v0.4.9

What's Changed

• fix: replace bincode with postcard in raft log by @kacy in #296
• feat: ember-client — native async RESP3 client library by @kacy in #297
• feat: CLI watch mode, batch mode, and proto-ts stub by @kacy in #298
• feat: add Redis 6.2+ commands (LMOVE, GETDEL, GETEX, ZDIFF, ZINTER, ZUNION) by @kacy in #299
• feat(server): automatic snapshot scheduling by @kacy in #300
• fix: rustfmt and clippy cleanup after sprint 1-4 merges by @kacy in #301
• refactor(cli): remove duplicate connection code by @kacy in #302
• feat(client): typed API and pipelining by @kacy in #303
• feat(client): close gRPC API gap — 31 new commands + pub/sub subscriber + vector commands by @kacy in #304
• feat(client): add hincrby by @kacy in #305
• feat(grpc): close api gap — 39 new rpcs by @kacy in #306
• docs: update ember-client readme, add crates index, regenerate protos by @kacy in #307
• feat(ts): scaffold ember-ts client with proto codegen by @kacy in #308
• feat(ts): ember-ts client — 65+ methods, full api parity by @kacy in #309
• fix: launch readiness — auth docs, healthcheck, metrics, info fields by @kacy in #310
• fix: rustfmt cleanup by @kacy in #311
• feat: WAIT command and CONFIG SET live-apply (maxmemory, maxmemory-policy) by @kacy in #312
• feat: keyspace notifications by @kacy in #313
• observability and reliability: prometheus counters, crash recovery, TLS tests by @kacy in #314
• feat: EXPIRETIME, PEXPIRETIME, SMOVE, and SINTERCARD by @kacy in #315
• feat: bitmap commands (GETBIT, SETBIT, BITCOUNT, BITPOS, BITOP) by @kacy in #316
• feat: LMPOP and ZMPOP commands by @kacy in #317
• feat: EXPIREAT, PEXPIREAT, GETSET, MSETNX, LPOP/RPOP count by @kacy in #318
• feat: HRANDFIELD and ZRANDMEMBER by @kacy in #319
• feat: surface 17 new commands across cli, grpc, and client libraries by @kacy in #320
• docs: accuracy pass — concurrent mode removal + command parity by @kacy in #321
• fix: eliminate production panics in client decoder and server startup by @kacy in #322
• fix: auth failure metrics, command memory budget, migration progress by @kacy in #323
• launch prep: FLUSHALL, MEMORY USAGE, and docs clarity by @kacy in #324
• fix: replication send failure counter and go subscribe error propagation by @kacy in #325
• docs: remove status section, relocate stats to benchmarks by @kacy in #326
• fix: deps and ci hardening by @kacy in #327
• style: cargo fmt by @kacy in #328
• fix: remove atomic-polyfill from dep tree by @kacy in #329
• fix: bitop cross-shard correctness; update test count by @kacy in #330
• feat: pre-launch quick wins (COMMAND, HINCRBYFLOAT, Z*STORE, helm pvc, npm publish) by @kacy in #331
• chore: rename npm package to emberdb, switch to oidc trusted publishing by @kacy in #333
• chore: rename npm package to emberdb, add oidc trusted publishing by @kacy in #332
• chore: add HEALTHCHECK to Dockerfile by @kacy in #334
• docs: update command count to 190+ by @kacy in #335
• style: cargo fmt — fix formatting from pr #331 by @kacy in #336
• fix(go): regenerate protobuf stubs for 17 new rpcs by @kacy in #337
• chore: bump version to 0.4.9 by @kacy in #338
• fix: make cluster produce cluster_state:ok on bootstrap by @kacy in #339

Full Changelog: v0.4.8...v0.4.9

v0.4.8

performance

• entry struct optimizations: version field moved to lazy side table, cached_value_size packed as u32, ENTRY_OVERHEAD tightened from 128 to 104 (#284-287)
• skip touch() timestamp updates when eviction is disabled (#287)
• packed hash encoding — hash memory reduced from ~451 to ~240 B/key (#276)
• vector insert throughput optimization with binary-encoded VADD_BATCH (#271-272)

fixes

• ENTRY_OVERHEAD bumped from 100 to 104 for cross-platform CI compatibility (#292-294)
• rate-limited ENOSPC handling for AOF writes (#249)

docs

• removed concurrent mode references — sharded is now the only execution mode (#275, #291)
• refreshed all benchmark numbers from 2026-02-25 GCP run (#288-290)
• added documentation section, code of conduct, performance tuning guide, production checklist (#270, #281-282)

