iiinova — where your digital self keeps evolving

ArcadeDB @arcadedb.bsky.social · 08/10/2026

Kainora gives retrieval more paths to evidence: a knowledge graph in ArcadeDB, next to keyword and vector search, running on Docker and Kubernetes. Case study: arcadedb.com/case-studies...

000

ArcadeDB @arcadedb.bsky.social · 06/10/2026

ArcadeDB 26.10.1 is out. The hardening release: - ~300 HA / Raft fixes - 984 bug fixes (75 critical) - 4 security advisories patched - TLS for the Postgres wire protocol Open source, Apache 2.0, multi-model (graph, document, key/value, search, time series, vector). arcadedb.com/blog/arcaded...

arcadedb.com

ArcadeDB 26.10.1: HA Hardening, Lost-Update Fixes, Consistent Equality, and 4 Security Advisories

ArcadeDB 26.10.1 closes 1,615 issues and pull requests across about 3,100 commits. Raft high availability is hardened end to end, lost updates and silent data loss are closed, index, SQL and Cypher no...

052

ArcadeDB @arcadedb.bsky.social · 01/10/2026

To be fair: if relationships are a side feature of your app, PostgreSQL will still be a fine choice. The post spells out where that line is. We also keep graph, documents, vectors and geospatial in one transaction.

101

ArcadeDB @arcadedb.bsky.social · 01/10/2026

Index-free adjacency: each vertex points directly to its edges, so a hop is a pointer follow, not an index search. Traversal cost stays independent of total database size. Star patterns (Q4) showed 229x, Q7 showed 561x.

100

ArcadeDB @arcadedb.bsky.social · 01/10/2026

SQL/PGQ is a rewriter. MATCH patterns become JOINs and the plans are identical to hand-written JOIN queries. Nothing changes in storage: edges are still rows, hops are still B-tree lookups. A nicer syntax, the same engine underneath.

100

ArcadeDB @arcadedb.bsky.social · 01/10/2026

The PostgreSQL community is working on SQL/PGQ. Will it turn PostgreSQL into a graph database? We ran LDBC LSQB (SF1: 3.9M vertices, 17.9M edges, 9 queries) on PostgreSQL and ArcadeDB. 2.21 s vs 69.07 s total. Per-query gap from 1.8x to 561x. Thread on why

110

ArcadeDB @arcadedb.bsky.social · 28/09/2026

ArcadeDB is now the graph database backbone of Tom Sawyer Digital Engineering from @tsawyersoftware 🎉 6 modules (Planning, BPMN, SysML v2, Data Streams, Perspectives, Explorations), one graph underneath. Apache 2.0, no proprietary DB license cost. arcadedb.com/blog/arcaded...

010

ArcadeDB @arcadedb.bsky.social · 23/09/2026

Best advice in the post, whatever database you use: build a ~500-question gold set before you migrate. It turns opinions into a table and becomes your regression suite. "Licensing is an architecture constraint." ArcadeDB is Apache 2.0, HA included: arcadedb.com

010

ArcadeDB @arcadedb.bsky.social · 23/09/2026

What they reported: - p99 retrieval 10.7s → 140ms - Recall@8 roughly doubled - HA without a license gate - openCypher: port, not rewrite Their own caveat: much of the speedup was finally having a real vector index. Fair.

110

ArcadeDB @arcadedb.bsky.social · 23/09/2026

A team doing RAG over 40M+ legal/medical docs outgrew Neo4j Community (3.1B nodes, 8B edges, 640M vectors) and moved to ArcadeDB. Honest write-up, caveats included: www.reddit.com/r/Rag/commen...

110

ArcadeDB @arcadedb.bsky.social · 23/09/2026

Apache Arrow now reaches ArcadeDB over its PostgreSQL wire protocol. adbcBridge tested it for 3 weeks and found 4 gaps. First fix: 4.5 hours. Every fix went upstream. arcadedb.com/blog/arcaded...

010

ArcadeDB @arcadedb.bsky.social · 17/09/2026

ArcadeDB is now a first-class integration in Google's MCP Toolbox for Databases, the open-source MCP server for AI agents. Query graph, document, vector & time-series data in Cypher or SQL. arcadedb.com/blog/arcaded...

arcadedb.com

ArcadeDB Joins Google's MCP Toolbox for Databases

ArcadeDB is now a first-class integration in Google's MCP Toolbox for Databases, the open-source MCP server that connects AI agents to enterprise databases alongside PostgreSQL, BigQuery, Spanner, Sno...

