{"id":3068,"date":"2026-11-16T22:20:00","date_gmt":"2026-11-16T22:20:00","guid":{"rendered":"https:\/\/www.surgeryplanet.com\/blog\/?p=3068"},"modified":"2026-11-19T17:20:00","modified_gmt":"2026-11-19T17:20:00","slug":"serverlessops-team-story","status":"publish","type":"post","link":"https:\/\/www.surgeryplanet.com\/blog\/serverlessops-team-story\/","title":{"rendered":"Replay Friday: A Coordination Team Learns ServerlessOps"},"content":{"rendered":"<p>Let&#8217;s tell a story about an imaginary team. Picture a small group at a travel-and-surgery coordination service, five people who handle enquiries from patients planning care abroad. Their job is administrative: receive an enquiry, send documents to the right partner clinic&#8217;s coordinators, track replies, remind everybody about deadlines. Everything here is invented to make a point, and none of it concerns treatment. The point is about ServerlessOps, the quiet craft of keeping event-driven software dependable, which this team learned the slightly hard way.<\/p>\n<h2>The setup: a tidy little machine<\/h2>\n<p>The team&#8217;s tooling was built from small cloud functions. A new enquiry fired one function that created a case file. Uploading a document fired another that notified the partner. A daily timer fired a third that nudged anyone who hadn&#8217;t replied. It worked beautifully for a year, which, as engineers like to say, is how trouble starts.<\/p>\n<h2>The Tuesday of the double emails<\/h2>\n<p>One Tuesday, a patient&#8217;s family wrote in confusion: they had received the same &#8220;documents received&#8221; message six times. Nobody had changed anything. The team&#8217;s lead, Priya (a name made up for the story), opened the logs and found that the notifying function sometimes timed out after sending but before noting that it had sent. The platform dutifully retried.<\/p>\n<p>&#8220;So it isn&#8217;t broken,&#8221; she said. &#8220;It&#8217;s doing what it was built to do. We just never built for repeats.&#8221;<\/p>\n<p>That sentence, more or less, is the first lesson of the discipline. The Ops Atlas describes the same idea in its entry on <a href=\"https:\/\/www.xopsschool.com\/serverlessops\/\">operating serverless workloads<\/a>, where it sets out how a design that records the action against a real-world business identifier stops a repeat from doing harm. The team chose the case number plus the document name as the identifier, wrote a note to a small table before sending, and checked that table first. The six emails became one.<\/p>\n<h2>The Friday of the missing replies<\/h2>\n<p>The second lesson arrived on a Friday evening before a long weekend. A partner coordinator reported that nobody had answered their question since Wednesday. The team looked, and discovered that the reply-routing function had been failing for two days. Failed items had been sliding into a holding queue that was set to keep them for only as long as the source did, so some had already expired.<\/p>\n<p>The fix had two parts. They gave the queue an owner, which was Tomas (another invented name), and an alarm tied to how long the oldest item had waited. They also lengthened the queue&#8217;s retention beyond the source&#8217;s, so that a weekend could never again swallow evidence. The Atlas gives a rule of thumb along these lines, and the principle, not the exact number, is what the team took away.<\/p>\n<h2>The Monday of the slow partner<\/h2>\n<p>The third problem was subtler. One partner&#8217;s document portal slowed down on a busy Monday. The team&#8217;s functions did not crash; they waited, and a rising number of them ran side by side, each holding a connection open. The portal, already struggling, fell over entirely, and the cloud bill that month came with an unpleasant surprise.<\/p>\n<p>Priya&#8217;s answer was to cap how many copies of the function could run at once, keeping the number comfortably under what the portal could handle, and to shorten the time the function would wait on any single call. She also added &#8220;cost per thousand cases&#8221; to the weekly team sheet, next to success rate. The team discovered that a sudden cost bump was often the first sign that a dependency had slowed.<\/p>\n<h2>The rehearsal that changed the culture<\/h2>\n<p>After those three episodes, Tomas proposed an afternoon exercise. They replayed a batch of test enquiries twice, on purpose, while a dummy partner endpoint was made slow. They then counted duplicate emails, duplicate case files and stuck items. The first run found two leftover problems. The second run found none.<\/p>\n<p>In the Atlas&#8217;s maturity ladder, that sort of rehearsal with zero double effects marks the &#8220;safe&#8221; stage for critical paths. The team did not use the vocabulary at first; they just called it Replay Friday and put it in the calendar once a month.<\/p>\n<p>Something else changed too. Meetings got shorter. When a partner complained, the team no longer argued from memory; they opened the one-page sheet, read off the numbers and agreed the next step in minutes. Priya later said the biggest gain was emotional: the Friday dread had gone, because every failure now had a place to land and a person to look at it.<\/p>\n<h2>What the story teaches a patient-facing service<\/h2>\n<ul>\n<li><strong>Repeats are routine.<\/strong> Design every step so a second run does nothing new.<\/li>\n<li><strong>Failures need a keeper.<\/strong> An unwatched queue is lost mail with extra steps.<\/li>\n<li><strong>Surges need a ceiling.<\/strong> Protect the slowest partner, since it will fail first.<\/li>\n<li><strong>Measure what matters.<\/strong> Real completions, duplicates, age of stuck items and unit cost.<\/li>\n<li><strong>Practise.<\/strong> A rehearsal is cheaper than a bad Friday.<\/li>\n<\/ul>\n<h2>Why this matters to people planning care abroad<\/h2>\n<p>Travellers considering surgery overseas depend on a chain of messages, documents and deadlines. When the administrative layer is sloppy, families feel it as confusion and delay, even if the clinical side is excellent. That is why we encourage readers to ask practical questions of any service, as discussed in <a href=\"https:\/\/www.surgeryplanet.com\/blog\/essential-questions-to-ask-before-your-surgery-abroad\/\">our list of questions to raise before travelling for an operation<\/a>. Another helpful read is <a href=\"https:\/\/www.surgeryplanet.com\/blog\/choosing-with-care-how-hospital-comparison-protects-your-surgery-journey\/\">how careful comparison of hospitals protects a surgery journey<\/a>, which treats the planning stage with the same patience.<\/p>\n<h2>Epilogue<\/h2>\n<p>A year later, Priya&#8217;s team still has Replay Friday. New hires get a one-page map of every function: trigger, effect, owner, limits. The service has not become glamorous. It has become predictable, and for a service whose customers are anxious, predictable is the feature that counts.<\/p>\n<p>This tale is illustrative only. It is not medical, legal or travel advice, and it makes no claim about any particular provider, hospital or service.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>An invented coordination team meets double emails, missing replies and a slow partner, and learns ServerlessOps habits that make their service predictable.<\/p>\n","protected":false},"author":13,"featured_media":3069,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[1582],"tags":[],"class_list":["post-3068","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-medical-tourism-tech"],"_links":{"self":[{"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/posts\/3068","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/users\/13"}],"replies":[{"embeddable":true,"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/comments?post=3068"}],"version-history":[{"count":1,"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/posts\/3068\/revisions"}],"predecessor-version":[{"id":3192,"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/posts\/3068\/revisions\/3192"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/media\/3069"}],"wp:attachment":[{"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/media?parent=3068"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/categories?post=3068"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/www.surgeryplanet.com\/blog\/wp-json\/wp\/v2\/tags?post=3068"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}