DCIM as the lingua franca of the data centre


2 November 2015
Gavin OReilly of UBS

Will DCIM (data centre infrastructure management) become the lingua franca of the data centre? Gavin O’Reilly, APEC regional head of Data Centre Services, UBS AG, is optimistic.

In a presentation at Data Centre World, O’Reilly re-cast the men-are-from-mars quandary in the data centre context, highlighting the problems that beset communications between the facilities, infrastructure and application teams. “They have different sets of technologies, different terminologies, different planning an forecast timelines. The product lifecycles are completely different, but they co-habit the same ecosystem,” said O’Reilly.

Application teams, for example, talk about new release cycles in terms of months; infrastructure teams could take two years to spin up a new data centre; while facilities teams are more used to working with a 10 to 15-year data centre lifecycle. “There is going to be a certain amount of tension between the infrastructure guys and the facilities guys when planning for the future.”

To illustrate with a scenario: The facilities team asks for the kilowatt rating for each rack in order to plan the loading, but the IT environment is not static, so the figure provided by the infrastructure team could change dramatically in a matter of months. This could be difficult for the facilities team to adjust to as they tend to have a longer-term mindset.

The reality is that within the data centre, the facilities team, infrastructure team and applications team have largely been operating in silos, and quite a bit of translation is required between the separate stacks.

But, said O’Reilly, things are changing. “We are seeing abstraction, virtualisation, consolidation and convergence at all levels within the IT space. IT infrastructure is changing at a pace that we have not seen before.”

And with virtualisation, there has to be a lot more communication between the three teams. Facilities will need to cater to the demands of Infrastructure; Application needs to know how the technology infrastructure works to manage failover and resilience. There has to be a greater degree of communication than in the past, he said.

“The question now is whether DCIM can provide the platform that allows us to have a common language that can be used to facilitate these conversations.”

DCIM has the potential to provide a single plane of glass into the data centre, giving visibility into work flow systems, asset tags, building management systems, cable management systems and IT hardware. For example, it will be able to get data directly from a server (for example, data on CPU utilisation, input-output operations and inlet temperatures), and monitor and manage power utilisation. During weekends when the server is not utilised, it can be put into idle mode. Potentially, all these can be measured, communicated and controlled through DCIM, said O’Reilly.

“From an infrastructure and facilities perspective, it allows you to model your data centre. You can model changes, failure scenarios and growth. And this can be done in conjunction with the three teams.”