lib.rs 2.08 KB
Newer Older
1
2
3
4
5
6
7
8
9
10
// SPDX-FileCopyrightText: Copyright (c) 2024-2026 NVIDIA CORPORATION & AFFILIATES. All rights reserved.
// SPDX-License-Identifier: Apache-2.0

//! KV Router - Radix tree data structures for LLM KV cache routing.
//!
//! This crate provides the core radix tree implementation and protocols for
//! efficient KV cache lookup and routing in distributed LLM inference systems.

pub mod indexer;
pub mod protocols;
11
pub mod recovery;
12
13
pub mod scheduling;
pub mod sequences;
14
pub mod zmq_wire;
15

16
17
// Backward-compat re-exports: old top-level module paths still work
pub use indexer::concurrent_radix_tree;
18
pub use indexer::concurrent_radix_tree_compressed;
19
20
21
22
pub use indexer::positional as nested_map;
pub use indexer::pruning as approx;
pub use indexer::radix_tree;

23
24
25
26
27
pub use scheduling::config;
pub use scheduling::queue;
pub use scheduling::selector;
pub use sequences::multi_worker as multi_worker_sequence;
pub use sequences::single as sequence;
28

29
30
31
#[cfg(feature = "standalone-indexer")]
pub mod standalone_indexer;

32
33
#[cfg(any(test, feature = "bench"))]
pub mod test_utils;
Yan Ru Pei's avatar
Yan Ru Pei committed
34

35
// Re-export key types for convenience
36
pub use self::multi_worker_sequence::{
37
38
39
    ActiveSequencesMultiWorker, SequenceError, SequencePublisher, SequenceRequest,
    SequenceSubscriber,
};
40
41
pub use self::sequence::{ActiveSequences, RequestId};
pub use concurrent_radix_tree::ConcurrentRadixTree;
42
pub use concurrent_radix_tree_compressed::ConcurrentRadixTreeCompressed;
43
pub use config::{KvRouterConfig, RouterConfigOverride, RouterQueuePolicy};
44
pub use indexer::{MaybeError, SyncIndexer, ThreadPoolIndexer};
Yan Ru Pei's avatar
Yan Ru Pei committed
45
pub use nested_map::PositionalIndexer;
46
pub use protocols::{
47
48
    KvCacheEventError, LocalBlockHash, OverlapScores, RouterEvent, RouterEventSink,
    WorkerConfigLike, WorkerId, compute_block_hash_for_seq,
49
};
50
pub use queue::SchedulerQueue;
51
pub use radix_tree::RadixTree;
52
pub use scheduling::LocalScheduler;
53
pub use scheduling::policy::{FcfsPolicy, RouterSchedulingPolicy, SchedulingPolicy, WsptPolicy};
54
55
pub use scheduling::{KvSchedulerError, PotentialLoad, SchedulingRequest, SchedulingResponse};
pub use selector::{DefaultWorkerSelector, WorkerSelector};