Microsoft Excel is still a wildly popular data center management tool, and one in 10 data center managers walk around the raised floor with a tape measure, according to results of a recent Intel survey of 200 data center managers in the US and in the UK.
As much as we like to write about data center automation and sophisticated DCIM software in trade pubs and talk about it in conferences, the reality is close to half (43 percent) of data center managers still use manual processes to do their jobs.
This is not limited to small data centers. Manual data center management is still a big hit in facilities that support 1,500 servers and more.
The stats put Google’s use of machine learning in data center management to maximize efficiency in perspective, or, more precisely, in the realm of science fiction as far as the bulk of the world’s data centers are concerned.
But it doesn’t have to be this way.
The most popular explanations for lack of automation are high cost of the tools themselves and the resources needed to deploy them. While those are valid, in Intel’s opinion they “represent false economies in the longer run.”
It’s important to keep in mind that Intel is not a neutral observer here. The company’s Data Center Manager middleware broadcasts device vitals, such as server temperature and power consumption, to DCIM software solutions by vendors that partner with Intel for this purpose, and it’s a business the company has been actively promoting in recent years.
More than half of manual planners surveyed spend more than 40 percent of their time on capacity planning and forecasting every month. From this, the report concludes that many data center managers are “locked in a vicious circle.” They don’t have the time to implement DCIM tools that automate capacity planning because they’re too busy managing manually.
Nearly 60 percent of respondents said they experienced “thermal-related” challenges that impacted operational efficiency in the past year, yet 70 percent of data center managers rely on thermal sensors and spreadsheets to identify hot spots or areas that are overcooled.
This doesn’t mean that’s all they use to manage their cooling systems, however. More than 60 percent use analytics capabilities in DCIM software to help optimize cooling efficiency. Another method is hot-spot audits.
But operators without DCIM are less likely to do hot-spot audits and cannot employ more sophisticated optimization methods, such as Computational Fluid Dynamics. Only 20 percent use CFD simulation to identify problem areas.
DCIM software can also help with addressing space and power constraints and quantifying costs of downtime, according to Intel.
It’s hard to say whether the issues of cost and deployment complexity are going to be resolved to accelerate DCIM adoption. Some newer vendors are offering lower-cost solutions, but they’re not as comprehensive as big DCIM software suites that often require a dedicated implementation team to deploy.
That DCIM tools are useful doesn’t appear to be a controversial point. What remains a big hurdle – and a deal breaker for many data center managers – is the upfront cost.