An estate is rarely bad by design. It becomes bad by accumulation — each decision defensible on the day it was taken, the sum of them impossible to change.
System review · System architecture review · Network architecture review
Identify improvement in architecting components (services, data, app, compute, storage and network) to enable most optimal landscape which helps in composability, open standards, in enabling speed of change
Most architecture reviews produce an inventory. An inventory tells you what exists; it does not tell you why a straightforward product change takes nine months, why the same customer record is authoritative in four places, or why the annual run cost rises faster than the business it supports.
So the review is framed around change velocity rather than completeness. Across services, data, applications, compute, storage and network, the test is the same: does this landscape let the organisation move at the speed its market demands, and if not, precisely which components are holding it?
The estates that age well share two properties. Capabilities are composable — assembled from parts that can be replaced individually rather than a monolith that must be replaced entirely — and the interfaces between them follow open standards rather than a single vendor's conventions. Neither is an ideological preference. Both are simply what makes the next decade of change affordable.
That said, a target architecture the organisation cannot staff or fund is not a target, it is a wish. The recommendation has to survive the client's actual capability and cash profile, which is why remediation is sequenced rather than presented whole.
Every application against business capability, cost, condition and criticality — identifying overlap, orphans and obsolescence.
Where each entity is mastered, how it flows, where it is duplicated, and what that duplication costs in reconciliation effort.
Point-to-point sprawl against managed integration, and whether interfaces can be changed without coordinated release.
Placement, utilisation and sizing across on-premise, colocation and cloud, with the actual economics rather than the list price.
Topology, resilience, latency to the places work happens, and whether the design still matches how the organisation operates.
Architectural exposure — segmentation, identity, privileged access and the assumptions the design quietly depends on.
What is unsupported, what is nearly unsupported, and the deadline each imposes whether or not it is acknowledged.
Availability, recovery, scalability and performance measured against what the business genuinely requires, not what was specified.
A defensible destination, the distance to it, and the order in which the distance should be closed.
Six to ten weeks for a mid-sized estate. The output is a decision paper, not a catalogue.
What decision the review must inform — an investment, a migration, an acquisition, a stalled programme. A review without a decision attached is rarely acted upon.
Applications, data flows, infrastructure, contracts and cost, drawn from records and interviews and reconciled where the two disagree.
Composability, open standards, resilience, security and cost-to-change — applied as tests with evidence, not as adjectives.
Which components genuinely limit change, distinguished from those that are merely old. Age is not a defect; inflexibility is.
A target state sized to the organisation's funding and capability, with the trade-offs of each option stated plainly.
Ordered by dependency, risk and payback, with the first phase specified closely enough to start.
Clients are described by sector rather than named.