{"id":1331,"date":"2026-11-16T04:20:00","date_gmt":"2026-11-16T04:20:00","guid":{"rendered":"https:\/\/bhaso.com\/blog\/?p=1331"},"modified":"2026-11-18T08:20:00","modified_gmt":"2026-11-18T08:20:00","slug":"cloudops-one-account-story","status":"publish","type":"post","link":"https:\/\/bhaso.com\/blog\/cloudops-one-account-story\/","title":{"rendered":"One Account, Six Engineers: A CloudOps Story in Five Chapters"},"content":{"rendered":"<p>This is an invented story, assembled to show how a pattern plays out. A six-person team runs a ticketing and travel-booking product for a regional operator. Nobody on it has the word &#8220;cloud&#8221; in their job title, yet for three years they have been the people who decide how the company&#8217;s cloud estate is organised. They did not plan it that way. The story below follows what happened when their CloudOps habits, or the lack of them, were put to the test.<\/p>\n<h2>Chapter one: everything lives in one account<\/h2>\n<p>The team started the way most teams start. One cloud account held the website, the reporting jobs, the experiments, a forgotten proof of concept and the nightly exports. A single invoice arrived each month and nobody could say which part of it belonged to which project.<\/p>\n<p>The Ops Atlas, a reference entry on this topic, lists this pattern first among its anti-patterns and gives three reasons it hurts: quotas are shared, any mistake spreads across everything, and the bill cannot be attributed. The team had all three. When an overnight reporting job grabbed most of the available compute allowance, the booking site could not add servers on a busy Friday. Nothing was broken in the application. The ceiling had simply been consumed by a neighbour in the same account.<\/p>\n<h2>Chapter two: splitting the estate<\/h2>\n<p>After a tense retrospective, the team moved toward what the Atlas calls a landing zone: a standard baseline that each new account inherits when somebody requests it. The idea is that a new project receives its own account with baseline settings already applied, rather than building them by hand. The Atlas names AWS Account Factory for Terraform and Azure subscription vending as examples of the mechanism.<\/p>\n<p>Splitting had a price, and the story is honest about it. More accounts mean smaller blast radius and separate per-account quotas, which was precisely what the team needed. They also mean more baselines to maintain, because every account must carry the same logging, alerting and access settings. The team discovered that the first account was easy and the fortieth was a chore until vending made it routine.<\/p>\n<p>One detail caught them out. The group had assumed their organisation-wide restrictions covered everything. As the Atlas explains, the organisation-level restrictions known as service control policies skip the management account entirely, and the provider itself advises keeping workloads out of it. Two small tools had been quietly living in exactly that account, outside every rule they thought was universal. Moving them cost an afternoon. Finding them cost a week of confusion.<\/p>\n<p>If you are weighing up how cloud work fits into a broader digital strategy, this piece on <a href=\"https:\/\/bhaso.com\/blog\/bridging-web-solutions-cloud-and-artificial-intelligence-for-digital-growth\/\">linking web solutions, cloud and intelligent tooling to business growth<\/a> covers the commercial side that the engineers in this story rarely get to discuss.<\/p>\n<h2>Chapter three: the tag audit that told two stories<\/h2>\n<p>With separate accounts in place, the finance contact asked a reasonable question: which product pays for what? The team enforced a small set of required labels on every resource and then measured compliance. Counting resources, 60 percent carried the labels. That sounded poor, but not catastrophic.<\/p>\n<p>Then somebody weighted the result by spend. The Atlas gives a worked version of this exact comparison: with 7,600 out of 12,000 in monthly spend sitting on labelled resources, compliance is 63.3 percent by cost against 60 percent by count. The gap is small in that example, but the point is the method. A count treats a tiny test volume and a large database as equals. A weighted figure shows where the unlabelled money actually is. The team adopted both numbers and began chasing the biggest unlabelled items first.<\/p>\n<p>They also learned a quirk worth passing on. The Atlas observes that AWS tag policies leave unlabelled resources out of the non-compliant list altogether, so a clean report can hide the very items you were hunting for. On Azure, labels set on a resource group do not automatically flow down to the resources inside it. The team stopped trusting a green dashboard until they had checked what it was able to see.<\/p>\n<h2>Chapter four: guardrails that people could live with<\/h2>\n<p>Next came the temptation to lock everything down. The team started with restrictions on which geographic Regions could be used, which is a common, sensible guardrail. They quickly met its limits, as the Atlas predicts. Resources that already exist in a blocked Region can become unreachable once the rule lands. Certain global services behave differently from regional ones. And the Google-side constraint, the Atlas adds, governs where resources are created, not where data is processed.