My technique nowadays tends to lead people gently into the whole topic.
The analogy of building a Biplane or a Jumbo Jet is a good one and can be extended quite a long way. You can get across the English Channel safely in both, but you could only take a few people in a Biplane and the safety levels are not as good as in a Jumbo. A Jumbo tends to be faster but limited in where it can land, it tends to be more robust but costs much more to build and maintain. I'd happily cross the Atlantic in a Jumbo, but would be much less happy in a Biplane.
Once you've used this analogy to lead the client gently into thinking about the robustness and availability of the system, I then tend to get them thinking about the cost implications of the various levels of Service that the system could be offering. Point out that 99.999% availability would cost several tens of millions and do they really want to spend that much to build a system that would guarantee only a few minutes of downtime in a year? When you frame it that way they tend to become more amenable to 99.9% or less availability. It is also worth coming prepared with the staffing costs for a 24/7 system, they often realise that the system is only needed 5 days a week and for only 12 hours a day...
Until you know the Disaster Recovery and Availability requirements for a system you can't architect or design it. However you must be careful in asking the right questions so that your clients think thoroughly about what they need to be.
No comments:
Post a Comment