Skip to content
EzraDB

REAL-TIME ANALYTICS

Real-time & user-facing analytics

Dashboards and in-product analytics on data that is seconds old — served from the same database that takes the writes.

THE STATUS QUO

The usual setup: OLTP in Postgres, a warehouse behind an ETL pipeline, and "real-time" dashboards that are an hour stale. Serve analytics from the transactional database instead and it melts under scan load.

HOW EzraDB DOES IT

Writes hit the row store; dashboard scans read the columnar side of the same table. One copy of the data, no pipeline to operate or lag behind.

The engine that finishes all 22 TPC-H queries and beats general-purpose engines 3–5× keeps user-facing queries interactive, not batch.

Counters, leaderboards, and rollups update in under a second as events land — differential dataflow, not nightly refreshes.

Hot counters and session state live beside the source of truth — no separate cache tier to keep coherent.

BI tools and data apps pull columnar Arrow batches straight out on port 5434 — no row-by-row driver overhead.

IN PRACTICE

Fresh analytics is not a pipeline problem. It is a database problem — solved in the engine.

one table, both workloads
CREATE TABLE accounts (id INT PRIMARY KEY, owner TEXT, balance NUMERIC) USING row;    -- OLTP
CREATE TABLE events   (id INT, payload TEXT)                            USING column; -- OLAP
CREATE TABLE ticks    (ts TIMESTAMPTZ, sym TEXT, px NUMERIC, vol BIGINT) USING hybrid; -- HTAP