001

ArcadeDB @arcadedb.bsky.social · 10/09/2026

4. Drift gate Hand-written drivers rot silently. The ArcadeDB drivers are generated from the server's own contracts, and CI fails the build the day the contract moves.

000

ArcadeDB @arcadedb.bsky.social · 10/09/2026

3. The truncated flag A result that stopped early and a result that ran out of matches look identical by shape. The ArcadeDB drivers return the truncated flag so you can tell them apart.

100

ArcadeDB @arcadedb.bsky.social · 10/09/2026

2. HTTP or gRPC Two ArcadeDB drivers per language. HTTP for browsers, serverless, and ordinary traffic. gRPC for server-to-server streaming and bulk inserts. Start with HTTP.

100

ArcadeDB @arcadedb.bsky.social · 03/09/2026

New SQL in ArcadeDB 26.9.1: INSERT ... ON DUPLICATE KEY SKIP, so one duplicate key no longer aborts a whole batch. Plus CHECK DATABASE ... DEEP and -onRowError skip in the bulk importer.

000

ArcadeDB @arcadedb.bsky.social · 03/09/2026

New in ArcadeDB 26.9.1 for clusters: optional mTLS on the Raft gRPC transport, live Raft membership in /api/v1/cluster, a per-protocol HA routing table, and jittered exponential backoff on tx retries.

100

ArcadeDB @arcadedb.bsky.social · 03/09/2026

ArcadeDB 26.9.1 for Neo4j and GQL users: 12 APOC-compatible functions and procedures db.index.fulltext.queryNodes over our BM25 index Neo4j 5 dynamic labels GQL Quantified Path Patterns, Phase B

100

ArcadeDB @arcadedb.bsky.social · 03/09/2026

Vector search in ArcadeDB 26.9.1: Constant-time database open at any index size Rebuilds scheduled, not blocking the first query Location index ~32 bytes per vector, from ~90 Graph builds on every core

100

ArcadeDB @arcadedb.bsky.social · 03/09/2026

ArcadeDB 26.9.1 puts an index behind more query shapes: IN (...) literal lists, BETWEEN, composite prefix + ORDER BY, and @rid IN [...]. That last one at 400k documents: 82.12 ms to 0.07 ms. 1143x.

100

ArcadeDB @arcadedb.bsky.social · 03/09/2026

Backups in ArcadeDB 26.9.1: parallel compression, parallel restore, and a page-level copy-on-write snapshot instead of suspending the flusher. 27.6x faster, and writers keep running at 77% of baseline throughput.

100

ArcadeDB @arcadedb.bsky.social · 07/08/2026

ArcadeDB 26.8.1 is out. 669 commits. 381 issues closed. - 13 security advisories patched. - Concurrent insert conflicts: ~1750 to 0 - Super-node deletes: 5x faster - Geospatial indexes: 11x smaller on disk No schema migration. Just upgrade. arcadedb.com/blog/arcaded...

023

ArcadeDB @arcadedb.bsky.social · 17/07/2026

ArcadeDB 26.7.3 is out — a focused hotfix for 26.7.2: 🛠 1 critical fix: silent edge loss on super-node vertices under concurrent writes 🔒 3 security advisories fixed (MCP permission bypass, MCP settings disclosure, trigger privilege escalation) github.com/ArcadeData/a...

github.com

Release 26.7.3 · ArcadeData/arcadedb

ArcadeDB 26.7.3 Overview This is a focused hotfix on top of 26.7.2. It closes three security advisories from the internal audit - two in the MCP server transport and one in the JavaScript/Java trig...

110

ArcadeDB @arcadedb.bsky.social · 08/07/2026

ArcadeDB! Apache 2 license, free for any usage without Neo4j limitations: arcadedb.com/neo4j.html

000

ArcadeDB @arcadedb.bsky.social · 07/07/2026

New ArcadeDB post: cloud observability for a database on Kubernetes. 🔭 OpenTelemetry tracing (through Raft replication) 📊 RED metrics via Micrometer 🧾 Structured JSON logs with correlation IDs ❤️ K8s liveness + readiness probes Works with Grafana/Prometheus/Tempo. arcadedb.com/blog/arcaded...

arcadedb.com

ArcadeDB Cloud Observability: OpenTelemetry Tracing, Structured Logging, and Kubernetes Health Probes

ArcadeDB adds vendor-neutral OpenTelemetry tracing, RED latency metrics with OTLP export, structured JSON logging with correlation IDs, and Kubernetes-standard health probes, all opt-in and fully back...

