Check, check. Alright, check. One, two Beyond the code that gets pushed through Is users building value and earned reliance Building dreams engineered in art and science Gotta earn the love, our central goal is expressed But how exactly is that abstract aim assessed? We need visibility, build our own objective sight Of customer experience, day by day, and night to night Lest we offend, we display the trends Through the eyes of users who depend (on us) Directors, engineers, and customer support Did we understand the assignment or fall short? One view where we all get to see each service dependency Tell me what was the reality, for each precious VIP Then leadership can act before a pandemic Of accumulating toil and errors systemic Threaten to sink us beneath cold, frigid seas Of discouraged customers and failed guarantees Alright, one view. We see what you're sayin' Just grab all of our metrics and get to displayin' Want a cache hit ratio? Want the limit we're obeyin'? We got so much telemetry that when we start surveyin' I dunno what we wanna show, it feels a bit dismayin' I think we better step back and think what we're portrayin' Before we go to measure -- the how can wait First define the what, and clearly articulate Gotta capture the reality, the end user's experience What exactly are they seeing, time to take it serious True service quality on their viewpoint depends Here it is point blank, shed the internal lens Find the ideal SLOs, the concept's not the same We quantify success within the frame of the game If our current telemetry just doesn't align Note the deficiencies for a future design Now map your dependencies, trace 'em upstream Carefully negotiate promises from each team Hold up. Precision counts Define the terms before confusion mounts SLI is an indicator; a service level definition When your code be instrumented query to see its condition Measure latency and error ratio, but frame it clean The event, the method, help me build the scene What is success? What's counted in the mix? How is it measured? Nail the specifics Next, SLO: the objective you attach A target value for the metric you must catch Can ninety-nine percent uptime be attained? Set internal standards that must be sustained Now, SLA: that's an agreement A formal contract that we signed with intent With fees that we pay, or credits we remit If service levels fail what SLOs commit The crucial difference now, SLAs guard the biz Contractual terms define what liability is So we might aim lower to help manage the dangers But SLOs are internal, for friends not strangers Aim higher for the heart of the users we need Help our business help our customers succeed Right, terms are set. Now who's job is it? What services can we include or omit? Any system we manage that's customer-consumed Whose poor performance could leave users doomed By delays or errors. Any backing systems deep If others rely on promises you keep Internal platforms, and tools for the crew If their work depends on your system too If your system's outage would cause pain elsewhere Then you gotta define your SLOs. Show you care Hey thanks for tuning in, stopping by to lend an ear We still got a lot to share with every software engineer But these verses gettin' lengthy, got me cookin' with the crew We gon catch our breath a minute, stick around to hear part two
