Cisco Support in the Philippines After End of Life

The Lifecycle Plan Usually Stops at the Server

Most organisations track end-of-support dates carefully for compute and storage. Switching is reviewed on a separate cycle by a different team, and the two plans are rarely laid side by side. The result is that the component every server in the rack depends on for connectivity is frequently the one furthest past its support date and the one nobody has assigned to anybody.

A Switch Failure Is Larger Than a Server Failure

This is the asymmetry that makes the oversight expensive. A failed server affects the workloads on that server. A failed supervisor module, line card or switch power supply removes connectivity for everything behind it, which on a consolidated rack is every system at once. The blast radius of the untracked component is the largest in the room.

Redundancy Covers the Failure and Not the Interval

A redundant switch pair does exactly what it should: the survivor carries the load and service continues. From that moment the rack is running with no network redundancy at all, and the length of that state is set by how quickly a matched component can be obtained. On a platform Cisco no longer sells, that is a sourcing exercise rather than a support call, and it is measured in weeks unless a spare is already held.

The Cheapest Items Cause the Most Outages

Optics are the clearest example: inexpensive, easily held, and the most common single reason a link does not come back up after maintenance. A spare power supply and fan tray for each switch family, plus the transceivers used on active uplinks, covers most of what actually happens. ICD supplies Nexus, Catalyst, UCS and optical components for platforms past end of support, tested and warranted for one to three years.