Target infrastructure sizing study

What I do

• —collection of real workloads over a representative period (LiveOptics or equivalent): average and peak CPU, memory consumed, IOPS, latency, data volume
• —application mapping machine by machine: environment, criticality, operating system, database, recovery requirements
• —cross-referencing of the collection and the mapping, with verification of the match rate
• —recalibration of allocations from actual consumption, graded by criticality
• —estimation of the resources required by upcoming projects, component by component
• —construction of growth assumptions and margins, with separate horizons for production and non-production
• —compute sizing under three constraints — peak performance in degraded N-1 mode, vCPU-to-core consolidation ratio, memory occupancy
• —storage sizing: raw, usable and effective capacity after data reduction, with a ratio weighted on the actual composition of the estate
• —sizing of backup and of the recovery policy by criticality level
• —analysis of processor density scenarios and of reuse of existing hardware
• —assessment of the impact of the sizing on licensing (Oracle, VMware)
• —sensitivity analysis of the assumptions and risk register

Deliverables

• —sizing report: findings, method, results, target architecture and decisions required
• —complete and editable calculation model, from per-machine detail through to the summary — the client can vary an assumption and watch the result move
• —expected functional specifications of the target components, directly reusable in a requirements document
• —documented assumptions with, for each one, the effect of a variation on the result
• —risk register and compared scenarios
• —simulation of degraded mode following the loss of a node

What changes

you negotiate with integrators on figures that are yours, and you can justify every euro requested — including the margin — before an investment committee.