The Situation
A global travel tech SaaS platform was ingesting up to 0.5GB of data every 30 seconds — 3.3 billion records a month — through an architecture that was never designed for it. Frequent connection disruptions, maxed-out IOPS, and spiralling costs were degrading application performance for hundreds of hotel and OTA clients worldwide.
The Problem in Numbers
0.5 GB
Every 30s
of ingestion
3.3B
Records/month
at peak load
Frequent
Outages disrupting
platform
Escalating
Infra costs
with no ceiling
What Cuedo Built
Architecture flow
A cloud-native, near real-time streaming platform on AWS — replacing the legacy database entirely.
01
Backend Application
02
Kinesis Firehose
03
S3 Data Lake
04
AWS Glue / DMS
05
Athena
06
QuickSight + Alerts
Key engineering decisions:
- Kinesis Firehose for high-throughput streaming ingestion with no data loss during spikes
- Bronze-to-Silver data lake architecture on S3 for reliable staging and transformation
- Spark-based parallel reporting jobs via AWS Glue — highly optimised for speed
- Reverse ETL feeding processed data back into the client's live application
- Terraform and Jenkins for full infrastructure-as-code and CI/CD
- One-click automated deployment end to end
The Results
Before
Database breaking under load
Hours to process reports
Frequent outages & data loss
Unpredictable infra costs
Manual deployment
After
Handles 5× current volumes
25M records in under 5 min
Zero data loss during spikes
Under $1,000/month
One-click automated deploy
A platform now processing 3.3 billion records a month— reliably, cost-efficiently, and built to scale.