111

ArcadeDB @arcadedb.bsky.social · 03/07/2026

ArcadeDB 26.7.1 is out 🚀 420+ commits, 238 issues resolved, no migration needed. Highlights: Raft/HA hardening, native BM25 full-text scoring, OpenTelemetry tracing, ISO GQL alignment, and tighter security. arcadedb.com/blog/arcaded... #ArcadeDB #OpenSource #Database

arcadedb.com

ArcadeDB 26.7.1: Raft Hardening, OpenTelemetry Tracing, BM25 & GQL

ArcadeDB 26.7.1 is a large stability, resilience and security release with 420+ commits and 238 resolved issues: Raft/HA hardening with per-database quarantine and divergence self-recovery, OpenTeleme...

030

ArcadeDB @arcadedb.bsky.social · 16/06/2026

Deploy a production-ready ArcadeDB cluster on Kubernetes in one command with the official Helm chart. From zero to a single pod in 30 seconds. arcadedb.com/blog/deploy-... #k8s #graphdb #multimodel

000

ArcadeDB @arcadedb.bsky.social · 04/06/2026

ArcadeDB 26.6.1 is out! 🔒 End-to-end TLS/SSL for HA clusters 💾 Durability + crash-recovery hardening (WAL fsync on commit, gap-safe recovery) 🛡️ Security hardening + full CodeQL cleanup 280+ commits, 66 issues fixed. arcadedb.com/blog/arcaded...

arcadedb.com

ArcadeDB 26.6.1: TLS for HA Clusters, Durability Hardening & Security

ArcadeDB 26.6.1 brings end-to-end TLS/SSL for HA clusters, a deep durability and crash-recovery hardening pass across the WAL and storage layers, a broad security hardening sweep, and a long list of O...

021

ArcadeDB @arcadedb.bsky.social · 12/05/2026

🚀 ArcadeDB 26.5.1 is out! + Sparse Vector Index (BM25/SPLADE) + Server-side Hybrid Retrieval (RRF, DBSF, LINEAR) + INT8 quantization end-to-end + EXTERNAL property storage + Query partitioning 270+ commits, 128 issues closed arcadedb.com/blog/arcaded...

arcadedb.com

ArcadeDB 26.5.1: Sparse Vector Index, Hybrid Retrieval & INT8 End-to-End

ArcadeDB 26.5.1 ships a brand-new sparse vector index with server-side hybrid retrieval, INT8 quantization end-to-end, EXTERNAL property storage, query partitioning, offline HA cluster bootstrap, and ...

022

ArcadeDB @arcadedb.bsky.social · 28/04/2026

If you find a real linearizability violation, a lost write, or an isolation anomaly we missed, please open an issue: github.com/ArcadeData/arcadedb-jepsen/issues PRs adding aggressive nemeses, longer durations, or tighter checkers are very welcome.

github.com

000

ArcadeDB @arcadedb.bsky.social · 28/04/2026

Why publish this? Most graph databases (Neo4j, ArangoDB, FalkorDB, Kuzu) have never published distributed correctness work. The default is "trust the marketing page." We wanted to put up or shut up. The repo is the receipts.

100

ArcadeDB @arcadedb.bsky.social · 28/04/2026

The claim is not "provably correct." The claim is: here is what we tested, here is the harness, run it yourself, find what we missed. Repo (Apache 2.0): github.com/ArcadeData/arcadedb-jepsen Full write-up: arcadedb.com/blog/arcadedb-jepsen-tests-34-pass/

github.com

100

ArcadeDB @arcadedb.bsky.social · 28/04/2026

Beyond Jepsen: an in-house HA suite runs on every commit to the apache-ratis branch. 81 test classes. 327+ test cases. Leader election. Split-brain recovery. Dynamic membership under live writes. Leader crashes between commit phases. Snapshot corruption.