full changelog: v0.4.7...v0.4.8

What's Changed

• fix: phase a quick wins — replication, dead code, CI coverage by @kacy in #241
• ci: enable cluster + CLI integration tests by @kacy in #242
• fix: use serial execution for integration tests by @kacy in #243
• fix: diagnose and fix cluster integration test failures on CI by @kacy in #244
• feat: persistent raft log storage by @kacy in #245
• docs: update raft persistence references by @kacy in #246
• feat: SSCAN, HSCAN, ZSCAN collection scanning by @kacy in #247
• feat: CLIENT ID / SETNAME / GETNAME / LIST by @kacy in #248
• fix: rate-limited ENOSPC handling for AOF writes by @kacy in #249
• feat: WATCH optimistic locking + INFO completeness by @kacy in #250
• feat: COPY, OBJECT, TIME, LASTSAVE, ROLE (audit #11-13) by @kacy in #251
• feat: ACL per-user access control by @kacy in #252
• docs: update READMEs for recent feature additions by @kacy in #253
• feat: RANDOMKEY, TOUCH, SORT commands by @kacy in #254
• feat: ZREVRANK, ZREVRANGE, ZCOUNT, ZINCRBY, ZRANGEBYSCORE, ZREVRANGEBYSCORE, ZPOPMIN, ZPOPMAX by @kacy in #255
• feat: SUNION, SINTER, SDIFF, SPOP, SRANDMEMBER, SMISMEMBER + STORE variants by @kacy in #256
• feat: LINDEX, LSET, LTRIM, LINSERT, LREM, LPOS list commands by @kacy in #257
• feat: SETNX, SETEX, PSETEX string commands by @kacy in #258
• feat: GETRANGE, SETRANGE, SUBSTR string commands by @kacy in #259
• feat: disk-full handling for AOF writes by @kacy in #260
• feat: OOM rejection visibility by @kacy in #261
• feat: config validation bounds checking by @kacy in #262
• feat: improve graceful shutdown drain visibility by @kacy in #263
• chore: switch license from MIT to Apache 2.0 by @kacy in #264
• refactor: split command.rs into command/ module by @kacy in #265
• refactor: split connection.rs into connection/ module by @kacy in #266
• refactor: deduplicate expiry-check pattern with get_live_entry helper by @kacy in #267
• refactor: eliminate unnecessary clones in sorted set and crossslot paths by @kacy in #268
• feat: data type, eviction, and stress benchmark workloads by @kacy in #269
• docs: changelog, migration guide, production checklist, troubleshooting, performance tuning by @kacy in #270
• perf: vector insert performance optimization by @kacy in #271
• perf: optimize vector insert throughput by @kacy in #272
• docs: update vector benchmarks after insert optimization by @kacy in #273
• docs: add SIFT1M recall and throughput results by @kacy in #274
• docs: refresh all benchmarks, remove concurrent mode by @kacy in #275
• perf: packed hash encoding to reduce memory ~451 → ~240 B/key by @kacy in #276
• fix: remove stale --no-grpc flag from bench scripts by @kacy in #277
• docs: update hash memory benchmark from 451 to 243 B/key by @kacy in #278
• docs: fix outdated compatibility matrix and stale references by @kacy in #279
• fix: cargo fmt on hash module by @kacy in #280
• docs: add documentation section to readme by @kacy in #281
• docs: add code of conduct by @kacy in #282
• docs: fix stale claims in readmes by @kacy in #283
• perf: move version field to lazy side table by @kacy in #284
• perf: pack cached_value_size as u32 by @kacy in #285
• perf: tighten ENTRY_OVERHEAD from 116 to 100 by @kacy in #286
• perf: skip touch() when eviction is disabled by @kacy in #287
• docs: update memory benchmark numbers by @kacy in #288
• docs: update stale memory numbers in readme by @kacy in #289
• docs: update benchmark results from 2026-02-25 GCP run by @kacy in #290
• docs: remove concurrent mode references, update readme benchmarks by @kacy in #291
• fix: bump ENTRY_OVERHEAD from 100 to 104 for CI platform compat by @kacy in #292
• fix: update safety_margin test for ENTRY_OVERHEAD 104 by @kacy in #293
• fix: update stale overhead comment in eviction test by @kacy in #294
• chore: release v0.4.8 by @kacy in #295

Full Changelog: v0.4.7...v0.4.8

v0.4.7

What's Changed

• docs: add homebrew install instructions and badge by @kacy in #240

Full Changelog: v0.4.6...v0.4.7

ember-0.1.1

a low-latency, memory-efficient distributed cache

ember-0.1.0

a low-latency, memory-efficient distributed cache

v0.4.5

what's new

batch pipeline dispatch

sharded mode pipelining had a major throughput regression after the thread-per-core PR (#230). per-command channel sends caused head-of-line blocking at high pipeline depths (P=16 dropped from 1.2M to 202k ops/sec).

