Idempotency key
A caller-generated string sent with a request so that a retry of the same request is recognised as a repeat and not performed twice. The server stores the key with the result of the first attempt and replays that result for every later request carrying the key. It is what stops a timed-out payment from being charged again when the client retries.
The pitfall is the key's lifetime: once it expires, the same retry becomes a new request. Pick a window longer than your longest client retry schedule, and say what it is in the API documentation.
Source: IETF HTTP API working group, The Idempotency-Key HTTP Header Field (Internet-Draft)
The first additional hire on a growing data team should be a data engineer if raw data infrastructure is the bottleneck, a data scientist if structured data already exists and business questions remain unanswered, and a data analyst when the priority is operationalizing reporting.
A take-home assignment only earns its place in your hiring funnel if it answers one specific, pre-identified question about a candidate's fit. Before it ever reaches an applicant, the hiring team must already know exactly which one to three competencies it measures, and they must have proven the task can be finished within two hours. Anything longer tests endurance, not ability.
A workable on-call rotation is not a schedule on a wall. It is a promise that someone will answer a page within minutes, even at 2am, and that the waking will be rare enough to keep judgment intact. The Google SRE Workbook, last updated in 2026, defines the hard limit: two distinct paging incidents per 12-hour shift.
The promotion to manager lands as a change in accountability, not a change in output. Your job is no longer measured by what you personally build, but by what the team delivers—and how it delivers it. The first thirty days are explicitly for learning that team, not fixing it, according to guidance published by Balancing Life's Issues. That single restraint defines everything else that follows.
In distributed computing, latency measures the time required to transmit a message from one location to another. Response time, as as defined in computer science, is the time from when a job is submitted until the last task finishes executing. A 2013 paper on the Sparrow scheduler puts a finer point on it: job response time spans from submission to scheduler through the final task completion.
Independent trade desk. We take no commission on anything we describe and run no affiliate programme of our own. Every figure on this page names the standard, filing or organisation it comes from; where a number could not be verified the page says so. How we work and how we correct. Reviewed:
Cookies, and what this site stores. The Dispatch sets no advertising or analytics cookies and loads no third-party tracker. Closing this notice writes one key — icd-notice — into your browser’s local storage, so that the notice does not return. Nothing else is kept. The one thing a page here sends onward is what a reader types into the form on the contact page, and that is described before the form is used. What the policy says.
