LStore, or Logistical Storage, is a customizable storage architecture for organizations managing large datasets, high-throughput workflows, and long-term data growth. It aligns performance and capacity to how your data is used, rather than forcing a one-size-fits-all model.
Lstore is a hardware-agnostic, tiered storage architecture that eliminates appliance lock-in, reduces blended cost per TB, and easily scales with capacity demand.
It enables organizations to:
Built on commodity hardware and designed for horizontal scale, LStore gives teams more control over both performance and long-term cost.
Physically and logically isolate multiple tenants without partitioning the storage pool. Enable encryption at rest only for file paths that need it.
Traditional storage often requires large upgrades or replacements. LStore grows in smaller increments, making expansion more predictable and easier to manage.
By aligning storage tiers to data value and usage, LStore helps organizations avoid overbuilding expensive high-performance storage for data that doesn’t need it.
A typical LStore architecture includes multiple tiers working together:
These tiers can be unified under a single namespace, allowing data to move between them without disrupting users or applications. This approach can be enabled by technologies such as Hammerspace as illustrated here:
ACCRE, Vanderbilt’s Advanced Computing Center for Research and Education, needed to modernize its storage environment to support growing AI and HPC workloads.
By implementing a tiered, commodity-based storage architecture, ACCRE was able to:
ACCRE can provide shared infrastructure for researchers across the university, enabling compute-intensive workloads while delivering storage services that scales with changing needs.
SourceCode engineers work with customers to co-design each deployment based on GPU density, expected dataset growth, concurrency profile, backup policy, and budget.
