Switch Firmware & OS Upgrade Services

Stacks, chassis and standalone switches behave differently

A standalone access switch upgrade is one of the more straightforward changes in networking. A stack is not. Stack members must run compatible software to form a stack at all, which means the upgrade has to bring every member forward together — and a member that fails to load the new image can leave the stack in a split or degraded state that takes the whole closet with it. Stack cable integrity, member priority and the state of the stack master are all things worth verifying before rather than after.

Chassis platforms add supervisor redundancy to the picture. Dual-supervisor systems may support upgrading the standby supervisor and failing over, reducing the outage substantially — but only if the software versions involved support that path and the redundancy mode is configured correctly. Establishing which is the case is part of the assessment rather than something to discover in the window.

Core and distribution switches carry an additional consideration: everything downstream of them is offline while they reboot. Sequencing an estate upgrade so that redundant pairs are never down simultaneously, and so that access-layer work does not coincide with the distribution upgrade above it, is the difference between a planned maintenance window and an unplanned outage.

Switch platforms we support

How we sequence a switching estate upgrade

1. InventoryFull estate discovery — model, software version, stack membership, chassis and supervisor configuration, uplink topology and support status.
2. GroupDevices grouped into upgrade waves by layer and redundancy, so that no redundant pair is ever taken down together and access work never coincides with the distribution above it.
3. PilotOne representative closet upgraded first to confirm image behaviour, timings and validation criteria before committing the estate.
4. Roll outWaves executed in agreed windows with a go/no-go gate between them, so a problem found early does not propagate across the estate.
5. ValidatePer-wave verification of uplinks, VLAN propagation, spanning-tree state, PoE delivery, access-port connectivity and management reachability.

Frequently asked questions

How do you upgrade a switch stack safely?

The core principle is that all members must come forward together, so the image is distributed to every member and verified present before any reload is issued. We confirm stack cable integrity and member health beforehand, since a marginal stack cable that has been quietly working becomes a real problem when members are renegotiating stack membership after a reload. Rollback for a stack means having the previous image retained on each member where storage permits.

Can core switches be upgraded without an outage?

If the core is a redundant pair and the network is genuinely designed for one to fail, then yes — the members are upgraded one at a time with validation in between. In practice we frequently find that redundancy which was designed in has quietly degraded, with a link that was never brought up or a protocol relationship that only exists on one side. Verifying that redundancy actually works is part of the assessment, before it is relied on.

What about PoE devices during an access switch upgrade?

Everything powered by the switch loses power for the reboot duration — access points, IP phones, cameras and door controllers included. This matters more than it first appears, because security and safety systems on PoE can have their own implications for a maintenance window. We identify PoE dependencies during the assessment so that the business can make an informed decision about timing.

Do you upgrade switches from multiple vendors in one programme?

Yes. Mixed switching estates are common, particularly where sites were acquired or built at different times. We run them under one programme with a consistent process and consistent reporting, while respecting that each platform has its own upgrade path rules and validation specifics.

How long does an access switch upgrade take?

A single standalone access switch is typically 20 to 40 minutes including reboot and validation. A stack takes longer because all members reload and rejoin. In practice the constraint on estate upgrades is rarely per-device time — it is how many closets you can safely take down in one window and how many windows the business will grant.

Related services

• Network Device Firmware Upgrade
• Router Upgrade Services
• Core & Distribution Switching
• Network Device Health Check
• Request an Upgrade Assessment