nOps, a Databricks Constructed On associate managing over $4 billion in annual cloud spend, migrated their manufacturing utility to Databricks Lakebase. The end result was a sooner, less complicated structure that eradicated the glue between their app and their analytics, and a playbook for ISVs seeking to do the identical.
Each ISV constructing on Databricks finally hits the identical architectural crossroads: your analytics stay within the Lakehouse, however your utility wants a relational database for low-latency reads and writes. So that you bolt on a separate Postgres occasion (possibly RDS, possibly one thing self-managed) and immediately you are sustaining ETL pipelines, cron jobs, and change-detection logic simply to maintain two methods in sync.
nOps lived that actuality for years. After which they discovered a greater means.
nOps: Automating Cloud Financial savings at Scale
For these unfamiliar, nOps is an automatic cloud value optimization platform that manages commitment-based reductions throughout AWS, GCP, and Azure. Their strategy is distinctly “always-on.” They monitor, buy, and alternate cloud commitments on an hourly foundation, utilizing machine studying to steadiness efficient financial savings charges in opposition to dedication lock-in threat. The mannequin is performance-based: nOps solely expenses a proportion of the incremental financial savings they generate.
It is a data-intensive operation. Each hour, nOps analyzes utilization patterns throughout hundreds of buyer accounts, evaluates dedication portfolios throughout three main cloud suppliers and dozens of providers, and makes automated buying choices. On high of that, they floor value visibility, forecasting, and anomaly detection via a centralized FinOps platform.
The analytical spine for all of this has lengthy been Databricks Lakehouse. However the front-end utility, the platform clients log into to see their financial savings, handle budgets, and discover value knowledge, wanted one thing extra.
The Drawback: Two Worlds, Loosely Linked
nOps’s earlier structure was a well-recognized sample for ISVs on Databricks. Superior analytics and metric computation ran within the Lakehouse. Buyer-facing knowledge (account configurations, person preferences, quickly altering client-specific state) lived in a separate relational database powered by third-party distributors and homegrown options.
The seams between these two methods created actual friction. Scheduled jobs and cron-based change detection had been required to maintain the front-end database and the Lakehouse in sync. Information that was “stay” in a single system would possibly take minutes or longer to look within the different. And the operational overhead of managing a separate database stack, with its personal scaling, backup, and safety considerations, pulled engineering time away from what nOps really does finest: constructing dedication automation.
When nOps expanded from AWS-only to multi-cloud protection throughout GCP and Azure in early 2026, the rising workloads strained this structure. The workforce determined to rebuild the platform, this time specializing in their specialty and selecting infrastructure that merely works.
The Determination: Why Lakebase
nOps chosen Databricks Lakebase, a completely managed PostgreSQL database built-in instantly with the Lakehouse, because the OLTP spine for his or her new platform.
Jordan Stein, Director of Product at nOps, pointed to 3 elements that made Lakebase the suitable match:
- Tight coupling to the Lakehouse. This was the largest issue. With Lakebase, nOps’s knowledge engineering groups can instantly entry incessantly altering buyer knowledge from their Lakehouse pipelines with out scheduled jobs, crons, or lag. As Jordan put it: “We’re speaking scheduled jobs that needed to run, crons which might be coming and choosing up these adjustments, whereas now we all know that the second it is stay, we will eat it. This has been a recreation changer for us.”
- Auto-scaling and auto-stop. Even with aggressive auto-stop settings throughout improvement, the nOps workforce was “shocked by the efficiency.” Lakebase’s serverless compute adjusts to workload calls for and scales to zero when idle, which issues for a cost-optimization firm that practices what it preaches.
- Ease of adoption. Level-in-time restore has already confirmed helpful. Versatile OAuth roles simplify entry management. And since Lakebase lives inside the Databricks workspace, their groups are working in a platform they already know. No new device to be taught, no separate console to handle.
The Structure: One Platform, Tightly Built-in
This is what nOps’s new structure seems to be like:
Lakebase serves because the central Postgres database and single supply of fact for each the front-end utility and their AI infrastructure.
Databricks Lakehouse constantly consumes knowledge from Lakebase for evaluation and metric computation.
The nOps platform robotically discovers and surfaces Databricks Metric Views, so standardized metrics computed within the Lakehouse present up constantly within the front-end.
Information flows in a single path, from Lakebase into the Lakehouse for analytics, with no direct write-back wanted. This retains the structure clear and the supply of fact unambiguous.
The remainder of the stack follows the identical strategy: Vercel for internet hosting and observability, WorkOS for authentication, and Databricks for every little thing knowledge.
Hear It from nOps
Jordan Stein not too long ago walked via the complete nOps Lakebase migration story in a associate highlight presentation. Watch the video to listen to how the transition went, what stunned them about efficiency, and the way the Lakehouse integration modified their knowledge engineering workflows:
The ISV Playbook: Why Lakebase Modifications the Recreation
nOps’s story is not distinctive. Practically each ISV constructing on Databricks faces the identical OLTP-meets-analytics pressure. What’s value taking note of is how cleanly Lakebase resolves it.
Remove the sync tax. The most costly code in any ISV’s stack is commonly the code that strikes knowledge between methods. Lakebase’s native integration with Unity Catalog and one-click Delta Lake sync replaces customized ETL pipelines with managed infrastructure. That is engineering time you get again.
One governance mannequin. When your OLTP database is registered as a Unity Catalog asset, you get unified governance, lineage, and entry management throughout operational and analytical knowledge. No extra managing safety insurance policies in two locations.
Postgres compatibility means zero rewrite. Lakebase is absolutely managed PostgreSQL. Your present libraries, ORMs, and SQL instruments work out of the field. Extensions like pgvector and PostGIS are supported. You migrate by pointing your app at a brand new connection string, not by rewriting queries.
Scale economics that make sense. Utilization-based pricing with scale-to-zero means you are not paying for idle capability. For ISVs with variable workloads (and which ISV does not have variable workloads?) this instantly impacts unit economics.
Ship sooner. When your utility database and your knowledge warehouse are the identical platform, a complete class of integration work disappears. Your workforce ships options as an alternative of sustaining plumbing.
Early Adopters, Actual Impression
nOps is an effective instance of what an modern Constructed On associate seems to be like. Slightly than ready for Lakebase to mature via a number of launch cycles, they acknowledged the architectural match early, dedicated to a manufacturing migration, and are already seeing outcomes: sooner knowledge pipelines, decrease operational overhead, and a greater expertise for his or her clients.
That willingness to maneuver early is strategically sensible too. By constructing on Lakebase now, nOps has a tighter integration with the Databricks platform than rivals who’re nonetheless duct-taping separate database stacks collectively. Their platform is easier to function and sooner to increase.
Get Began
Discover Lakebase. Should you’re an ISV constructing on Databricks, or contemplating it, be taught extra about Lakebase and the way it can simplify your structure.
Discover nOps. In case your group is seeking to scale back cloud prices throughout AWS, GCP, or Azure with out the dedication threat, go to nOps to see how their automated optimization platform, now powered by Databricks Lakebase, can assist.
