HPE Support in Austria After End of Life

A Two-Node Array Survives One Failure and Nothing After It

3PAR arrays are deployed in node pairs and the pair genuinely protects against a controller failure. What it does not do is restore protection. From the instant one node is lost the array runs on a single controller with no remaining redundancy, and the duration of that state is determined by the availability of a matching node rather than by anything in the design.

A Node Is Not a Generic Chassis

Unlike a host, a replacement node has to join an existing cluster at a compatible level and assume the array personality. It cannot be substituted with any available server, and on a generation HPE no longer supports it cannot be ordered from a catalogue either. That combination is what turns a one-day repair into a multi-week exposure.

Single-Node Operation Is Slower as Well as Unprotected

Cache mirroring between nodes stops, and the surviving controller may adopt a more conservative write behaviour to protect data it can no longer mirror. The result is an array that is simultaneously unprotected and noticeably slower, so users experience a degradation during a period when, from their perspective, nothing has broken.

The Cheap Version of Preparedness

Holding a spare node is expensive and for many estates not justified. Knowing whether one is obtainable, from where, and in what time, costs nothing and is the difference between a planned response and a decision taken under pressure. ICD supplies 3PAR node hardware, cages and drives for generations past end of support, tested and covered for one to three years.