Case Study
Weaviate
Automatically Rightsizing EBS Volumes Across Dedicated Customer Infrastructure
Cuts Wasted EBS Spend by 50%
- Weaviate’s AWS Environment
- Dedicated customer clusters
- >100 AWS accounts
- >140 EBS volumes managed by Datafy
Company
Weaviate is a database platform built for developers and enterprises creating AI applications.
Its platform includes a vector database alongside capabilities such as Query Agent and built-in embedding service.
Infrastructure
Weaviate’s platform team operates dedicated environments for its customers. Each customer gets an isolated cluster, which means storage capacity needs to be managed independently across those environments.
As the number of clusters grows and customer workloads change, the platform team needs to ensure every environment has enough capacity while avoiding unnecessary overprovisioning.
Before Datafy, average storage utilization across the relevant Weaviate infrastructure was approximately 35%.
“The decision was driven by the seamless integration. Without requiring CSI driver replacements, filesystem modifications, or additional management effort, the solution delivered more than just functional improvements. For Weaviate, this translates into a strategic edge: reducing the cost of our AI services allows for more aggressive pricing, helping us capture a larger market share.”
Spiros Andreou, VP Platform Operations and Security, Weaviate
Rightsizing EBS, Including Downsizing, Without Manual Effort
The Challenge
Every Weaviate customer cluster needs enough storage capacity to safely support its workload and future growth. But actual storage consumption varies between environments and changes over time.
To reduce the risk of running out of space, platform teams need to provision sufficient headroom. The challenge is that while EBS volumes can grow, they cannot natively shrink.
As a result, capacity added to accommodate growth can remain allocated even when it is no longer required. Across dedicated customer clusters, manually managing these capacity decisions does not scale.
Several factors contributed to this challenge:
- AWS does not support shrinking EBS volumes natively
- Storage requirements vary across customer environments and over time
- Manual rightsizing across dedicated clusters creates operational overhead
- Storage changes need to be made without putting customer production environments at risk
Weaviate needed a way to automatically rightsize EBS volumes based on actual requirements, including downsizing disks when excess capacity was no longer needed.
The goal was to achieve this without replacing CSI drivers, modifying filesystems, or introducing another storage layer to operate and support.
Deployment Approach
Datafy was introduced gradually into Weaviate’s existing infrastructure.
The process started with Datafy Sensor, providing visibility into the storage environment and identifying opportunities to improve utilization.
The rollout process included:
- Infrastructure integration
- QA and PoC
- Initial deployment
- Production deployment of Datafy EBS Auto-Scaler
- Automatic grow and shrink operations based on storage requirements
Importantly, the deployment required no changes to Weaviate’s existing storage architecture. This was particularly important given where Datafy sits in Weaviate’s infrastructure.
Every Dedicated Cloud cluster runs a named enterprise customer’s live AI service. There is no margin for error and no acceptable window for downtime, making reliability and support critical parts of Weaviate’s evaluation.
The platform team looked at several storage vendors, and every alternative required them to change how they run storage, whether through a different CSI driver, a different filesystem, or a new layer to operate and support.
Datafy required none of those changes. The same standard has applied to support: fast, specific answers when something needs attention, allowing Weaviate’s team to maintain the reliability its customers depend on.
“Plug-and-play is what got the contract over the finish line. Every other vendor we looked at wanted us to change how we do storage. We can’t afford a support relationship that puts our customers’ production environments at risk, and Datafy hasn’t.”
Brave Okafor, Senior Platform Engineer, Weaviate
The Results
Datafy enabled Weaviate to automatically rightsize EBS volumes across its dedicated customer infrastructure, including downsizing disks when excess capacity is no longer required.
Key outcomes include:
- 50% reduction in wasted EBS spend
- >140 EBS volumes currently managed by Datafy
- Automatic EBS rightsizing, including downsizing
- Reduced manual effort involved in managing storage capacity
- Less EBS capacity sitting unused
- No CSI driver replacements or filesystem modifications required
Weaviate has also been able to test additional capabilities, including enhanced EBS performance, while gaining a unified console with visibility into volumes across approximately 100 AWS accounts.
Higher utilization means less unnecessary EBS capacity, reducing infrastructure costs while maintaining the capacity required to safely support customer environments.
Storage can now adjust more closely to actual workload requirements without requiring the platform team to manually manage every capacity change.
Operational Impact
For Weaviate, infrastructure reliability is directly tied to the AI services running in its customers’ production environments.
Storage needs to adapt as those environments change, without introducing downtime, additional operational complexity, or unnecessary capacity.
Datafy allows Weaviate to make EBS rightsizing:
- Automatic
- Bi-directional, including downsizing
- Continuous
- Compatible with its existing storage architecture
This turns EBS rightsizing from an ongoing manual infrastructure task into a practical operational capability that can scale across Weaviate’s customer environments.
Secure. Compliant. Reliable.
Gain control over your EBS with Datafy