Router and switch replacement and migration
A replacement switch has to behave exactly like the old one, VLAN for VLAN and port for port. When the vendor changes, that means translating a configuration, not copying it.
We do that work remotely at a fixed price, under your brand or as your named partner. The client’s network comes out the other side documented, with a port map that matches what’s plugged in.
- Staff
- Phones
- Access points
- Printers and cameras
- Uplink
Ports that filled up one at a time on the old switch can be grouped by purpose on the new one. The port map records every move.
What the config doesn’t show
A switch configuration records what someone changed. Everything left at the default is invisible, and the defaults are where vendors differ most. Here is an ordinary uplink, a desk port and a gateway from a Cisco switch, with what we read into each line, including two lines that aren’t there.
interface TenGigabitEthernet1/1/1description Uplink to coreswitchport mode trunkswitchport trunk allowed vlan …1There’s no allowed-VLAN list, so this trunk carries every VLAN on the switch. Several platforms carry only the VLANs you name. Copy it literally and VLANs drop off the uplink; allow everything and VLANs turn up where they were never meant to be.
switchport trunk native vlan 992Untagged frames on this link belong to VLAN 99, and the far end has to agree. When it doesn’t, traffic from one VLAN lands in another, and it usually looks like a DHCP fault rather than a switching one.
!interface GigabitEthernet1/0/1description Reception PC3Descriptions get written once and rarely updated. We check them against the MAC address and neighbour tables, which show what’s plugged in today.
switchport mode accessswitchport access vlan 10switchport voice vlan 204The phone on this port learns its voice VLAN through CDP, Cisco’s own discovery protocol. Most other switches hand it out with LLDP-MED instead. Without it, phones come up on the staff VLAN.
spanning-tree portfast!spanning-tree mode …5With no mode shown, the switch runs Cisco’s per-VLAN spanning tree. Most other vendors run RSTP or MSTP. While old and new switches share the network, the root bridge has to be placed on purpose, or it can move by itself and traffic stops while the network settles.
!interface Vlan10ip address 10.10.10.1 255.255.255.06The gateway for VLAN 10 lives on this switch. Whether routing between VLANs stays here or moves to the firewall is worth deciding deliberately, rather than inheriting.
ip helper-address 10.10.0.57DHCP for this VLAN is relayed to a server elsewhere. Leave it out and nothing breaks on the night. