5 min read

Part 7: The Data Model [Nexus Newsletter #131]

Hey friends,

Today, we're continuing our series on Nexus Lore: the core concepts that come up again and again in this newsletter, on the Nexus podcast, in the Nexus Foundations course, in Nexus Pro gatherings, and in the community chatroom.

If you want to start at the beginning, check out our white paper with all 10 parts: Nexus Lore.

Lore is never written by one person, so send us your feedback for v2!

Let's jump into Part 7, our nerdiest building block...

The Data Model

Let's pause our exploration of each horizontal layer and zoom in on one key part of the Independent Data Layer: the Data Model.

In order for our applications (which we'll cover next week!) to do their thing, they're going to need context.

The applications need a way to understand all the data they're consuming, including all the underlying devices and how they fit together into a system of systems. As an example, the diagram below shows just two siloed systems (HVAC and lighting) and just two rooms of a building.

Source: Brick Schema 

See how all these devices, rooms, zones, and systems overlap with each other? Both the lighting circuit and the AHU and the VAV all “serve” room 101. The AHU and the VAV also serve room 102, which has the thermostat and the CO2 sensor that control the operation of the system.

Any application that uses HVAC data will need to know these facts—so how do we communicate those concepts in a standardized way? In most buildings today, we don't. All the silos and teams of humans represent these concepts differently.

Take naming for example: The architect might call it Room 101; it’s in the CMMS as Room_101; the BAS doesn’t even have a concept of rooms; the lighting controller calls it Zone-1 (and every controller has a Zone-1); the ACS calls it a 'Conf room'; the tenant calls it the Zeus conference room because they’ve named all their rooms after Greek gods. Crazy, right?! And naming is just one aspect of data modeling!

The key detail here is machine-readability. Most of the time we humans can infer that it's the same room, but a computer would see them as distinct concepts unless programmed otherwise. The migration to more interoperable systems comes with more machine-to-machine communication—which requires data to be understood by a computer natively. If a common data model doesn’t exist, then us humans need to map disparate data models to each other for two systems to work together.

And trust me (since I've done it): mapping is time-consuming, mind-numbing, and costly.

This is where standards come in. Using a standard ontology—like Project Haystack, Brick Schema, or Google's Digital Buildings Ontology—minimizes mapping and ensures the data model is ready for any application. However, although Project Haystack (the original) has been around for over 10 years, we're still in the early days of our industry's data modeling standards development process.

Here's a quick snapshot of where we're at today:

  • There are many standards—Haystack came first, and then Brick, and then Google added a third. The Microsoft ecosystem also added their DTDL, which other efforts have extended to cover real estate (E.g. RealEstateCore). The Passive Logic team is now creating Quantum.
  • They don't yet cover all silos and use cases—Haystack was created mostly for HVAC and metering. Brick and Google cover other silos, but there isn't one standard that covers all smart buildings use cases.
  • Most vendors have created their own—Naturally, each vendor creates their own standard to save their team time, enable their own use cases, and develop their products. Some are extensions of industry standards, most are not.
  • There are many efforts at convergence and/or alignment—Efforts like ASHRAE's proposed standard 223, the DOE's BENEFIT project, and the Ontology Alignment Project are hoping to create commonality.
  • Convergence is difficult, slow, and mostly unpaid work—This topic is a really deep rabbit hole and lots of people get lost in it, including me! The work of convergence/alignment is mostly volunteer... and it ain't fun.

Given that, should each building owner just give up and do their own thing? I don't think so...

While the status quo is messy, the north star is to start with an existing standard and continue to adapt your data modeling methods (and your IDL) as the standards evolve. That way, you're not expecting the rest of the industry to adapt to your new standard...

xkcd: Standards

Do you agree? What did I miss?

Let us know on LinkedIn,

—James Dice, Founder of Nexus Labs

P.S. Nexus Pro members can go deeper by checking out last year's panel discussion with members of Haystack, Brick, RealEstateCore, and more.

✖ At the Nexus

Here’s everything worth sharing from Nexus HQ this week:

★ PODCAST: 🎧 #104: New York City's unique RTEM programEpisode 104 is a conversation with Ryan Baxter, PropTech Advisor to the New York State Energy Research & Development Authority, AKA NYSERDA.

We talked about how New York City real estate is unique, the top 5 pitfalls for technology vendors trying to get a foothold there (and anywhere), the past, present, and future of NYSERDA’s Real Time Energy Management Program, AKA RTEM, and finally, the exciting Hackathon initiative the RTEM launched this year.


★ FROM THE ARCHIVE: Pro Member Gathering with all the Major Ontology Efforts (Pro members only)



  • Subject Matter Expert Workshop: CEO at BuiltSpace Technologies Corporation, Rick Rolston will present on Motor Monitoring
    and Analytics.
  • Subject Matter Expert Workshop: Co-Founder and CTO at Senseware, Julien Stamatakis, will join us as the subject matter expert leading the conversation on how Real-Time IAQ Data Drives Modern Building Management.
  • Member Gathering: James and Pro member Brian Vaughn of Cushman Wakefield will discuss the key ways technology is transforming building operations. James and Pro member Andrew Knueppel, also of Cushman Wakefield, will discuss the three categories of MSIs in the marketplace today. Plus, breakout rooms for networking!

Join Nexus Pro now to get the invites and access to the recordings.


★ ON LINKEDIN: Is AI another layer in the stack?


★ READ OF THE WEEK: Rob Speyer on the long arc of megadevelopment, New York’s renaissance and proptech bets


👋 That's all for this week. See you next Tuesday!