“99.5% availability” sounds like “almost always.” In reality, it means up to 43.8 hours of downtime per calendar year — and that’s normal for the cloud. Here’s what the numbers mean and what to do if your provider doesn’t meet them.
Percentages in hours
| Availability | Downtime per year | Where you see it |
|---|---|---|
| 99% | 87.6 h (3.65 days) | budget hosting |
| 99.5% | 43.8 h (1.83 days) | VPS, cloud services |
| 99.9% | 8.8 h | Tier III data centers, banks |
| 99.99% | 53 min | geo-redundancy, multiple data centers |
SLA is calculated per calendar year: 99.5% doesn’t mean “3.6 hours every month” but up to 43.8 hours in total over the year (0.5% × 8 760 h).
Each additional “nine” costs many times more: backup links, a second data center, duplication of everything. For most companies, 99.5% with fast recovery is a better deal than 99.99% at five times the price.
What the SLA covers
- Physical infrastructure: power, cooling, the data center network, the hypervisor.
- Availability of the virtual server itself — it’s powered on and responding over the network.
What it doesn’t cover
- Your software: the website, 1C, the database. If the server is reachable but the application crashed after an update, that’s not SLA downtime (unless you have an administration contract).
- Scheduled preventive maintenance the client was notified of in advance.
- DDoS attacks on your server, problems with your ISP, force majeure.
What happens if it’s breached
That depends on the contract. There are no universal “compensation percentages”: some providers credit your balance, others refund the cost of the service for the downtime. Under the MyCloud offer, for downtime caused by the provider, the cost of the service for that time is credited back to your balance on request (clause 24.12). Either way, you need to document the downtime: a support ticket with the time, a monitoring screenshot. That’s why your own monitoring (an external check every minute) is worth having — and not just for the SLA.
What to look for in the contract
- How availability is measured — by the provider’s data or by independent monitoring.
- Response time and recovery time — these are different things.
- How and how far in advance scheduled maintenance is announced.
- Whether there’s a status page with incident history.
At MyCloud
The infrastructure SLA is at least 99.5% per calendar year, i.e. up to 43.8 hours of downtime per year, excluding scheduled maintenance you were notified of in advance. Support response time depends on the nature of the request: we reply as quickly as possible — where possible, from 10 minutes. The scheduled maintenance procedure, the parties’ liability and the full SLA text are in the public offer agreement. The current status of services is on the status page.
Need help with the task from this article?
We’ll pick a configuration, calculate the cost and help with the move.