, you can get a thinner spreading of energy overhead. Higher utilisation equals greater energy efficiency.”


A mixed-density data centre gives organisations the ability to match infrastructure to application requirements. So applications with low power density requirements can be deployed on raised floor (capable of supporting 3-5 kW per rack). And applications with higher power density requirements can be deployed in modular data centre infrastructure (capable of supporting up to 30 kW per rack).


When that mixed-density data centre also offers colocation customers access to an on-premises private cloud, then there are even more options for matching applications to the most suitable infrastructure. That “hybrid” solution – making use of data centre modules and cloud – is particularly ideal for organisations with some workloads that are predictable and others that are not (the variable workloads can burst into the cloud).


Scalability to meet changing demands
So a mixed-density data centre with an on-premises private cloud gives organisations the ability to deploy applications on the infrastructure that best suits the application, and in so doing to maximise utilisation. But still important is the ability to scale capacity as demands change. Many of the features of a mixed-density data centre also enable scalability – both vertical scalability (densification) and horizontal scalability (expansion).


Modules can accommodate much higher densities than traditional raised-floor data can. In a raised-floor data centre, then, organisations more quickly reach the power density limit – and then the only way to scale is to expand the footprint. A modular data centre, in contrast, accommodates up to six times higher density, enabling organisations to scale without expanding the footprint.


Of course, even the highest density data centre still has a maximum number of racks it can accommodate. So what happens when the organisation reaches that level? It has to expand horizontally – that is, expand its entire data centre footprint.


If the organisation is in a data centre with only traditional raised-floor capacity, data centre expansion can take many months or even years. By contrast, new capacity in a modular data centre can be deployed in days. In the case of a global investment bank, for example, the time to deploy modular capacity in an existing data centre was 67 percent faster than deployment of the company’s traditional raised-floor capacity.


Capacity planning
Being able to match infrastructure density to application requirements and to scale, vertically and horizontally, as demand changes are essential pre-requisites to managing capacity in a world of relentless data generation. But how does an organisation know when and where new – or different – capacity is needed?


Data centre infrastructure management (DCIM) software in concert with predictive analysis helps organisations more efficiently allocate current resources, identify capacity bottlenecks, and anticipate future resource needs. Providing a single view of the organisation’s data centre assets, even across multiple data centres, DCIM software enables organisations to optimise their decisions. As Gartner analyst Jay Pultz explains, “We get new equipment and say, ‘Here’s a place we can put this,’ but it’s really about what’s the best place to put this new equipment.”


It is clear that today data centre capacity needs are no longer static. Demand for capacity is rising fast, and that is not going to change. So data centres must offer deployment options that allow organisations to match infrastructure density to application requirements. To scale – up and down – as capacity demand fluctuates, and to have full visibility into capacity across all data centres and the ability to optimise capacity planning decisions.


9 Questions to Ask About Data Centre Capacity Management
1. How does the data centre enable me to match density to application needs?
2. Does the data centre support mixed densities?
3. Does the data centre offer an on-premises private cloud option?
4. How does the data centre enable me to scale as my demand for compute capacity changes?
5. What is the maximum watts per square metre (or kW per rack) the data centre can support?
6. How many square metres of unused space are available in the data centre?
7. What is the total capacity available at data centre (# MW and M2)?
8. How long does it take to deploy new IT infrastructure?
9. How does the data centre enable capacity planning?