Live Architecture Review with Dylan DuFresne

Abelara Lead Architect Dylan DuFresne walks Zack Scriven through the multi-site industrial automation stack Abelara planned and deployed for its Prove It 2026 sponsorship — naming specific vendors at every layer of the stack and explaining the architectural reasoning behind each choice. Starting at the edge (Opto 22 controllers), moving through site-level infrastructure (Ignition, TimescaleDB, Flow Software, TimeBase), into data operations (HighByte, Litmus Edge), across the enterprise UNS (HiveMQ broker), and up through multi-site enterprise applications (Fuuz, MaintainX, Google Cloud), Dylan builds the architecture layer by layer and takes live audience questions throughout. The through line isn't the specific tool list — it's the *why* behind each decision. Why a separate tag server in Ignition 8.3 for DevOps isolation. Why HighByte for data in motion but Flow Software for data at rest. Why vision for SCADA but perspective for MES. Why the blue-vs-red namespace distinction matters. Why Fuuz sits where it sits. The result is one of the clearest end-to-end walkthroughs of a real, deployable Industry 4.0 stack you'll find. Three things to take away 1. "Edge" means whatever the vendor wants it to mean — know what they mean.** Dylan's working definition: *"Edge is shorthand for as far into the stack as I care to look."* To a cloud vendor, the entire plant is the edge. To an OT engineer, it's the PLC. When a product labels itself "edge," you have to ask which edge. 2. HighByte for data in motion, Flow Software for data at rest.** The two tools live at the same layer of the stack, but do different jobs. HighByte is the ETL / data ops engine — pulling data from disparate sources, modeling it, moving it. Flow Software is the analytics engine — pulling data at rest from multiple stores and calculating KPIs against it. Not an "or" decision; an "and" decision. 3. Fuuz is to the enterprise what Ignition is to the site.** If Ignition is the Swiss Army knife platform for the plant floor — SCADA, MES, IoT connectivity — Fuuz is the same kind of platform for L3/L4 enterprise apps: iPaaS foundation, MES, WMS, supply chain. Different jobs, different data models, both Swiss Army knives at their respective layers. The stack, in one pass - Device layer (L0/L1): Opto 22 controllers - Edge connectivity: Ignition Edge (IOT licenses), running on bare-metal Linux or Docker - Site SCADA: Ignition Vision (for local multi-monitor operator screens) - Site MES: Ignition Perspective (for browser-based MES frontends) - Site tag server: Ignition 8.3 dedicated tag server — single source of truth for site current state - Historian: TimeBase (Flow Software) — telemetry, scalar data, in-flight time series - MES database: TimescaleDB on Postgres — contextual event data, state changes, tabular MES functions - Analytics: Flow Software — connects all the above, calculates KPIs, contextualizes - Data ops (Site A): HighByte — ETL from tag server to UNS broker - Data ops (Site B): Litmus Edge — standalone data pipeline, no local SCADA/MES needed - Enterprise UNS broker: HiveMQ (the Prove It shared broker) - CMMS: MaintainX - Enterprise apps / L3: Fuuz — MES, WMS, supply chain, ERP integration, multi-site consistency - Data warehouse: Google BigQuery / Google Cloud - Deployment: Portainer Timestamps:00:00 — Intro + the planned stack05:11 — The credibility moment: "All 8 title sponsors, by coincidence"07:45 — Edge devices, Opto 22, and why IPCs run Linux13:00 — "Edge is shorthand for as far into the stack as I care to look"15:16 — The Ignition 8.3 tag server18:02 — TimeBase vs. TimescaleDB, MES core vs. MES custom24:51 — SCADA in Vision, MES in Perspective — why27:31 — Medallion architecture, HighByte, and the enterprise UNS34:34 — Fuuz as the enterprise Swiss Army knife45:09 — Q&A