The GizmoData Dispatch — April 2026 Edition
Momma didn't raise no April fool!
Happy April, GizmoData community! 🙋
March was about meeting your users where they are on the desktop (Windows, Power BI, OAuth everywhere). April is about something we couldn’t have pulled off without last month’s groundwork:
GizmoSQL now fits in your pocket. 📱
Let’s get into it.
🔎 What is GizmoSQL?
New here? GizmoSQL is an open-source SQL engine that puts the speed of DuckDB behind a proper server. It uses Apache Arrow Flight SQL as its wire protocol, so data moves between the server and your tools in columnar Arrow format — no row-by-row serialization overhead.
Think of it as “DuckDB as a service”: DuckDB’s blazing-fast analytics, accessible over the network from any language, BI tool, or notebook that speaks JDBC, ODBC, ADBC, or Flight SQL. Free and open-source (Apache 2.0), with an optional Enterprise Edition for teams that need OAuth/SSO, audit trails, catalog filtering, and fine-grained access control.
📱 GizmoSQL on iOS — Live on the Apple App Store
The headline of April: GizmoSQL is now available on the Apple App Store.
A full GizmoSQL server, running on your iPhone or iPad, with an embedded Arrow Flight SQL client built into the app. It’s a complete analytics stack in your pocket:
Server tab — start/stop a local GizmoSQL server with a tap, view live logs, manage self-signed TLS certificates, see your WiFi network status for LAN access
Query tab — a terminal-style SQL client with syntax highlighting, box-drawing table output, query history, and an automatic 500-row safety limit
Browser tab — navigate catalogs, schemas, tables, and columns with instant data previews
File sharing — export your DuckDB database out of the app via the Files app
Because Apple requires DISABLE_EXTENSION_LOAD for App Store compliance (Guideline 2.5.2), we statically link the extensions iOS users care about most:
🐘
postgres_scanner— “Postgres in your pocket.”ATTACH 'postgresql://…' AS pgfrom your phone.🦆
ducklake— A lakehouse in your pocket. Query a shared DuckLake catalog directly from mobile.httpfs,aws,azure,parquet,json,iceberg, and friends — all bundled.
On Linux/macOS/Windows, these remain normal runtime-loadable extensions. The iOS-only static linking was the missing piece that let us ship.
🚀 …and yes, it actually flies.
To prove it’s not a toy, we pointed the full TPC-H 1 GB benchmark at a phone — and not politely, either. The demo video shows 25 concurrent Arrow Flight SQL client sessions, each independently running the complete 22-query TPC-H suite against a single GizmoSQL server instance running on the device. No warm-up tricks, no query caching — just the phone doing real analytical SQL work under realistic multi-user load.
550 analytical queries, executing in parallel, on a phone. That’s the kind of thing that used to require a data warehouse cluster.
💡 If GizmoSQL can do this on an iPhone — pause for a second and picture what it can do turned loose on enterprise-grade hardware. A beefy cloud VM. A rack-mounted server on-premises. A fleet of GPU-adjacent compute nodes in your data center. The phone is the punchline; the enterprise deployment is the feature film. 🎬
🎬 Watch it run: TPC-H Benchmarks on GizmoSQL iOS, Fast
📝 A quick note on where iOS fits
To set expectations: the iOS edition is not intended for production workloads. It’s designed for learning, development, and experimentation — a beautiful way to explore GizmoSQL, prototype queries on the go, demo it to a colleague over coffee, or teach someone what Arrow Flight SQL feels like. iOS deliberately sandboxes what apps can do (background execution limits, no dynamic extension loading, memory caps, thermal throttling), so it’s not the right home for your nightly ETL.
