Optimise the resources of a data centre

Last modified by the author on 14/02/2018 - 09:13
Optimise the resources of a data centre

Projecting the building needs of a data storage company and finding ways to postpone new building work.

Dissertation summary - Certificate of Advanced Studies in Energy Management - 2017

Context of the project

Infomaniak currently operates 2 data centres in Geneva. The first was opened in Vernier in 2008. There is limited space available in the racks.
The second was opened in Satigny in 2014. There is a lot of available space and technological improvements such as miniaturised server components and a suitable building have led to improved energy efficiency, with lower power consumption and a reduction in the ground area required, or footprint.

Description of the project

As cooling has already been optimised (which has reduced the power consumption), the room for manoeuvre remaining to the company is to focus on the servers.

Here are the objectives of this project:

Calculate in how long a new data centre should be built:

   1. as things stand (at the current rate of growth), and;
   2. by proposing the progressive replacement of different servers, determined for the occasion. An analysis of both solutions was proposed, taking into account the following main criteria:

- the available space in each data centre and the average power consumption of each type of server;
- the needs of the company (estimating the increase in purchasing of servers and consequently the increase in space used according to the frequency adopted and the proposed changes of servers);
- an estimate of the date when the data centres will be full according to various developments mentioned above;

This project was evaluated as part of the company's energy management scheme, using the following plan:

Start, Step 1:

Inventory (business model, simplified audit)

Plan, Step 2:

Performance assessment

Energy balance (schematic diagram, separation of consumption according to uses)
Reference consumption
Identification of indicators
Comparative performance analysis and benchmarking

Plan, Step 3:

Action plan

Definition of objectives and targets of the action plan (roles and responsibilities, means & deadlines)
Description of the three chosen energy performance actions (selection criteria, financial analysis, non-energy benefits)
Measurement method and verification
Incorporating these EPAs (energy performance actions) into the existing Action Plan

Perform, Step 4:

Implement the action plan

Technical production file and operational control
Planning (Project Gantt)
Communication
Training

Check, Step 5:

Evaluation of progress and rewards

Evaluation of the improvement in performance
Monitoring and targeting (data collection, data analysis, reporting, action)
Legal compliance
Recognise and reward success

Act, Step 6:

Correction and reassessment for continuous improvement

Identify and treat nonconformities
Review the management system (Management review)

Conclusion

Given the high cost of building a new data centre, optimising the latter’s resources becomes a strategic issue that a company like Infomaniak must take into account in its investments. We have been able to translate this resource optimisation into this report, particularly in terms of the use of available space (server footprint): we have been able to determine when the Infomaniak data centres will be full. We have also identified key success factors for the company and key energy services related to Infomaniak's value propositions. The EPAs that we were able to identify have been incorporated into the company's existing action plan.

Calculating the non-energy benefits of each proposed EPA, which had not been done by Infomaniak in the past, will allow the company to more accurately calculate the benefits related to the EPA investments proposed in this report, but in a broader framework to all future EPAs that the organisation will set up. The company will, for example, include the cost of building its Satigny data centre in the next EPAs that it will set up. In this way, it will be able to take into account the non-energy benefit of “reducing the footprint” of the servers in its data centres. But this is not the only non-energy benefit that has been identified in this report: reduction of maintenance costs, reduction of the footprint, improvement of existing services.


Finally, we have demonstrated the point of setting up these EPAs, but above all the point of setting up an energy management system in an internet data storage company, not only for the sake of saving energy, but also to optimise the resources of its computing centres.
The energy management system makes it possible to better manage all the resources of a data centre and contributes to the business model of the company.

Share :
3912
Author of the page

Alexandre Patti

Gestionnaire sécurité, énergie et environnement