100

ArcadeDB @arcadedb.bsky.social · 28/04/2026

What we did NOT test: ❌ Multi-hour durations (most runs were minutes) ❌ Disk corruption / fsync lying ❌ Byzantine faults (Raft does not defend against these by design) ❌ Geo-replication ❌ Worst-case stack: expired lease + partition + clock skew together

100

ArcadeDB @arcadedb.bsky.social · 28/04/2026

What passed: ✅ Knossos: zero linearizability violations ✅ Elle: zero ACID isolation anomalies ✅ Zero lost writes under any nemesis ✅ Read-your-writes on followers via commit-index bookmarks ✅ Linearizable reads on demand (Raft 6.4 ReadIndex)

110

ArcadeDB @arcadedb.bsky.social · 28/04/2026

Setup: - 5 Debian nodes in Docker, Jepsen 0.3.11 - HA on Apache Ratis (Raft consensus) - 6 workloads: bank, set, elle, register, register-follower, register-bookmark - 7 nemeses: partition, kill, pause, clock skew, combined - 34 combinations

100

ArcadeDB @arcadedb.bsky.social · 28/04/2026

We ran 34 distributed-correctness tests on ArcadeDB's HA layer using the open-source Jepsen framework. All 34 passed. Caveat first: this is NOT an official Jepsen LLC analysis. It is our own use of the tooling on our own database. Repo & methodology in thread.

110

ArcadeDB @arcadedb.bsky.social · 24/04/2026

Plus: - Bolt supports TLS - Batch SQL UPDATE/DELETE/TRUNCATE - COLLATE case-insensitive indexes - Auto-fsync + startup security checklist in production - 80%+ RAM drop on Neo4j importer github.com/ArcadeData/arcadedb/releases/tag/26.4.2

github.com

Release 26.4.2 · ArcadeData/arcadedb

ArcadeDB 26.4.2 Release Notes Overview ArcadeDB 26.4.2 is a major release with over 340 commits and 100+ resolved issues. The headline is a brand-new Raft-based High Availability stack built on Apa...

000

ArcadeDB @arcadedb.bsky.social · 24/04/2026

OpenCypher fixes: - Filter pushdown for any/all/none/single - Index usage for MATCH WHERE ID(n) = <dynamic> - CREATE INDEX, CREATE CONSTRAINT IS TYPED - WALK / TRAIL / ACYCLIC path modes - Subquery fixes (EXISTS, CALL, COUNT, COLLECT, UNION ALL)

100

ArcadeDB @arcadedb.bsky.social · 24/04/2026

The new Raft-based HA stack is built on Apache Ratis. Leader election, log replication, and membership changes are delegated to a battle-tested Raft implementation. Stronger correctness for replicated writes, simpler K8s operations.

210

ArcadeDB @arcadedb.bsky.social · 24/04/2026

ArcadeDB 26.4.2 is out. 340+ commits. 100+ issues closed. Highlights: - Raft-based HA (Apache Ratis) - End-to-end Query Profiler - Major OpenCypher correctness round - Bolt + TLS - Batch UPDATE/DELETE/TRUNCATE - COLLATE case-insensitive indexes Thread below.

111

ArcadeDB @arcadedb.bsky.social · 13/04/2026

Building RAG but tired of stitching together Pinecone + Neo4j + Postgres? ArcadeDB does vector search, graph traversal, and document storage in one engine. Our Vector Search & RAG course shows you how. 22 lessons, free. arcadedb.com/academy

000

ArcadeDB @arcadedb.bsky.social · 08/04/2026

Each course ends with a real exam (not a participation trophy). Pass it and you get a verifiable certification with a unique ID. Put it on LinkedIn, your resume, your team wiki. Prove you know what you're doing. arcadedb.com/academy

arcadedb.com

ArcadeDB Academy - Free Database Training & Certification

Master ArcadeDB with free, self-paced online courses. Earn your ArcadeDB certification.

000

ArcadeDB @arcadedb.bsky.social · 08/04/2026

Migrating from Neo4j? We built an entire course for that. Step-by-step: data export, import, Cypher compatibility, BOLT driver setup, and how to unlock multi-model capabilities. Keep your Cypher. Ditch the license fees. arcadedb.com/academy

arcadedb.com

ArcadeDB Academy - Free Database Training & Certification

Master ArcadeDB with free, self-paced online courses. Earn your ArcadeDB certification.

100

ArcadeDB @arcadedb.bsky.social · 08/04/2026

The Vector Search & RAG course is especially worth checking out: - Vector embeddings + HNSW indexes - Building RAG pipelines - GraphRAG (knowledge graphs + retrieval) - Hybrid search strategies - LangChain integration All in one database, not a stack of 5 tools.

100

ArcadeDB @arcadedb.bsky.social · 08/04/2026

Why free? Most database vendors charge $500-2000+ for certification training. We think that's backwards. If you're going to build with ArcadeDB, you deserve world-class training from day one. No paywall. No enterprise sales call. Just open the browser and start.

100