Inferra Product Documentation
KV cache tiering, connectors for vLLM/SGLang/TensorRT-LLM, API reference, and benchmarking methodology.
docs.lightbits.ai → 3 docs to start withWhat Is Inferra
A purpose-built KV cache platform that keeps GPUs producing under long context
AI InferenceInferra Introduction
Learn how Lightbits is reinventing inference with Inferra—the intelligent KV cache engine built for superior token economics, predictable SLAs, and unbounded LLM scale.
AI InferenceOptimized AI Inference
This tech paper illustrates how Inferra is designed to restore—and then expand—GPU efficiency under long-context workloads by proactively managing KV cache movement across memory tiers.
Lightbits SDS Product Documentation
NVMe/TCP architecture, Kubernetes/OpenStack/VMware deployment guides, API reference, and migration guides.
documentation.lightbitslabs.com → 3 docs to start withRelease Notes
All Lightbits SDS release documentation.
StorageInstallation Guide
The Lightbits cluster storage solution distributes services and replicates data across different Lightbits servers to guarantee service and data availability
StorageSupport Guide
Lightbits provides full support for all of our software products. The Support Guide describes the support conditions and communication channels available for our customers.
Getting Started
What Is Inferra
A purpose-built KV cache platform that keeps GPUs producing under long context
InferraInferra Introduction
Learn how Lightbits is reinventing inference with Inferra—the intelligent KV cache engine built for superior token economics, predictable SLAs, and unbounded LLM scale.
InferraOptimized AI Inference
This tech paper illustrates how Inferra is designed to restore—and then expand—GPU efficiency under long-context workloads by proactively managing KV cache movement across memory tiers.
Architecture
KV Cache Tiering
How cache moves across HBM, DRAM, and NVMe without stalling generation.
InferraThe Memory Wall Explained
Why static weights and growing KV cache overwhelm GPU HBM — and how Inferra avoids it.
InferraCache Locality & Oracle Scheduler
The O(1) sparse-attention scheduler that predicts and prefetches the next cache block.
Integrations
vLLM Connector
Drop-in integration guide for vLLM — no framework fork required.
InferraSGLang Connector
Attach Inferra to an SGLang serving deployment.
InferraTensorRT-LLM Connector
Integrate Inferra with TensorRT-LLM inference servers.
InferraLMCache Connector
Bridge existing LMCache deployments onto Inferra’s tiering engine.
API & CLI Reference
Security & Ops
Benchmarks
Release Notes
Current (Rolling)
What’s shipping in the current rolling release.
InferraDA — Aug 31, 2026
Design Availability feature set: sub-linear prefetch, disaggregated inference, RDMA KV encryption.
InferraGA — Sep 15, 2026
General Availability feature set: production hardening, observability, 10M-token context.
No documentation matches your search. Try a different term or clear the search.
Getting Started
Release Notes
All Lightbits SDS release documentation.
Lightbits SDSInstallation Guide
The Lightbits cluster storage solution distributes services and replicates data across different Lightbits servers to guarantee service and data availability
Lightbits SDSSupport Guide
Lightbits provides full support for all of our software products. The Support Guide describes the support conditions and communication channels available for our customers.
Architecture
NVMe/TCP Protocol
The protocol Lightbits invented, explained from first principles.
Lightbits SDSDisaggregated Storage
Why separating compute and storage scaling changes your cost curve.
Lightbits SDSWhy Lightbits SDS
How Lightbits SDS compares to Ceph and traditional SAN architectures.
Deploy
Kubernetes & CSI
Deploy the Lightbits SDS CSI driver and provision persistent volumes.
Lightbits SDSOpenStack & Private Cloud
Integrate Lightbits SDS with Cinder for OpenStack environments.
Lightbits SDSVMware Environments
Connect Lightbits SDS block storage to vSphere and vSAN.
Lightbits SDSPublic Cloud
Run Lightbits SDS on AWS, Azure, or bare-metal cloud instances.
API & CLI Reference
Guides
Ceph Migration Guide
A step-by-step migration path from Ceph to Lightbits SDS with zero downtime.
Lightbits SDSSizing Guide
Right-size your cluster for your workloads, IOPS and capacity needs.
Lightbits SDSHigh Availability Setup
Configure replication and failover for mission-critical volumes.
Release Notes
No documentation matches your search. Try a different term or clear the search.