LOGIKASign in
On this page

Quotas and rotation

Every tier sells space: how much of your data a project holds at once. Not events, not hosts, not ingest. When you send more than the space holds, Logika keeps fewer days of history; it never stops taking your data, and it never bills you for more. This page explains how the space is counted, what gives when it runs out, and how to choose a tier.

TierSpaceHistoryAbout this much a day fills it
Free demo0.5 GB3 days0.17 GB
Solo50 GB30 days1.7 GB
Team200 GB30 days6.7 GB
Business500 GB30 days16.7 GB

What counts

Space is counted in raw bytes: your data as your application sent it, before compression. That's the size of each batch once decompressed, measured by ingest as it arrives and recorded with the batch, so the count never changes as files are compacted, merged or moved. We store everything compressed, often several times smaller, but that saving is ours to make; you're never charged by a ratio you can't predict.

  • Spans, log records and metric data points all count, the same way.
  • Each issue's stored evidence counts, since it outlives the raw data it came from.
  • Analytics don't count: they're derived, and Logika keeps them for you.

A typical declared span with six params is about 440 raw bytes, so Solo's 1.7 GB a day is close to four million of them, every day, kept for 30 days. Library spans and log lines are usually bigger; the console tells you what yours actually weigh.

Near the limit

The console's History card says how long your space lasts at the rate you're sending: "61.4/200gb over 30d · fits 95d at 2.1gb/d". An environment's rate is its average over the last 7 days, so a single busy afternoon moves it little. At 80% of the space an account notice tells you, and again at 95%, riding along on your next email.

Past the limit: rotation

Each environment keeps the smaller of two windows: its own history (the tier's 30 days, or fewer if you set fewer), and what the space allows at its rate. Every night, the nightly pass deletes each environment's oldest days beyond that window, its events, logs and metrics, from our servers and from the backup. It works in whole days and is approximate by design.

When the project's environments don't all fit, they don't all shrink alike. The environment with the shortest history gives up days first, down to a floor of 2 days, before the next one is touched; environments with the same history are trimmed together, oldest days first across all of them. So a staging environment you set to keep 14 days gives up its days before production, which keeps 30.

Say a Team project, 200 GB, sends 5 GB a day to production, kept 30 days (150 GB), and 3 GB a day to staging, kept 14 (42 GB). That fits. Then a load test pushes staging to 5 GB a day: 70 GB for 14 days, 220 GB in all. Staging drops to 10 days, and production keeps its 30. Only if staging were already at its 2-day floor would production start giving up days. Each environment trimmed below its own history raises an account item naming it, and Settings says so ("Space runs out first, so only 10/14d are kept: send less or change plan.").

Past every floor, with every environment down to 2 days, the space gives: nothing more is deleted and nothing is refused. Two days is what the evaluator needs, so issues are always judged on intact data.

What outlives the window

Rotation removes the raw rows: the days you can no longer read with logika query sql or follow with logika query journey. What Logika concluded from them stays.

  • Issues and their evidence last as long as the issue does, whatever happened to the data they came from, so a fix your agent makes next week reads the same evidence the email described.
  • Analytics last longer than the window: daily summaries for the life of the project, hourly ones 90 days, and the tables of who was active 13 months. Funnels, retention and change detection keep their history even when the raw days are gone. See Analytics.

Bursts and 429s

Ingest is limited by bytes, not requests, through a token bucket per SDK key that refills at about 5× the tier's average rate, with a bucket for the whole project above it. On Solo the average is about 19 KB a second, so a key can sustain about 96 KB a second; Team's keys about 386 KB, Business's about 965 KB. A deploy that restarts every instance at once, or a burst of retries, fits easily; a runaway loop doesn't.

A batch over the limit is answered 429 with Retry-After, and the OpenTelemetry exporter retries it within its deadline, about 30 seconds. Every bucket holds at least one maximum batch (4 MiB), so a legal batch is never refused forever. One throttled past its exporter's deadline is dropped and counted as loss, which shows in the evidence of any issue it might affect and in the weekly instrumentation feedback, never silently. In browsers and React Native, batches that couldn't be sent wait on the device, up to a week, and go on the next launch.

Never a surprise bill

The price is the tier's, flat, every month:

  • No overage fees. Sending too much shortens your history; it is never billed and never refused.
  • No billing by deletion. A missed payment never rotates your data away: the project keeps working in full for 30 days while you sort it out.
  • Your agent's runs are billed by your own agent subscription, never by Logika, and each member can cap their own with a spending limit.

Moving up a tier is free to do at any time, and the extra space lands through rotation: days already deleted stay deleted, and the window grows back as new days arrive. Moving down is allowed when the project fits: its members and pending invites within the seats, and what it stored over the last 30 days within the space.

Choosing a tier

Start with the free demo, which holds three days, for one person. Then:

  • Solo for one person and up to about 1.7 GB a day;
  • Team for up to 5 people and about 6.7 GB a day;
  • Business for any number of people and about 16.7 GB a day.

The calculator on the home page estimates your bytes from the events you'd send, with event sizes measured through the real enhancer, and prices the same numbers at other vendors; pricing has the tiers in full. After a week of real traffic, the console's History card gives you the number that matters: how many GB a day you actually send.