Data centres should no longer be on the CIO’s dashboard

We’ve been kicking around building blocks of the ideal ‘business model’ for data centres.  The exercise has taken us through provider reviews, the role of location, marketing and how strategic business model considerations remain low priority. We believe part of the ongoing challenge lies in what the CIO measures. What you focus on grows in importance. By changing how data centre capacity is framed in team dashboards, CIOs can steer progress towards better data centre business models.

Management bias: production versus consumption

In most enterprise data centres today, consumption forecasts are made by the same team that plans for production. This inherently creates the tendency to trade off demand management needs against service delivery realities. We think there are lessons from other industries that are worth adopting.

Take oil and gas for example. Upstream activities (exploration and extraction) and downstream (refining and distribution) are now quite distinct, and in many cases, separated by the open market. This means oil extracted by Company A is sold to the market, and Company A’s downstream operations will buy from the market for its refining needs. This leaves market efficiencies to manage the costs of physical distribution and matching demand with supply.

The same is true in the finance sector. Many banks have separated their deposit taking business management from their lending businesses. It is not unusual for a bank to place deposits it holds on the money market in return for the best prevalent rate. Its lending business in turn goes to the open market to raise funds necessary to meet the needs of its loan order book.

We believe the time has come for CIOs to view data centre capacity along similar lines. Separating out the management of consumption from that of production. Hang on, we hear you say, isn’t that what clouds are all about? Well, yes. BUT, we have yet to see the promise of technology filtering through to related roles, responsibilities and dashboards.

What could be

IT governance and associated dashboards come in many shapes and sizes, but on the whole they cover three aspects: what business lines expect from IT, what IT processes intend to deliver and what level of efficiency is expected within the process cycle. The majority of enterprise IT dashboards we have reviewed translate business expectations into services and technical requirements and immediately assign producers of these services.

Managing consumption demand has to be more than the automated translation into capacity. Workload analysis should lead to evaluation of best provider of capacity. Even when there is an internal data centre, objective comparison of price performance needs to occur. For example, enterprises that have engineered in-house data centres to handle real time workloads might find it makes more sense to use an external provider for archiving.

Segregating duties is the best way to ensure control and promote transparency. Separating consumption management from data centre production management will be the first step towards de-constructing the traditional IT dashboard. This separation will shed light on alternative options and help define a more focused role for in-house resources. Defined metrics should show a good ratio between cost and relevance, allow for internal and external comparison and be easy to measure.

Skills and experience to manage transparency

Arguably the main barrier to such a separation lies in the lack of skills. Most managers in the data centre world have risen through the ranks and are predominantly concerned about production matters. Few have any experience in demand management. CIOs who have attracted or developed data centre demand managers have moved on to remove the data centre from their overall dashboard. Instead, they monitor workloads and trust in their fulfilment engine which may, or may not, include an in-house data centre.

Image credit: Observation Deck Stanserhorn by Ivan Rigamonti

This post first appeared on the old Cassini Reviews website.