(*) In IT, resilience is a system's capacity to function even in situations of breakdown or extreme stress. Fair performance consists of implementing adapted measures and resources without having to over-dimension them.
What it's about: The IT sector is governed by the "always on" principle as well as the race towards performance. By taking into account the green dimension in service level agreements, a better definition of reasonable high availability thresholds can be obtained.
The IT sector must confront all potential issues (breakdowns, peaks in usage, and resumption of operations, etc). In order to manage this, it must have resources available for use other than those used at "ordinary times". Availability problems have lead to the introduction of “standby” systems at all levels of industry, which are kept ready to use in case of breakdown to keep periods of out-time and recovery time to a minimum.
Issues: Availability stakes encourage the creation of redundant architectures at all levels, as well as their maintenance on standby, to minimise out-times in case of breakdowns. Ecological costs and impacts can be disproportionate in comparison to these stakes. This must be taken into consideration right at the initial SLA definition stage.
This rule leads to several key questions:
Some of the main conditions when planning a hosting project, at the core of the contractual commitment between the service provider and their client, include making availability stakes, plans for resumption of activities after breakdown, and maximum capacity levels, transparent.
The “always on” culture is sometimes aimed at providing the hosted service maximum possible availability. These stakes must be compared with their carbon footprint, in order to avoid unnecessary energy wastage (eg: redundant systems which are still available for use have a significant carbon impact.)