OBJECT STORAGE FOR YOUR ENVIRONMENT
Your app runs on AWS. Your object store can too.Your app runs on AWS. Your object store can too.
Run self-hosted object storage beside your application, with the same query and permission model you use elsewhere. Choose AWS compute and disks to suit your workload instead of changing your product model to match a service.
Your AWS account. Your deployment. Original files you can take elsewhere.
Example deployment — not an automatic installer
Your AWS environment
Your application
Your chosen compute environment
Compute and disks
For example, EC2 and EBS
↓
Keldra
Self-hosted object storage beside your app
Your application
Your chosen compute environment
Compute and disks
For example, EC2 and EBS
→
Keldra
Self-hosted object storage beside your app
Choose actual instance, disk, network and failure-domain placement for your own requirements.
Why run your own object store on AWS?
Renting infrastructure does not mean outsourcing every part of your application platform.
One deployment model
Use the same Keldra release across development, AWS and your own servers. Test the same object and query workflows.
Place storage near the app
Choose the location and network path for application traffic. Measure latency and transfer costs for that actual topology.
Keep queries with storage
Use configured JSON, text and vector indices with Keldra's supported authorisation instead of stitching every result together in the app.
Control the allocation
Select instance resources, disks and configuration. Capacity planning and maintenance remain yours.
Price your real workload
Compare infrastructure, licence and operational costs with the requests and access patterns your application actually generates.
One deployment model
Use the same Keldra release across development, AWS and your own servers. Test the same object and query workflows.
Place storage near the app
Choose the location and network path for application traffic. Measure latency and transfer costs for that actual topology.
Keep queries with storage
Use configured JSON, text and vector indices with Keldra's supported authorisation instead of stitching every result together in the app.
Control the allocation
Select instance resources, disks and configuration. Capacity planning and maintenance remain yours.
Price your real workload
Compare infrastructure, licence and operational costs with the requests and access patterns your application actually generates.
One deployment model
Use the same Keldra release across development, AWS and your own servers. Test the same object and query workflows.
Place storage near the app
Choose the location and network path for application traffic. Measure latency and transfer costs for that actual topology.
Keep queries with storage
Use configured JSON, text and vector indices with Keldra's supported authorisation instead of stitching every result together in the app.
Control the allocation
Select instance resources, disks and configuration. Capacity planning and maintenance remain yours.
Price your real workload
Compare infrastructure, licence and operational costs with the requests and access patterns your application actually generates.
Compute, disks, network and operations still cost money
You pay for the AWS infrastructure you choose, including applicable network and snapshot charges, and operate Keldra yourself. Self-hosting is not automatically cheaper than a managed service. Compare your workload, resilience, backups and staff time before choosing.
Amazon S3 is a managed option
Amazon S3 is a managed object-storage service with its own storage, request and transfer pricing. Delegating infrastructure can be the right choice when you do not want to operate an object store.
Both approaches have a place. This is an operating-model decision, not a claim that one product wins every workload.
Provider references checked on 14 October 2026. No AWS endorsement, automatic savings or unrestricted S3 compatibility is implied.