For those involved in SaaS development or management, I'm curious about your thoughts on acceptable downtime per month for a website. Is it under 5 minutes, under 30 minutes, under an hour, or do you think it doesn't matter much? Additionally, do you actively track downtime incidents, or do you only become aware of them when users start to complain?
9 Answers
When it comes to downtime, I believe the focus should be on the impact rather than just minutes lost. A short outage during peak hours can be way more damaging than a longer one during off-peak times like 3 a.m. The real question is how much business you stand to lose during that downtime and how it affects user trust.
Totally agree, but remember that in legal contexts, those minutes can come up as important in discussions.
It really does depend on your specific use case and the expectations set with your customers. Some businesses can manage with a few hours of downtime, while others need zero—they're all about those five nines!
Our goal is a 99.99% uptime SLA, though we usually exceed that—sometimes hitting 100% since the start of the year. That's about 2 minutes of downtime per month, which is manageable with good processes and monitoring in place. Constant checks every minute help us stay on top of performance and downtime.
Impressive numbers, particularly with all your proactive checks! I love that you measure responsiveness too—not just uptime.
What monitoring tools do you use? I'm a bit old school with Nagios, but I’m always interested in alternatives.
It's simple: any downtime is a problem, especially during crucial times when traffic peaks. Tracking those minutes and their impacts is vital.
Honestly, I just take the server down when it’s time for bed!
I knew a company that had a simple message, "We’re closed for the day" on their site overnight!
The maximum acceptable downtime is whatever you agree upon in contracts based on client needs. What works for one may not work for another.
Really, it just comes down to business needs and customer expectations. If clients demand high uptime, then you better deliver; if it's a blog, it can be down for a while without much concern.
This is an impossible question to answer without context! It greatly depends on service types, use cases, and even the time of day. For instance, AWS going down for an hour might cost millions, while a downtime in payroll software on non-pay days isn't a big deal.
Couldn't agree more! Every situation is unique, and user impact can vary significantly based on the service.
Right?? Timing and context are critical in determining how serious downtime is.
The real concern should be about what each minute or hour of downtime costs the business. Without that context, the time metric alone is meaningless.
Exactly! Tracking uptime is dull until you tie it to potential revenue loss or trust issues.
Exactly! The amount of downtime doesn't tell the whole story, it's really about the losses and user perception. During a busy time, even a few minutes can have major consequences.