What Is Uptime? SLAs and What 99.95% Really Means
What is uptime and what does an SLA mean? The yearly and monthly downtime behind 99.9%, 99.95% and 99.99% uptime, outage causes, monitoring and tips.
Uptime is the percentage of time a server, website or service is running and reachable over a given period. 99.95% uptime means the service is reachable 99.95% of the time, so downtime doesn't exceed 0.05% of the total time. The agreement in which a provider promises this rate in writing is called an SLA (Service Level Agreement).
In this post we explain how much downtime uptime percentages really allow, what an SLA covers, the most common causes of outages and how to monitor and improve the uptime of your own services.
How much downtime do uptime percentages allow?
Small differences between percentages make big differences in downtime:
| Uptime | Maximum yearly downtime | Maximum monthly downtime |
|---|---|---|
| 99% | 3 days 15 h 40 min | 7 h 18 min |
| 99.9% | 8 h 46 min | 43 min 50 s |
| 99.95% | 4 h 23 min | 21 min 55 s |
| 99.99% | 52 min 36 s | 4 min 23 s |
| 99.999% | 5 min 16 s | 26 s |
In the industry these rates are referred to by "the number of nines": 99.9% means "three nines" and 99.99% means "four nines". Each extra nine cuts the allowed downtime tenfold and noticeably raises the cost of delivering it.
What is an SLA and what does it cover?
An SLA is a provider's written commitment to a level of service quality. When evaluating an SLA, look beyond the percentage at these details:
- Scope: What does the commitment cover? Usually it's the hardware, network and virtualization infrastructure the provider is responsible for. The software inside the server and the configuration you do are out of scope.
- Measurement period: Is uptime calculated monthly or yearly? Monthly calculation keeps a long outage in a single month from being spread across the year and disappearing.
- Scheduled maintenance: Does maintenance announced in advance count as downtime?
- Compensation: What happens when the commitment isn't met? The common practice is to refund part of the service fee as credit.
The most common causes of outages
Some outages happen on the provider's side, but most happen inside the server:
On the infrastructure side:
- Hardware failures (disk, memory, power supply)
- Network problems and routing errors
- DDoS attacks, which can make an unprotected server unreachable within minutes. See our What is DDoS? post for details.
Inside the server:
- Applications that crash or slow down because of memory leaks
- Full disks; the database can't write and logging stops
- Faulty updates and configuration changes made without testing
- Expired SSL certificates and domains
- Services suspended because a payment wasn't renewed
Even if your provider delivers 99.95% uptime, the real uptime your visitors see will be much lower if your application crashes regularly.
How to monitor uptime
You need to notice an outage before your users do. Use a monitoring service for this:
- External monitoring services: Check your site or server regularly from different locations and send a notification by email, SMS or channels like Discord and Telegram when they get no response.
- Your own monitoring tool: You can set up an open-source tool like Uptime Kuma with Docker in a few minutes; we use it as an example in our Docker installation guide. Run the monitoring tool on a different server from the one you're monitoring; otherwise, when the server goes down, the notification won't arrive either.
- Status page: A public status page for your users reduces support requests during an outage and builds trust.
For a website, don't just check that the homepage loads; if possible, also check that a critical page, like login or checkout, returns the right content.
Ways to improve uptime
- Run applications as services. Tools like systemd or pm2 automatically restart an application that crashes.
- Monitor disk and memory. A disk that's about to fill up or memory usage that keeps climbing is a sign of an upcoming outage.
- Plan your updates. Update during the hours with the lowest traffic and take a backup before major changes; we explain how in our Server backups guide.
- Automate renewals. Turn on automatic renewal for SSL certificates and domains.
- Choose DDoS-protected infrastructure. Attacks are the most common external cause of outages, especially for game servers.
- Build redundancy for critical services. For services that truly must stay up, running the application on more than one server with a load balancer in front removes the dependency on a single server.
Uptime at Nodesty
Nodesty offers a 99.95% uptime SLA on all its services. AMD Ryzen 9 9950X VPS and EPYC VPS plans run with L3–L7 nShield DDoS protection and free daily backups kept for 7 days, and web hosting plans run with LiteSpeed, CloudLinux isolation and JetBackup backups. While we monitor the infrastructure, you can secure the uptime of your own application with the steps in this guide.
Frequently Asked Questions
Is 100 percent uptime possible?
In practice, no. Hardware fails, software needs updating and networks run into unexpected problems. Be skeptical of any service that promises 100 percent uptime; serious providers offer a measurable target and an SLA that explains what happens when that target isn't met.
What happens if the SLA isn't met?
That depends on the provider's service agreement. The common practice is to refund part of the service fee as credit for the period that fell below the promised uptime. Read the agreement to learn the scope, the deadline for claims and how scheduled maintenance is counted.
My server is running but my site won't load. Does that count against uptime?
A provider's SLA usually covers the infrastructure it's responsible for, such as the server being powered on and connected to the network. Problems like software crashing inside the server, an expired domain or a wrong configuration are on your side and don't affect the provider's uptime.
How is uptime measured?
A monitoring service checks your site or server at set intervals, for example every minute, and records the periods when it gets no response. The share of time with a response out of the total time gives the uptime percentage. Services that measure from different locations can tell network problems in a single region apart from real outages.
What does the uptime command on Linux show?
The uptime command on Linux shows how long the server has been running since its last restart and the load average. This is different from the uptime percentage that measures a service's availability; even when the server is on, your application may not be running.
Is scheduled maintenance deducted from uptime?
That depends on the provider's SLA terms. Some providers don't count scheduled maintenance that was announced in advance as downtime, and some do. When evaluating an SLA, look at how maintenance is handled as well as the percentage.
About the author
Nodesty Team · Infrastructure & Support
The Nodesty team runs our servers, network and DDoS protection day to day and answers customer support requests around the clock. These guides come from that work.
Related articles
What Is a Subdomain and How Do You Create One?
What is a subdomain and how do you create one? Subdomains with DNS records, creating one in cPanel and DirectAdmin, pointing it to a game server, and SSL.
What Is Bandwidth and Traffic? Calculate Your Needs
What is bandwidth, and how does it differ from traffic? Mbps vs. MB/s, port speed, unlimited traffic, sizing for websites and game servers, and monitoring.
How to Change the PHP Version and php.ini Settings
How to change PHP version: choosing a version in cPanel and DirectAdmin, upgrading safely, and php.ini settings like memory_limit and upload_max_filesize.