RECAP | ROUNDTABLE | Modern capacity plans

Introduction: Data center owners and operators have long struggled to develop accurate IT capacity plans. Changes in technology, density and business plans constitute a large part of the challenge. While increases in individual server compute power and increased deployment of virtualization strategies seem to argue for a decrease in requirements for data center space, power and cooling capacity, the simultaneous increase in new software applications and demand for IT services make it unlikely that the capacity demand curve will flatten any time in the near future. Fred Dickerman, Uptime Institute senior vice president management services, joins this Inside Track roundtable discussion, during which participants will share their approaches to capacity planning.

Roundtable Wrap-up:

The Inside Track July 8, 2020, roundtable on Modern Capacity Planning was attended by company representatives from financial, healthcare, and colocation providers. Attendees were eager to hear what other companies were doing in the way of capacity planning.

Fred Dickerman, VP Management Services for Uptime Institute, started the discussion by saying that maximum return on investment determines how you manage capacity. He also mentioned that communications issues still exist between IT customers, Capacity Planners, and Facilities causing forecasting to be an issue. Companies are creating plans that go out 3-5 years, but they become outdated within months.

The enterprise attendees all indicated they have what they would call mature and sophisticated capacity planning processes. They are managing the constraints caused by space, power, cooling, and in some cases network as well. They create plans that are in a range of 3 to 5 years, and manage to them monthly. Some are managing in an IT growth environment, and some are in a consolidation mode, but the processes are in place to manage capacity and balance risk.

The challenges and difficulties to capacity planning, as stated by the attendees:
• Ability of IT to forecast is not sufficient. Plans go out 3-5 years, but are barely valid a year, and sometimes much less than a year. This means the capacity plan relies on trending, which can end up not being very effective.
• The rapid changes in IT technology lead to issues in managing asset life-cycles, which lead to ineffective capacity planning.
• It was stated the COVID-19 pandemic has caused hardware lead time increases, which can cause capacity planning issues.
• As workloads are virtualized and moved to the cloud, managing IT services is now more difficult and requires balancing asset placement to balance application risk. An attendee discussed their detailed business workload placement policies around risk and application resiliency. Another attendee indicated they follow a similar approach.
• Hardware having different depreciation schedules creates a challenge. Because of this there is constant hardware refresh and churn to stay ahead of hardware end-of-life.

In regard to colocation, the feeling was that colos tend to be on top of capacity planning because it is critical to their livelihood. Without capacity to sell, there is no added revenue. The attendees indicated they are concerned more on colo physical security and access.

Finally, an attendee brought up how they have installed modular self-contained capacity as a strategy to quickly add high density computing capacity when needed to an existing data center site. The modular data hall containers are self-contained and house Dell and HP equipment. The issue being experienced is that the hardware is rusting and corroding. This would include corrosion of the internals to the hardware (motherboards, etc.). It is an interesting issue and will be followed up more with the attendee directly.