About Hyperbyte

Hyperbyte Cloud builds HyperbyteDB, a time-series database for teams whose metrics have outgrown InfluxDB v1. We're based in Winchester, UK.

The problem we work on

Time-series systems fail in a predictable order. Tag cardinality creeps up — a container ID here, a device serial there — and a database that was comfortable at a hundred thousand series starts taking seconds to answer queries that used to take milliseconds. Dashboards go stale. Then someone's pager goes off at 2 a.m.

Every option at that point is bad. Drop the tags you actually needed. Downsample the data you'll want next quarter. Or discover that replication and retention policies live behind an enterprise tier and a sales call.

HyperbyteDB exists so that high cardinality is a capacity question rather than a crisis, and so that keeping more raw data costs you less hardware instead of more.

Where this came from

Fifteen years running observability and monitoring in the gaming industry, across large, globally distributed fleets. Long enough to have hit every one of those failure modes personally, and to have sat on the wrong side of the procurement call.

HyperbyteDB isn't a reaction to InfluxDB v1 being badly built — it wasn't. It's a reaction to what happens when a workload outgrows what it was designed for, and the only routes forward are a rewrite or a price list.

How we work

We publish the numbers we'd put in our own capacity plan, not the ones that look best in a deck. Our own ingest benchmark spends most of its length explaining why the headline figure is 66× higher than the number you should plan around.

The database is open source, and so are the tools built alongside it — an SLO engine, a terminal process monitor, a log tailer, a metrics plotter. Every performance claim we make is reproducible, and we've published a walkthrough of how to re-run the benchmarks against your own hardware before you believe us.

What we're building

HyperbyteDB — the database. Open source, and a drop-in for InfluxDB v1: same endpoints, same InfluxQL, same Telegraf and Grafana.

HyperbyteDB Cloud — the same database, run by us. Early access.

And sharding, so that cardinality stops having a ceiling at all.

Where to find us

Winchester, United Kingdom
[email protected] — an actual person reads it.

Some artwork on this site is by Fajar Trio Sukma from Noun Project (CC BY 3.0).

Running InfluxDB v1? HyperbyteDB is open source and free to self-host, with no feature gates. HyperbyteDB Cloud runs it for you — early access is open. See pricing, or request early access.