All innovations in the field of IT production (practices, tools, infrastructures) meet 3 needs of IT departments: cost reductions, security and flexibility.

Of these 3 needs, security is certainly the one that most concerns production managers and meaning the availability of the infrastructure and applications. Several "systems" contribute to improving availability in the Data Centers, systems that are structured as follows:

These needs are prioritised (in a similar way to the famous Maslow pyramid): unless for specific goals (such as flexibility ....), we will not launch a Security project for example by starting a virtualisation project in old Data Centers which are many and need renovating..

The need for secure hosting has revived and revitalised the market for "data rooms for co-location" by adding value to it: facing growing energy issues, , the increasingly complexity of the field (management of communication links, virtualisation, increasing growing density...), many outsourcing providers have invested heavily in value-added hosting solutions: Bull thus renewed, in early 2012, a hosting contract for 6 years with La Poste (French postal operator), thereby guaranteeing it the very high availability necessary for its continuity of service needs for La Poste Group's strategic activities. The concentration of more than 1,800 servers previously hosted in over 8 data centers limits risks by improving control of infrastructure which has been relocated in 2 state-of-the-art Bull data centers (one DC in Dual Building mode): electrical chains and dual, crossed and recent cold chains, the most advanced techniques in the urbanisation of the rooms (cold and hot aisles, cold corridors), cooling systems (chilled water doors, free cooling...). Management of the machine base, hardware, cabling and racking also allows improved change management and a reduction in the identification and resolution times in the event of incidents.

Through secure hosting services, outsourcing also returns to one of its fundamentals, the mutualisation , which allows costs to be shared, to benefit from mutual experiences and from the capacity for innovation and continued investment by an industrial player specialised in the field. It is only when it is relocated and fully managed that one infrastructure can undergo the second step: consolidation. This consolidation of physical servers has been made possible by the standardisation of the equipment and new technologies (blades, multiplication of computing power and storage, modular structures...). Consolidation improves the rate of use of equipment, optimises licence and maintenance costs and facilitates the management of capacities. Consolidation also contributes to improved availability because due to a better control of limited numbers of more uniform and standardised devices. Care is needed however to strengthen the recovery plan (DRP/BCP) since consolidation also concentrates risks!

We will then derive the full benefit of a new modern and consolidated infrastructure by deploying virtualisation solutions: These natively include the principles of availability since a virtual machine can be in a dual state (active and awaiting activation) and moved from one infrastructure to another to take over in the event of an incident. Finally, automation and "Suites for Cloud" will bring a real industrial dimension to IT production:

• Service catalogues, portals and workflows will channel the demands of users and limit uncontrolled and unanticipated changes.
• The use of library objects (infrastructure stacks, application components) ensures that the products have been subject to a qualification process upstream.
• The automation of provisioning accelerates deployments and, in particular, allows manual intervention to be reduced which is the main source of incidents.
• Orchestration automates the integration of VMs into the Data Center's Information System and ensures that a newly created VM is properly backed-up, monitored, secured...
• Configuration management and capacity planning have a significant impact on the availability of applications: the configuration management process limits non-compliant configurations to standards, reduces downtime in the event of incidents since a status report of differences between the failed configuration and the reference one is automatically prepared which enables fast reversal to the previous configuration. Capacity management warns about over-use of the servers and optimises their use.
• Automated management of SLAs will improve the visibility of the behaviour of the Data Center, the infrastructure and applications and will support an active and continuous improvement approach. We are at the beginning of a real industrial revolution that will overturn our practices and enable new uses for IT still unknown today: this revolution has started in the basements of our Data Centers and is continuing in the upper reaches of the Cloud.

We are at the beginning of a real industrial revolution that will overturn our practices and enable new uses for IT still unknown today: this revolution has started in the basements of our Data Centers and is continuing in the upper reaches of the Cloud.