5 posts tagged with "KV Cache"

A request arrives and its 48K-token prefix is already sitting in a KV cache, but the endpoint holding it - one schedulable serving instance: either a complete engine instance or a rank within a multi-rank deployment - is busy. Route the request to the cache owner and it queues behind that backlog; route it to an idle endpoint and it spends seconds recomputing work the cluster already did. Waiting and recomputing are both wrong answers. P2P (peer-to-peer) KV cache sharing adds the third option: send the request to the best endpoint for load, and copy the finished KV to it.

A busy owner is only one way locality breaks. Load balancing may select another endpoint, or the required KV may have been generated by a different serving role. In every such case the llm-d router already knows how much of a request's prefix each candidate endpoint holds. With P2P, that knowledge becomes a transfer instruction: route the request to the best endpoint overall, then tell it where to fetch the missing prefix.

This is not a universal speedup. When routing already produces a local hit, P2P correctly does nothing. When locality conflicts with load balance or serving topology, it can replace seconds of repeated computation with a peer transfer. The operating principle is simple: local hits first; portable reuse when locality breaks.

The rest of this post answers four questions: how expensive is a pull; whether it improves load-balanced serving; whether it preserves session history across prefill/decode (P/D) roles; and when it should stay inactive.

llm-d's well-lit paths are tested, benchmarked recipes for serving large language models in production. The agentic-serving path starts with the workload and combines the capabilities it needs into a deployment for each accelerator. This post applies that pattern to GLM-5.2-FP8 on H200, using the 219 production Claude Code sessions in SemiAnalysis's Weka trace corpus.

GLM-5.2 is a 744B-parameter (39B active) mixture-of-experts model with 78 layers. Its Multi-head Latent Attention (MLA) stores one compressed latent instead of separate KV heads, so tensor parallelism cannot shard it: TP=8 keeps eight copies of the KV cache. Tensor-parallel all-reduce is also expensive across nodes: every layer performs the column-parallel, row-parallel, and all-reduce sequence twice. llm-d's Wide-EP path avoids both costs with data-parallel attention and expert-parallel MoE, making the KV budget viable for long-context agentic serving.

At a measured output rate of roughly 37-41 tokens per second per user, a 64-GPU llm-d deployment works out to $0.05 per million input tokens and $3.55 per million output tokens at an H200 rental rate of $2.71 per GPU-hour. Z.ai's public GLM-5.2 row lists $1.40 input and $4.40 output; Claude Opus 4.8 lists $5 input and $25 output. These figures assume sustained load and assign prefill rental to input and decode rental to output.

For most of the transformer era, the KV cache rested on a quiet assumption: one model, one uniform cache. Every layer attended the same way, every block was the same size, and everything built on top of the cache (allocators, offload connectors, schedulers) could treat it as a single pool.

Hybrid models broke this assumption. Many recent frontier and open-weight models increasingly mix attention types within a single model (full attention next to sliding-window, linear, or Mamba layers), making the cache heterogeneous: different layers now hold different amounts of state, in different shapes, with different reuse rules. A cache block that used to be allocated as one uniform unit is now constituted of several distinct parts.

To serve a hybrid model efficiently, an AI inference platform has to handle that heterogeneity in at least three aspects of the stack:

• GPU Memory Allocation: How the cache is laid out and allocated on the GPU. vLLM solved this with its Hybrid Memory Allocator (HMA), rebuilt around a unified allocator (see Hybrid Models as First-Class Citizens in vLLM).
• KV Offloading: Extending the KV cache to CPU and storage. Without HMA awareness, an offloading connector turns the HMA off and therefore discards the GPU memory improvements or potential data movement savings.
• KV-Aware Routing: Sending each request to the right model-server replica. Ignoring hybrid memory structure may erroneously list nodes as having or not having the required KV data based on information stemming from just part of the layers.

vLLM's HMA solved hybrid GPU memory allocation when handling a single vLLM instance. This post shows how llm-d extends that to tiered KV cache management - including KV offloading to CPU and storage, and KV-aware request routing - significantly improving throughput and latency at scale for hybrid models.

Most production inference clusters today are single-vendor because that is often the simplest way to configure and operate a cluster.

That is starting to change. Procurement cycles bring new generations alongside older ones, supply planning spans multiple accelerator options, and cost/performance profiles differ by workload. Real production fleets are accumulating heterogeneity whether or not the architecture planned for it.

This is an opportunity to unlock real value: different accelerator classes can be matched to workload requirements, stranded capacity gets reclaimed, and operators gain more flexibility in capacity planning. The case is stronger still for sovereign and on-premise deployments, where data residency, regulatory alignment, and the long-term economics of high-volume inference make local fleet optimization especially important.

Making that work in practice is a non-trivial systems problem. Each accelerator stack brings its own optimized drivers, firmware, container images, runtime settings, and attention kernels. A coherent serving layer needs to preserve those platform-specific optimizations while still giving operators one control plane for routing, observability, and policy.

llm-d is a distributed inference platform spanning multiple vLLM instances. KV cache hits are critical to achieving high inference throughput. Yet, in a distributed environment, cache hits do not occur across different nodes as the KV cache is local to each vLLM instance. In addition, this local cache is limited in size, further limiting KV data reuse. This blog presents a new way to offload KV cache to storage, tackling both aforementioned challenges – KV cache sharing and KV cache scale. llm-d's filesystem (FS) backend is a KV cache storage connector for vLLM that offloads KV blocks to shared storage based on vLLM's native Offloading Connector. While the llm-d FS backend can speed up serving of single requests (improve TTFT), its main goal is rather to preserve stable throughput and low latency at scale, as concurrency and context lengths grow. This is accomplished by significantly enlarging the cache space and enabling KV reuse across multiple replicas and nodes in llm-d.

While there are a number of existing solutions for KV cache offload to storage (e.g. LMCache or Dynamo KVBM), the new connector offers simplicity, can run with llm-d and vLLM as the only dependency, and exhibits improved performance over state-of-the-art shared storage connectors.