Case Study
NoTraffic
Safely Optimizing Hundreds of Terabytes of ClickHouse on AWS. Without Downtime.
NoTraffic Boosts ClickHouse Disk Utilization from 30% to 95% Without Downtime
- Key environment characteristics
- AWS infrastructure (EC2 and EKS)
- Self-hosted ClickHouse clusters
- EBS volumes reaching up to 10 TiB per node
- Hundreds of terabytes under active management
Company
NoTraffic operates large-scale data platforms that power real-time traffic intelligence systems used by municipalities and transportation networks.
Infrastructure
Their infrastructure processes massive time-series datasets using self-hosted ClickHouse clusters running on AWS. These clusters manage hundreds of terabytes of data and support real-time analytics that require high reliability and continuous availability.
“We self-host Clickhouse database for massive time series tables, with each volume reaching up to 10 TiB of data. As data usage fluctuates, we have opted to over-provisioning our volumes to reduce manual overhead and downtime. Datafy’s solution auto-scales these volumes with almost no effort, improving our disk space utilization dramatically from 30% to 95%.”
NoTraffic, Alexander Rozenberg, IT Director
NoTraffic Boosts ClickHouse Disk Utilization
Maximizing Disk Utilization Without Downtime
The Challenge
NoTraffic’s ClickHouse clusters handle massive volumes of time-series data. As traffic data grows and fluctuates over time, storage requirements shift continuously.
To avoid operational overhead and reduce the risk of downtime, the infrastructure team intentionally over-provisioned storage volumes.
Once EBS volumes were created and attached to production nodes, they were rarely adjusted. This was largely due to AWS limitations and the operational risk associated with modifying storage on live systems.
Several factors contributed to this challenge:
- AWS does not support shrinking EBS volumes natively
- ClickHouse workloads are highly stateful
- Manual volume migration or rebalancing introduces significant risk
- Any downtime would directly impact production services
As a result, storage capacity continued to grow over time while utilization remained relatively low.
The issue was not a lack of awareness about inefficiency.
It was the absence of a safe and reliable way to optimize storage in production.
Deployment Approach
Datafy was introduced gradually across NoTraffic’s ClickHouse environment to ensure stability.
The rollout process included:
- Initial validation on a small subset of volumes
- Continuous monitoring of real storage utilization under live workloads
- Controlled in-place grow and shrink operations
- Progressive expansion after verifying stability
Importantly, the deployment required no changes to the application environment.
The following components remained untouched:
- ClickHouse configuration
- Database schema
- Query behavior
- Cluster topology
- Production availability
All nodes remained fully operational throughout the entire process.
The Results
Datafy enabled NoTraffic to safely optimize storage across hundreds of terabytes of data without interrupting production systems.
Key outcomes included:
- Disk utilization improved from ~30% to ~95%
- Hundreds of terabytes continuously aligned with real usage
- Large amounts of over-provisioned EBS capacity safely reclaimed
- Zero downtime during grow or shrink operations
- No infrastructure or application changes required
Storage now dynamically adjusts to actual workload requirements while maintaining full operational stability.
Why This Matters
For NoTraffic, infrastructure reliability is directly tied to services used by municipal traffic systems and transportation networks.
Storage failures or downtime are not acceptable.
Datafy allowed storage infrastructure to become adaptable in production environments, enabling optimization that is:
- Safe
- Predictable
- Continuous
- Downtime-free
This transformed storage optimization from a theoretical improvement into a practical operational capability.
Secure. Compliant. Reliable.
Gain control over your EBS with Datafy