Case study · Fintech · Cloud cost
Reducing cloud cost by 50% for Klub, one of India's largest revenue-based financing investors.
Klub · Revenue-based financing
Limited visibility into what drove the bill, and a non-production environment per workload. We assessed it, right-sized it, and shared it.
50%
Reduction in cloud costs
The opportunity
A cloud bill with no visible cost drivers.
Klub had limited visibility into the factors driving its cloud costs, which made identifying areas to optimise difficult-it is hard to reduce spend you cannot attribute.
Underneath that were two concrete drivers: a large number of unused and underutilised resources carrying cost for nothing, and a non-production environment per workload, so development cost scaled with the number of environments rather than with the work.
No visibility into cost drivers
Limited insight made optimisation opportunities hard to identify.
Unused and underutilised resources
A large number carried cost without carrying load.
One non-prod environment per workload
Development cost equalled the total number of non-production environments.
The impact
Half the cloud bill, and a view of where it goes.
Assessment first
A comprehensive cost optimisation assessment identified opportunities across the AWS environment.
Right-sized and resized
Recommendations eliminated unused and underutilised resources rather than guessing at capacity.
Shared non-production
Sharing resources across non-production workloads is where a significant part of the saving came from.
Visibility that persists
Klub can now attribute cloud cost and allocate resources on evidence rather than estimate.
The stack
What it runs on.
- Cost analysis
- AWS Cost Explorer
- Compute
- Instance right-sizing, AWS Graviton instance types
- Cleanup
- Decommissioning of unused services
- Environments
- Shared resources across non-production workloads
The road ahead
Costs are visible. Keeping them down is now routine.
With drivers attributable and non-production shared, cost control becomes a standing practice against a known baseline rather than a periodic clean-up exercise.
- 01
Continuous cost anomaly detection
- 02
Wider Graviton adoption
- 03
Per-team cost accountability
Talk to us
Bring one process. Leave with an outcome architecture.
30 minutes on one function, mapped to the architecture behind it. No slides.