Availability

Every request receives a (non-error) response

“Availability means that every request receives a response, without guarantee that it contains the most recent write.” – Eric Brewer

When to use

When uptime is more important than perfect data accuracy (Social Media feeds, Shopping carts).

Why it matters

  • User Experience: Users hate “System Down” or timeouts. It’s better to show a slightly old list of posts than an error page.
  • Revenue: Amazon found that 100ms latency cost 1% in sales. Availability (always responding) is king for retail.

Signs of Violation

  • “Service Unavailable” errors when a database node is down.
  • System locking up waiting for a sync that never happens.

Explanation

Problem

If you insist on perfect Consistency, and a network cable is cut, you have to stop accepting writes (System Down).

Solution

Accept the write on the available node, even if it can’t talk to the others yet. Resolve conflicts later.

Real world analogy

Calling a call center. If the “Main” agent is busy, you get routed to “Backup” agent in another city. They might not know you just called 5 mins ago (State consistency), but at least they answer the phone (Availability).

Pros and Cons

Pros Cons
  • Uptime
  • Low Latency
    		•
  • Stale Data
  • Write Conflicts (Split Brain)
    		•

    Comparison

    • High Availability (HA): A general engineering term, but in CAP it specifically means “choosing uptime over consistency during a partition”.

    Code example

    Typescript

    Bad (Violation - CP)

    Good (Adherence - AP)

    PHP

    Good (Adherence)