ETA Calculator

The arithmetic is trivial, the assumption is not

An ETA is one division: remaining work divided by rate. If 40,000 rows are left and you are managing 500 rows a minute, you have 80 minutes to go. This calculator does that division, converts the answer into readable units and adds it to your start time so you get an actual clock time rather than a duration you then have to add up in your head.

The hard part is not the division. It is that the division assumes the rate you measured will hold for the rest of the run, and that assumption is wrong more often than it is right. Bulk imports slow down as indexes grow. Downloads speed up once TCP finishes ramping. Batch jobs hit a stretch of unusually large records. An ETA is a projection of the present, not a prediction of the future, and it is worth saying that out loud before anyone writes the number into a status update.

Pick the rate window deliberately

Because the estimate inherits whatever rate you feed it, the window you measure over is the most consequential input on the page. A rate taken from the last ten seconds is responsive and jumpy: it will swing wildly and give you a different ETA every time you look. A rate averaged over the whole run so far is stable but slow to react, and it will keep quoting an optimistic number long after the job has bogged down.

For anything short, use the whole-run average. For a job measured in hours, a rolling window of a few minutes usually strikes the right balance: recent enough to notice a slowdown, long enough not to react to every hiccup. If the two numbers disagree sharply, that disagreement is itself the useful signal - it means the rate is changing, and the honest answer is a range rather than a single timestamp.

Read the total duration alongside the remainder

The tool reports total duration as well as remaining time, and the pair is more informative than either alone. If a job is 5% done and the total duration reads eleven hours, you know immediately that this is not something to sit and watch, regardless of what the remaining figure says. It also gives you a sanity check: if the projected total is wildly different from how long the same job took last week, either the rate measurement is wrong or something has genuinely changed, and both are worth investigating before the run finishes.

Progress percentage is shown for the same reason. Early in a run, when the completed count is small, the rate estimate rests on very few samples and the ETA is close to meaningless. Watching the percentage climb tells you when the estimate has enough data behind it to be worth quoting.