Access Switch Selection for Campus Closets

Decide the Power Requirement First

Whether the closet has to power anything is the decision that cannot be reversed in the field, so it belongs at the top rather than as a line item checked at the end.

If access points, cameras, phones or door controllers are anywhere in the design now or in the funded roadmap, the switch has to be a PoE model and the budget has to be calculated before the port count is fixed.

Then Set the Port Count From the Real Endpoint Population

Port count is usually chosen from the number of wall ports rather than the number of live endpoints, which overprovisions in older buildings and underprovisions where wireless has displaced wired desk connections.

• Count endpoints that are actually live rather than sockets that exist
• Add powered devices separately since they consume both a port and a share of the budget
• Allow for growth over the switch replacement cycle rather than the project
• Consider whether two smaller switches suit the closet better than one large one

Uplink Capability Is Fixed Earlier Than People Expect

On fixed switches the uplink is either built in on an L variant or supplied by a network module ordered alongside the switch. Either way the decision is made at purchase, and the module has to match the switch family.

The question to answer is not what the uplink needs to carry today but what it will carry once the access layer behind it has been refreshed. An access switch outlives at least one generation of the endpoints it serves.

Stacking Decides How the Closet Is Managed

Stacking turns several physical switches into one managed unit, which reduces the number of devices to configure, monitor and upgrade. Where a closet will hold more than one switch, stacking capability and the stacking cables belong in the original order.

Stacking bandwidth differs between families and between the standard and L variants of the same family, and this is one of the specifications worth reading on the product page rather than assuming from the family name.

The Software Tier Is a Design Decision Not a Price Decision

The licence tier supplied with the switch sets which capabilities are available. Choosing it on price and discovering the gap during implementation turns into a licence purchase plus a delay.

The question worth asking is whether the design relies on segmentation, automation or assurance capability. If it does, the tier is already decided.

Hold the Choice Across the Estate

The single largest operational saving in campus switching is not the unit price. It is running one model, one software train and one spares pool instead of four.

That only works if the chosen model stays available for the length of the programme, which is a supply question rather than a design question. It is worth confirming availability horizon before fixing a standard, not after.

• One access model held for the life of the programme
• One software train so change windows are uniform
• One spares pool that covers every site
• One configuration template that new sites inherit