ROI & costs

Optimising cloud costs in data projects

The most common causes of cloud overspend in data and the techniques to cut it without losing performance: sizing, formats, queries and frequencies.

DLData Layer Team Mar 29, 2025 4 min read
Optimising cloud costs in data projects

Key takeaways

  • Cloud overspend in data usually comes from idle capacity and inefficient queries.
  • Optimising sizing, formats and queries reduces the bill.
  • Matching refresh frequency to real need saves a lot.
  • Optimisation must be continuous, not one-off.
  • It cuts spend without losing performance.

The cloud makes scaling easy, but also makes overspending easy without noticing. In data projects — compute- and storage-intensive — overspend is especially easy to accumulate. The good news: much of it is avoidable.

What it is

Cloud cost optimisation in data is the set of practices to reduce infrastructure spend without sacrificing performance or reliability, matching consumption to what the business actually needs.

The causes of overspend

The saving levers

  1. Right-size and shut down idle resources.
  2. Optimise queries and partitioning.
  3. Use compressed columnar formats.
  4. Adjust refresh frequency to real need.
Measure
Attribute spendper process
Optimise
Right-size, tuneRemove idle
Review
Repeatperiodically
Cloud cost optimisation is a continuous measure-optimise-review cycle.

Continuous optimisation

Overspend is not eliminated once and for all: workloads change, data grows and new inefficiencies appear. In a managed service, this watch and adjustment is part of the operation, so each process consumes the minimum necessary.

Optimisation is not a one-off cleanup but a continuous cycle, or savings erode as workloads grow.

In summary

Cloud overspend in data comes from idle capacity, inefficient queries, heavy formats and excessive refresh frequencies — most of it avoidable. Right-sizing, query and format optimisation and right frequencies cut the bill without losing performance, but only if done continuously.

Sources & further reading

Frequently asked questions

Why does the data cloud bill spike?

Idle capacity, inefficient queries, heavy formats and excessive refresh frequencies. Most is avoidable with optimisation.

Does optimising mean losing performance?

No. Done well, it reduces spend while maintaining or improving performance, matching consumption to real need.

Is it a one-off effort?

No. It must be continuous, because workloads and data change. In a managed service it is part of the operation.

What are the main saving levers?

Right-sizing and shutting down idle resources, optimising queries and partitioning, using compressed columnar formats, and adjusting refresh frequency.

How much can I save?

It varies, but removing idle capacity and tuning queries and formats often cuts a significant share of the data cloud bill.

Who handles optimisation in a managed service?

The provider, continuously, as part of the operation — so each process consumes the minimum necessary without you watching it.

Turn this data into results

Tell us what you want to achieve. Data Layer connects, processes and delivers the result up and running, with no infrastructure for you to manage.