For production, reach for the enterprise-grade editions:
🐧 Linux — the production default; scales from a laptop to a 96-core bare-metal box or a Kubernetes pod
🍎 macOS — native arm64 and x86_64 builds; great for team workstations and CI runners
🪟 Windows — with the signed MSI installer we shipped in March, zero-friction for Windows-native shops
☁️ GizmoData Cloud (coming soon) — fully managed GizmoSQL with OAuth/SSO, autoscaling, and built-in observability already wired up
Think of iOS as the tasting menu, and Linux/macOS/Windows/Cloud as the full kitchen.
🦆 DuckDB gave us a shout-out
The DuckDB team published a write-up about the launch on their website:
👉 Introducing the GizmoSQL iOS App
Thank you to the DuckDB Labs folks — it means a lot to have the mothership signal-boost this.
🎨 dbt-gizmosql: Dramatic Adapter Improvements
Get it: GitHub • PyPI — pip install dbt-gizmosql
March shipped the adapter. April made it production-grade. A steady stream of improvements tackled concurrency, correctness, developer ergonomics — and unlocked server-side pushdown for Python models, which is the standout feature of the month.
🚀 Server-side pushdown for Python models — a new
session.remote_sql()helper lets your Python models execute SQL directly on the server and receive results as Arrow. The heavy lifting stays on GizmoSQL where the data lives, instead of dragging a billion rows into your Python process just to aggregate them. (v1.11.15)💾 Write results straight to files or object storage — a new external materialization writes model output via server-side
COPYto local files, S3, Azure Blob, or GCS. No more pulling large result sets back through the client just to save them. (v1.11.13)🔁 Smarter incremental + snapshot merges — incremental and snapshot materializations now use
MERGE BY NAME, which gracefully survives column additions, reorderings, and schema drift. Ships with dramatically expanded test coverage. (v1.11.12)⚡ True parallel dbt runs — removed a stray
CHECKPOINTthat was serializing concurrent writers. Thanks to community contributor @airhorns for the report and reproducer. (v1.11.9)🛡️ Fixed a remote-instance catalog race — concurrent
dbt runthreads could corrupt the shared catalog metadata cache on remote GizmoSQL instances. No longer. (v1.11.10)🐍
dbt showno longer crashes on Python models — a long-standing papercut is gone. (v1.11.14)🧹 Simpler, cleaner adapter internals — removed a client-side workaround now that the underlying fix has landed upstream. (v1.11.11)
🏷️ GizmoSQL Server — Tagging, Instrumentation & Reliability
🏷️ Instance, session, and query tagging for instrumentation (Enterprise) — attach arbitrary JSON metadata to every instance, session, or individual query for cost attribution, multi-tenant identification, and observability. Slice your Grafana dashboards by team, tenant, environment, or scheduled job with zero extra plumbing. Existing instrumentation databases migrate automatically. 📖 See the docs (v1.21.0)
🦆 DuckDB upgraded to v1.5.2 — latest performance and correctness improvements from the DuckDB team, for free. (v1.21.1)
🛠️ Bulk ingest now works inside open transactions —
adbc_ingest()no longer fails when a client usesautocommit=False. Thanks to community member @fromm1990 for the detailed report and reproducer. (v1.21.2, #155)📱 Cross-instance tokens & session-count API — two small pieces of plumbing that made the iOS app possible, and also help load-balanced deployments survive session rebalancing.
☕ JDBC Driver — Cleaner Errors, Stable Download Links
Get it: GitHub • Latest Release
🔑 Clear, upfront credential errors — missing or partial credentials are now rejected early with a readable message, instead of failing cryptically mid-query. (v1.5.2)
🧩 Fixed duplicate columns in empty result sets — a subtle metadata bug that surfaced in BI tools that introspect schemas before fetching rows is now resolved. (v1.5.5)
🔗 Stable download URL — every release now ships an unversioned
gizmosql-jdbc-driver.jaralias alongside the versioned jar, so download links in internal docs and dashboards don’t rot when you upgrade.📝 Richer release notes — auto-extracted from
CHANGELOG.mdgoing forward.