the fix groups pipeline commands by target shard and sends one batch message per shard, reducing channel traffic from O(pipeline_depth) to O(shard_count). this eliminates contention and restores monotonic scaling:

pipeline depth	SET (ops/sec)	GET (ops/sec)
P=1	222k	222k
P=4	765k	764k
P=16	1.56M	1.96M
P=64	1.95M	3.21M
P=256	2.07M	4.08M

sharded mode now achieves 1.6x redis SET and 1.7x redis GET throughput at P=16 (up from 1.2-1.3x).

other changes

• thread-per-core workers with SO_REUSEPORT (#230)
• CompactString keys with SSO for ≤24 bytes (#228)
• zero-copy read loop via split_to (#227)
• hot-path optimizations: fused lookups, cached sizes, packed Entry (#226)
• O(log n) sorted set rank queries (#224)
• pipeline scaling and transaction benchmarks (#231)
• sort list commands alphabetically in CLI help
• resolve clippy warnings and protobuf CompactString migration

full changelog

v0.4.4...v0.4.5

v0.4.4

what's new

VADD_BATCH command

batch vector insert for dramatically improved throughput — send multiple vectors in a single command instead of one round-trip per vector.

VADD_BATCH key DIM 3 a 0.1 0.2 0.3 b 0.4 0.5 0.6 METRIC COSINE

• RESP3 and gRPC support
• up to 10,000 vectors per batch
• updated python client (vadd_batch()) and benchmarks

security hardening

comprehensive audit across all 6 crates:

• ember-protocol: reject oversized bulk strings (512 MB cap), aggregate frame limits (1M elements), finite ZADD score validation
• emberkv-core: eliminate panicking indexing, checked arithmetic for memory estimates, SCAN cursor bounds, RENAME self-check
• ember-server: pipeline depth limit (10k), gRPC concurrency limit (256/conn), gRPC auth interceptor with constant-time comparison
• ember-persistence: atomic AOF truncation (write-to-temp-then-rename), vector dimension validation on replay, snapshot shard_id validation, encryption key zeroize on drop
• emberkv-cli: filter AUTH from history, history file permissions (0600), ANSI escape sanitization in responses
• ember-cluster: release-mode slot range assertions, raft command validation (AssignSlots, BeginMigration, CompleteMigration), encoding truncation guards, gossip incarnation poisoning protection

