In order to measure the performance of a data centre, we require Key Performance Indicators (KPIs). A KPI is a measurable value that demonstrates how effectively a company is achieving key business objectives. Organisations use KPIs to evaluate their success at reaching targets.  Practically therefore, KPIs are designed to benchmark the performance against a set of metrics and depending on the outcome, it will provide valuable information for (senior) management for analysis and decision making purposes.

The EPI-DCOS® (EPI-Data Centre Operations Standard) provides a progressive improvement approach covering all 11 disciplines in a fully managed data centre, where each discipline can be benchmarked on its own maturity level. In simple terms, it means that the organisation considering adopting EPI-DCOS® has the freedom to select individual disciplines which they feel has a need for improvement, and for that discipline, choose a maturity level. There are five maturity levels in total which are based on the well-known ISO/IEC-15504 standard:

1. Initial
2. Repeatable
3. Defined
4. Managed
5. Optimised

Although important, let’s for now forget about the maturity levels and focus on what is important to reach the desired level of maturity, namely metrics and KPIs.

Every clause in the EPI-DCOS® is in principle a metric.  In simplified terms, metrics define what is to be measured in order to understand how well the organisation is performing.

Let’s look at an example based on the EPI-DCOS®:

18.2.1. of the standard mentions the following (standard extract):

‘The facilities monitoring matrix should detail for each piece of equipment which points are monitored. The organisation should have a justification document defining what is required to be monitored and why.’

18.3 of the standard goes into further detail (standard extract):

‘All monitoring matrices should be linked to the notification system. The notification matrix links alarm events to the operations teams ‘event response plans’ which indicates what actions to take and what mode of notification is required as well as the contacts to be notified.’

16.4.1 of the standard maps it into ‘recording of incidents’ (standard extract):

‘Each and every incident should be recorded. Incident generation as well as recording could be handled manually or with the use of a fully automated monitoring system feeding incidents into the incident management system’.

--

With this basic information I have established 3 metric points on which I can start designing my KPIs, such as, and not limited to:

1. The number of incidents not reported by the monitoring system
2. The number of incidents which require manual interference
3. The number of incidents which require escalation
4. The number of incidents not resolved in time
5. The number of notifications versus number of operators per shift
6. The number of notifications versus the type of operators per shift
7. Notification - time to response

The list of KPIs(*) can be almost endless but care should be taken in designing them. For KPIs to be successful, they need to connect to the business and be understandable for the subject matter experts who are involved. It needs to provide context, it must be based on legitimate data and must lead to actions where required, ideally having a (measurable) business impact. One should carefully look at the principle of diminishing returns since not every effort for improvement is worth the business benefit.

Depending on the outcome of KPIs, the data centre organisation will be able to respond wherever applicable and this will demonstrate control, being capable of meeting the business objectives.

With the above example, the organisation adopting the standard will be able to meet the requirements as stated in the clauses, and by integrating these it is capable of demonstrating maturity (to the highest level 5 – Optimised), thereby, delivering added value to the business.

--

(*) With a growing number of KPIs, it is recommended to consider some level of automation. There are many vendors offering systems for managing and monitoring KPIs, often in real time, and these solutions are available on both cloud and non-cloud platforms.

Contributed by, Jan Willem Mooren, EPI

Questions?  Email us!