Data center migration: no room for improvising
Data center migration requires careful preparation, at both organizational and technical levels, in order to limit or even avoid service interruptions. From shaping the migration masterplan to relaunching the servers, what steps need to be followed?
Data center migration: first of all, define target strategies
Relocation, replacement of an aging IT room, new hosting scenario, rationalization of IT infrastructures or even outsourcing: many situations can lead an organization to make the decision to move a data center. As with moving office, the relocation of an IT room cannot be improvised, and must be prepared well in advance by both the ISD teams and the business departments (accounting, production, logistics, etc.).
Data centers represent an increasingly important part (if not all) of a company’s activity. Therefore, even before any ideas for a sequence of events, the very first step consists in identifying those contacts who can provide all the information necessary to carry out this project while minimizing, or even avoiding, a service interruption.
First of all, it is a question of establishing a precise mapping of applications, their interconnections, their level of criticality, time slots for potential shutdown, etc. This is followed by an exhaustive inventory of the hardware (servers, racks, etc.) associated with the operation of these applications. An inventory which, incidentally, provides an assessment of real needs in terms of electrical power and necessary cooling of the target infrastructures. Depending on the criticality constraints and possible stoppage times, the relocation team can then establish priorities and decide on a complete migration or a phased move for a relocation in stages.
Since all the equipment will be de-racked, the relocation of a computer room also presents an opportunity for completely rethinking the layout: reorganization of the racks, high and low-voltage cabling, PDUs and even air-conditioning equipment. This is also the stage at which a decision can be taken on whether to retain or replace equipment.
Ultimately, all of these parameters will have to be considered in order to formulate a coherent scenario, based on the chosen target architecture and depending on downtime constraints and priorities.
Operational preparedness, the essence of the migration project
This first phase of shaping the migration masterplan takes between one and three months depending on the complexity and size of the infrastructure and is followed by the field work, which will lead to the project’s success: operational preparedness.
This starts with the establishment of a provisional calendar, taking into account the various constraints and the possibility of a BCP, defining whether or not it is possible to move the various units, if need be, on working days, weekends or even at night. Not to mention the necessary cooling and rest times for the equipment before and after the move (which can each require up to an hour) and of course the transportation times themselves, depending on how far the equipment has to be moved.
In a very concrete way, if migration to a hosting provider is simply a question of ensuring consumption limits per rack based on weighted server consumption, migrating to a dedicated room amounts to no more and no less than designing a new data center from scratch. This implies specific monitoring in terms of organization and governance: operator, consulting company, DSC, etc.
Everything must be studied and programmed: the mover, the number of trucks needed, the sites (access procedures, whether or not there is a delivery dock, freight elevator, etc.). And, on the future IT room side, distribution, cables (types, length, color, naming), labeling of equipment, network equipment, PDUs, racks, air conditioners, hot and cold air flows in compliance with best layout and operating practices, etc. This must all take into account the technical characteristics of the various IT devices, according to the manufacturers’ standards.
Relocating the data center: a seamless plan?
As long as the previous steps have been scrupulously followed, the physical relocation is undoubtedly the simplest part of the project. Each identified item of equipment must be able to be relocated (and reconnected) almost immediately in the new data center. Objective: this step should be a simple question of implementing the plan established in the preparatory phase, in order to minimize potential risks and hazards.
Once the operation has begun, the machines have been stopped and the cooling times observed, the equipment must be given special attention and be completely protected from the risk of falls or jolts. The use of flight cases and specialized carriers is therefore particularly recommended.
Despite all these precautions, it is advisable to take out ad valorem transport insurance: the unexpected can happen to anyone, and the cost of replacing IT equipment is substantial enough to avoid taking unnecessary risks. Indeed, this insurance is offered as standard by companies specializing in data center relocation.
In all cases, the degree of success of a data center relocation project depends on the project management skills of the internal and / or external team in charge. It is best to engage a team or third-party company who have extensive experience and/or can provide certification demonstrating their expertise.