Relait Architecture

Ok, what does the Relait architecture look like? Many stakeholders in the enterprise have first class concerns about introducing new software or technologies into their application stack. These concerns include security, privacy, system integrity, and supportability, among others. We’ve been thinking about these concerns from day one, and made key decisions along the way to address them.

We're happy to be transparent about our architecture and the thinking behind it. We use familiar, robust, and demonstrably secure technology throughout our solution. We won’t share our trade secrets here, but below is a diagram of our architecture:

The first thing to point out is that all of this is deployed in your (the customer) infrastructure. This is not a SaaS. We don’t hold, nor do we have access to, your data. There is no dial home capability. All our code is available to our customers for inspection (subject to a Non Disclosure Agreement).

The clever stuff happens in the control framework. It generates low impact scans to execute against your source data. It does this in a unique, privacy preserving way, making it impossible to reconstruct customer records in a re-identification attack. This generated metadata is sent over to the Relait Data Repository, where the heavy lifting data processing runs, and exposes the relationships in the data.

Note: this is all done by leveraging and honouring your existing RBAC frameworks. The diagram shows Active Directory, but we are agnostic about the solution in your enterprise.

We believe that metadata comes to life when it is visualised. We provide interactive visualisations of the relationships in your data. These results are not based on documentation or design decisions. They are based on what data is shared. What is actually shared, not what is supposed to be there.

But Relait can import your documented rules as well - and we recommend you do so. This is useful so you can compare the ideal world with the current state, and show exactly where you need to focus to get to the future state.

We provide Apache Superset as the main visualisation capability. This choice is driven by a few things. First, it is completely open source. No hidden premium enterprise features. Second, it has out-of-the-box support for RBAC through integrations with Active Directory and others - plus all code can be inspected. Third, it has powerful graph visualisation and interactivity options. Most of its peers don’t come close on this third reason (I’m looking at you - Power BI, Tableau, Qlik, Microstrategy!). None of them (from what I can see) can do all three.

Hopefully you agree that the architecture is pretty straightforward. If you want to learn more, get in touch. We’re happy to share more about the architecture and reasons behind it.

Edit: 11th April 2024
This blog was originally published on 13th March 2024. Today we updated it to show the new, simpler architecture for Relait. All functionality that previously required a Neo4j database has been migrated to SQL Server. This removes infrastructure complexity and costs, and minimises the data footprint while maintaining enterprise grade RBAC.