What's Changed

• harden server — startup, shutdown, overflow guards by @kacy in #91
• harden server — startup, shutdown, overflow guards by @kacy in #92
• harden round 3 — overflow, DoS, correctness fixes by @kacy in #93
• tasteful refactoring — clarity, dedup, structure by @kacy in #94
• feat: vector similarity search with HNSW by @kacy in #95
• harden vector operations — bug fixes, tests, docs by @kacy in #96
• fix: gate proto tests behind protobuf feature flag by @kacy in #97
• refactor: deduplicate accept loop, slowlog, pubsub in ember-server by @kacy in #98
• refactor: keyspace helpers for expiry, collection cleanup, and error strings by @kacy in #99
• refactor: extract_strings and extract_bytes_vec helpers in ember-protocol by @kacy in #100
• harden: replace unmaintained dep, enable overflow checks by @kacy in #101
• harden ember-server — auth rate limiting, pubsub limits, integer safety by @kacy in #102
• harden ember-protocol — vector dimension and HNSW parameter caps by @kacy in #103
• harden ember-persistence — safe u32 length casts in format layer by @kacy in #104
• harden ember-cluster — saturating arithmetic in migration progress by @kacy in #105
• perf: build + serialization fast paths by @kacy in #106
• perf: zero-copy command parsing + keyspace fast path by @kacy in #107
• perf: small wins — static bytes and inline hints by @kacy in #108
• update benchmark results post-performance audit by @kacy in #109
• add string helpers to concurrent mode by @kacy in #110
• fix: broken integration tests (46 failures → 0) by @kacy in #111
• perf: sharded mode pipeline throughput optimization by @kacy in #112
• update benchmark results from gcp c2-standard-8 by @kacy in #113
• add vector similarity benchmark suite by @kacy in #114
• harden: vector security audit fixes by @kacy in #115
• feat: grpc proto definitions + codegen infrastructure by @kacy in #116
• feat: grpc server implementation by @kacy in #117
• feat: go grpc client by @kacy in #118
• add python grpc client by @kacy in #119
• refactor: ember-protocol types.rs documentation by @kacy in #120
• refactor: ember-persistence docs and test cleanup by @kacy in #121
• refactor: ember-core unreachable removal and sorted_set optimization by @kacy in #122
• refactor: ember-server grpc error handling and protobuf fix by @kacy in #123
• harden: collection count validation in persistence deserialization by @kacy in #124
• harden: mix random bits into migration ID generation by @kacy in #125
• harden: grpc input validation, ttl overflow, cursor truncation by @kacy in #126
• refactor: extract collection-write helpers in ember-core by @kacy in #127
• refactor: replace grpc pipeline dispatch with macro by @kacy in #128
• refactor: group response tags, use let-else in proto handlers by @kacy in #129
• refactor: aof serialization helpers and snapshot utf-8 helper by @kacy in #130
• security: harden ember-protocol input validation by @kacy in #131
• security: eliminate panic sites in ember-core keyspace by @kacy in #132
• harden ember-server: grpc limits, scan cap, gossip port overflow by @kacy in #133
• bound vector deserialization total allocation in persistence by @kacy in #134
• update dependencies to latest compatible versions by @kacy in #135
• cargo fmt: fix formatting drift from merged refactoring PRs by @kacy in #136
• perf: tier 1 and tier 2 performance audit by @kacy in #137
• docs: add python client readme by @kacy in #138
• docs: add go client readme by @kacy in #139
• docs: add helm chart readme by @kacy in #140
• feat: complete grpc api — pub/sub, slowlog, operational commands by @kacy in #141
• feat: regenerate client stubs + add new client methods by @kacy in #142
• bench: add grpc-based ember vector benchmarks by @kacy in #143
• bench: add qdrant comparison to vector benchmark by @kacy in #144
• bench: add qdrant comparison to vector benchmark by @kacy in #145
• cargo fmt: fix formatting drift from grpc merge by @kacy in #146
• fix: add qdrant support to vm setup and vector benchmark defaults by @kacy in #147
• bench: expand memory benchmarks to all data types by @kacy in #148
• bench: add grpc vs resp3 throughput comparison by @kacy in #149
• bench: add pub/sub throughput benchmark by @kacy in #150
• bench: add protobuf storage overhead benchmark by @kacy in #151
• bench: add bench-all.sh master script by @kacy in #152
• docs: update readmes with new benchmark sections and results by @kacy in #153
• fix: use INFO memory for bench-memory.sh measurements by @kacy in #154
• fix: benchmark suite reliability improvements by @kacy in #156
• fix: proto benchmark key collision and memory DBSIZE parsing by @kacy in #157
• fix: rewrite memory benchmark bulk loading and mode handling by @kacy in #158
• fix: add --no-grpc to benchmarks that don't need grpc by @kacy in #159
• docs: update benchmark results from GCP c2-standard-8 by @kacy in #160
• feat: add VADD_BATCH command for bulk vector inserts by @kacy in #161
• fix: comprehensive security audit hardening by @kacy in #162

Full Changelog: v0.4.3...v0.4.4

v0.4.3

patch release — readme cleanup only, no code changes.

• removed duplicate architecture section with aspirational target metrics (now redundant with real benchmark data in the benchmarks section)
• updated status line: 989 tests, ~21k LOC

Full Changelog: v0.4.2...v0.4.3

v0.4.2

What's Changed

• test: cluster command and CLI integration tests by @kacy in #80
• feat: encryption at rest (AES-256-GCM) by @kacy in #81
• docs: update benchmarks and docs for encryption at rest by @kacy in #82
• feat: TLS support for ember-cli by @kacy in #83

Full Changelog: v0.4.1...v0.4.2

v0.4.1

what's new

• cluster subcommands — ember-cli cluster info, cluster nodes, cluster meet, and all other cluster management commands as typed CLI subcommands (#77)
• built-in benchmark — ember-cli benchmark with pipelining, concurrent clients, and latency percentile reporting (#78)
• CLI polish — syntax highlighting, inline hints, and subcommand tab-completion (#75)
• cluster server integration — wire cluster commands into the server (#72, #74)
• integration test suite — 45 end-to-end tests covering commands, data types, auth, persistence, and pub/sub (#76)
• security hardening — timing-safe auth, restrictive file permissions, password file support (#71)
• refactoring — consolidate cluster slot checks (#73)
• docs — updated all READMEs for new CLI features (#79)

stats

• 85 commands
• 861 tests (817 unit + 44 integration)
• ~18k lines of code (excluding tests)

install

cargo install ember-server
cargo install emberkv-cli