Microsoft Fabric and Azure data platforms in Cambridge: architecture, capacity and cost

The question we are asked most often by Cambridge organisations is whether they need Microsoft Fabric. The answer is frequently no, and saying so has cost us work we did not want.

This page is about building a data platform for an organisation in Cambridge, Microsoft Fabric, Azure Synapse, Data Factory and the lakehouse patterns underneath them, with the local economy taken seriously. Cambridge's base in software and technology, electronics and semiconductors, and pharmaceuticals and biotech determines the data volumes, the latency requirements and the retention obligations, and those three things decide the architecture. The client population here is shaped by a globally significant semiconductor design cluster and biotech and pharmaceutical research.

Does a Cambridge organisation actually need Fabric?

Fabric starts to earn its keep when several of the following are true at once: volumes that strain an import model, a genuine need for both SQL and notebook access to the same data, several teams requiring governed self-service, retention obligations that make a desktop model impossible, or an existing Premium capacity. One analyst maintaining three import models that refresh in twenty minutes does not need a capacity, and buying one will not improve anything.

What the data actually looks like in Cambridge

Software and technology

Product event streams at very high volume alongside subscription, billing and support data that must be joined at account grain to be useful. Source systems typically include billing and subscription platform, CRM, product event pipeline, support platform and finance system.

For a platform design, the consequence is specific. ARR comes from a finance spreadsheet with manual adjustments, so restating a prior period requires archaeology and diligence becomes painful. So the architecture has to make the raw grain replayable rather than aggregating at ingestion, and the first serving model should be an effective-dated contract and movement model so ARR is reproducible for any historical date, then product events joined at account grain.

Higher education and research

Student lifecycle, research grant and estates data in separate systems, with statutory returns and league-table submissions on fixed annual cycles. Statutory returns and internal management reporting are produced through separate routes, so published figures and internal figures disagree.

Systems to plan for: student records system, research grant management, finance system, virtual learning environment and HR. Student data is personal data with additional duties around special-category attributes, and research data may carry funder-specific restrictions.

Electronics and semiconductors

Very high volume test and yield data per die or unit, complex multi-stage processes, and component supply with long and volatile lead times. Test data is aggregated at ingestion, so when a field failure pattern emerges the unit-level evidence no longer exists. Export control classification applies to many components and to the associated technical data.

A medallion layout for a software and technology business

The medallion pattern is not complicated, and it is routinely implemented badly: a bronze layer that has already lost fidelity, a silver layer with no testable rules, and a gold layer that is really just the silver layer with different names. Here is how we would lay it out for a Cambridge client in software and technology.

Capacity sizing and what it really costs

Fabric capacity is billed by capacity unit. An F2 is roughly US$0.36 per hour pay-as-you-go, about US$260 a month, with around 40 per cent off for a one-year reservation. An F64, the smallest SKU at which report consumers no longer need an individual Pro licence, is roughly US$8,400 a month pay-as-you-go or around US$5,000 reserved.

The arithmetic that matters: F64 only becomes cheaper than per-user licensing somewhere in the region of 350 to 600 readers, depending on the discount you actually pay and whether you reserve capacity. Below that, Pro or PPU per user is normally the right answer, and buying capacity for prestige is a straightforward waste of money.

The sizing exercise is empirical: profile each workload, separate background refresh from interactive query, count real concurrency, find the true peak, then prove the number on a short-lived capacity before anything is bought. The output is a sizing model that states expected utilisation at launch, at double the readers, and the specific point at which you would need the next SKU. Always price from the current Microsoft rate card for your region and agreement. Any figure in an article older than a quarter should be treated as an order of magnitude, not a quote.

Direct Lake, import or DirectQuery

The appeal of Direct Lake is that nothing is copied and nothing is refreshed. The risk is the fallback: one unsupported construct and the model quietly behaves like DirectQuery instead. We test it against an import model on your own data and choose on the measured result. In practice Direct Lake tends to win on large, frequently landing tables, which software and technology workloads often are, while import still wins on smaller models with heavy DAX.

Migrating from what you have now

Few Cambridge organisations start from nothing. The realistic options are a serving-layer-first adoption, where Fabric takes the gold layer and reporting while existing engineering stays in place, or a domain-by-domain migration with the old platform retained until each domain is validated. Neither requires switching everything at once, which is the approach behind most of the failed migrations we are asked to rescue. The reference pipeline we build first is usually the system that hurts most, which here is normally billing and subscription platform.

Residency, latency and the United Kingdom question

Azure UK South and UK West give genuine in-country data residency, and Power BI tenants can be provisioned in the United Kingdom region. Note that some workloads, certain preview features and a handful of AI services, still process outside the region, so residency claims should be checked per service rather than assumed tenant-wide.

For Cambridge specifically, UK South (London) is the normal primary with UK West (Cardiff) as the pair. Latency from Cambridge to that region is not usually a constraint for analytical workloads; the constraints that bite are feature availability per region and the processor chain your data protection officer will ask about.

• UK GDPR data-minimisation and storage-limitation duties apply directly to analytics copies of personal data, which is where most estates quietly fail.
• Consumer Duty outcome monitoring has made customer-level analytics a supervisory expectation for regulated firms, not an optional capability.
• SECR energy and carbon reporting requires auditable activity data, which usually means the same lineage discipline as financial reporting.

Where to go next

If you are weighing a platform decision for a Cambridge organisation, a workload assessment will answer it with evidence in two to three weeks, and the written recommendation is yours whatever you decide. Book a technical review.

We work across United Kingdom and deliver into neighbouring markets regularly. Nearby pages worth reading if your group has sites elsewhere: Oxford, Dundee, Guildford, Londonderry, Manchester and Bristol.

Frequently asked questions