<\/p>\n<p>So they changed their approach. Rules first ran in audit mode, simply reporting what would have been blocked. Only after two weeks of clean reports did they switch to enforcement, and only with a documented exception route for teams that had a genuine reason to deviate. The Atlas lists guardrails without exception paths as an anti-pattern in their own right; people who cannot get a quick yes will find a quiet workaround.<\/p>\n<h2>Chapter five: the Friday that went differently<\/h2>\n<p>Months later, a provider event degraded the launch path in the team&#8217;s main Region. Compute that was already running carried on. New servers would not start. Because the booking site&#8217;s recovery plan had been reviewed to contain no launch steps, with spare capacity provisioned in advance, the failover required only a traffic shift. The team watched the incident from a dashboard hosted outside the affected provider, so their view of the situation did not disappear with the outage.<\/p>\n<p>The Atlas describes this design target as zero control-plane steps in top-tier failover, and reserves its highest maturity level for organisations whose rehearsals, called game days, prove it holds. The team had not read about it first. They had arrived at it by getting burned twice.<\/p>\n<p>The retrospective afterwards was short. Nobody had been heroic. The team had spent a few quiet months making the unexciting parts of the estate predictable, and the payoff was an incident that felt almost dull. That is a fair description of what good operations work is supposed to deliver: fewer surprises, and the surprises that remain arriving with a plan already attached.<\/p>\n<p>One choice is worth recording. The group stayed away from a second cloud for resilience. The Atlas cautions that multi-cloud adds another control plane, another identity system and another set of guardrails, and it quotes Google&#8217;s own warning that those complications might outweigh the first benefits. The team kept one provider and simply used a second Region for the pieces that mattered.<\/p>\n<h2>What the team would tell another team<\/h2>\n<ol>\n<li>Separate accounts early, because splitting later is slower than starting that way.<\/li>\n<li>Measure label coverage by spend as well as by count.<\/li>\n<li>Run every new rule in audit mode before enforcing it, and publish an exception path.<\/li>\n<li>Check where your own workloads and tools actually live before trusting an organisation-wide control.<\/li>\n<li>Review the recovery plan for hidden dependence on the provider&#8217;s control plane.<\/li>\n<\/ol>\n<p>Teams looking for a wider view of how modern delivery, platforms and cloud fit together may find <a href=\"https:\/\/bhaso.com\/blog\/transforming-software-delivery-through-cloud-devops-and-platform-engineering\/\">this overview of software delivery across cloud and platform engineering<\/a> a useful companion. And for the underlying definitions, scope boundaries and a staged 30, 60 and 90 day plan, the original <a href=\"https:\/\/www.xopsschool.com\/cloudops\/\">CloudOps entry in the Ops Atlas<\/a> is a short and well-organised read. Whatever your team size, begin by listing where accounts, quotas and labels stand today, and let the largest gap choose your first project.<\/p>\n","protected":false},"excerpt":{"rendered":"<p>An invented six-person team outgrows a single cloud account. Follow their account split, tag audit and guardrails to see what everyday CloudOps work looks like.<\/p>\n","protected":false},"author":6,"featured_media":1332,"comment_status":"","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[166],"tags":[],"class_list":["post-1331","post","type-post","status-publish","format-standard","has-post-thumbnail","hentry","category-business-technology"],"_links":{"self":[{"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/posts\/1331","targetHints":{"allow":["GET"]}}],"collection":[{"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/posts"}],"about":[{"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/users\/6"}],"replies":[{"embeddable":true,"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/comments?post=1331"}],"version-history":[{"count":1,"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/posts\/1331\/revisions"}],"predecessor-version":[{"id":1440,"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/posts\/1331\/revisions\/1440"}],"wp:featuredmedia":[{"embeddable":true,"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/media\/1332"}],"wp:attachment":[{"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/media?parent=1331"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/categories?post=1331"},{"taxonomy":"post_tag","embeddable":true,"href":"https:\/\/bhaso.com\/blog\/wp-json\/wp\/v2\/tags?post=1331"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}