☁️ Coming Soon: GizmoData Cloud
The managed GizmoSQL platform is now in private preview with our first customer, and continues to take shape. A preview of what’s coming:
One-click GizmoSQL provisioning — spin up fully configured instances with OAuth/SSO, TLS, OpenTelemetry, and instrumentation already wired in. No YAML, no Helm charts, no Terraform.
Elastic autoscaling — ephemeral sessions scale with demand, fronted by a local-NVMe cache layer for fast repeat reads against object storage.
Built-in observability — every instance emits traces and metrics into a managed Grafana + Tempo + Prometheus stack, so you get query-level visibility out of the box.
Multi-tenant catalog & project scoping — organize your GizmoSQL instances by project and environment, with unified secrets management and scope-aware access control.
Bring-your-own-cloud ready — the control plane is designed to provision into customer AWS accounts, not just ours.
We’ll have more to share as we open up the preview to additional customers. Want to be on the list? Contact us here.
🌐 Website, Videos, and Community
🎬 Introducing the New GizmoSQL iOS App — a walkthrough of the new iOS edition.
🚀 TPC-H Benchmarks on GizmoSQL iOS, Fast — yes, TPC-H running on a phone.
📱 iOS added to the “GizmoSQL Runs Everywhere” platforms — Linux, macOS, Windows, iOS.
🤝 Starlake joined as a GizmoData Technology Partner (welcome to the family!).
📖 Ongoing contributions to the Columnar Tech
adbc-quickstartsproject.
📅 Upcoming Events
🎤 Live Online — Introducing GizmoSQL for Power BI Users
We’re partnering with our friends at Fabricent for a joint live online event:
🎤 Introducing GizmoSQL for Power BI Users
ADBC Driver • Performance • Open-Source Innovation
🗓️ Tuesday, May 12, 2026
🕒 3:00 PM GMT-3 (2:00 PM – 3:00 PM your local time varies — check the calendar invite)
💻 Online — hosted on Microsoft Teams
👥 Hosted by: Fabricent • Gustavo Leo • GizmoData • Philip Moore
Gustavo Leo (Fabricent) and Philip Moore (GizmoData) will walk through how the new GizmoSQL ADBC driver brings blazing-fast, open-source analytics into Power BI — with a live demo and plenty of time for Q&A. Already 145+ attendees registered — come join us!
🌉 In Person — IoT Tech Expo North America 2026
GizmoData is a featured partner at IoT Tech Expo North America, co-located with AI & Big Data Expo, Cyber Security & Cloud Expo, Edge Computing Expo, Digital Transformation Week, Intelligent Automation & Robotics, and Data Center Congress — one of the largest enterprise tech events of the year.
🗓️ May 18–19, 2026
📍 San Jose McEnery Convention Center, California
🏷️ Find us at Stand 41
If you’re going to be in the Bay Area that week, come say hi — we’d love to chat GizmoSQL, DuckDB, edge analytics, or just swap stories. Drop us a note if you want to pre-arrange a meeting.
🏢 GizmoSQL Enterprise Edition — Pricing Reminder
Core Edition: Free forever (Apache 2.0)
Enterprise Edition: $500/month per server — OAuth/SSO, catalog visibility filtering, DuckLake-backed audit, instance/session/query tagging, KILL SESSION, multi-instance OAuth routing
SLA-driven enterprise 24x7 support is available as an add-on option for Enterprise Edition customers
Enterprise+: Custom pricing for volume discounts and SLAs
💡 Looking Ahead
GizmoData Cloud early access — we’re lining up the first external tenants
Power BI certification — still on the path; iOS was the April detour, Power BI is the May focus
More dbt adapter goodness — a few community-requested features are in flight
Continued DuckLake integration — both on the server and, now, on the phone
That’s a wrap for April 2026. Download the iOS app, try the new external materialization in dbt-gizmosql, and let us know what you build.
Questions, feedback, or want a demo? We’d love to hear from you. 🚀
— The GizmoData Team