What does high availability mean?

High availability: definition and key figures

Highly available systems are built so that the failure of individual parts goes unnoticed by users or is noticed only briefly. To achieve this, critical components are provided twice or more—servers, databases, network connections, power supply. Experts call this avoiding single points of failure.

Availability is measured as a percentage of the time a service is reachable. At 99 percent, around 3.6 days of downtime per year are possible; at 99.9 percent, about 8.8 hours; at 99.99 percent, only about 52 minutes. Each additional nine makes the system noticeably more complex and expensive.

How high availability is achieved

The basic principle is redundancy: several servers behind a load balancer share the work (active-active), or a standby system is ready to take over in case of failure (active-passive). Databases are continuously replicated to a second node so that no data is lost in an outage and the system switches over quickly.

For higher requirements, the system is spread across several fire compartments or data centers within a region. This also requires monitoring that detects failures and established procedures for updates without interrupting operations. Without regular tests, high availability remains an assumption: whether the switchover works only becomes clear in an emergency.

Real-world example: a wholesaler’s order portal

A wholesaler accepts orders around the clock through a dealer portal; many customers order early in the morning before opening hours. If the portal fails, they call, orders are delayed and sales staff take orders by hand. An outage of several hours is therefore far more expensive than a second server instance.

The wholesaler therefore moves the portal to two application servers with a load balancer and a replicated database. If one server fails, the other takes over within seconds; updates are applied one after the other. For the inventory management system in the background, a simpler safeguard with backups is enough, since short pauses are tolerable there.

Benefits, limits and common mistakes

High availability reduces the risk of outages but does not replace backups: accidental deletions or malware are copied to the standby system as well. In addition, cost and complexity rise with every additional nine. Common mistakes are an overlooked single point of failure—such as only one internet connection or only one database server—and a lack of testing.

Nano Concepts works with you to assess which systems truly need to be highly available, plans redundant architectures, sets up monitoring and tests the emergency case—sized economically rather than blanket overengineered, with clear availability targets for each system and transparent costs so that business and management can decide on solid ground.

• Prioritize critical systems by cost of downtime instead of securing everything equally
• Systematically look for single points of failure in technology, network and power
• Test failover regularly instead of merely documenting it
• Plan backup and disaster recovery in addition to redundancy
• Agree the availability target in writing with the business unit and provider