// project
GP-STOR
Tiered object storage for research data, distributed across Great Plains institutions.
GP-STOR — the Great Plains Scalable Tiered Object Repository — is a distributed storage initiative for the region, built around resilient, tiered object storage for research workloads.
My focus is the service architecture: Ceph object storage spread across institutions, fronted by data access that research computing environments already speak, so a dataset is reachable from a job without a bespoke transfer step.
Multi-site durability is the requirement that shapes everything else. Storage that survives a site outage but that researchers find awkward to read from will not be used, so the access path gets as much attention as the replication does.
I am the PI for GP-STOR.
Technologies
- Ceph
- Kubernetes
- Helm
- OSDF — Open Science Data Federation
- XRootD — XRootD distributed data access framework
- Linux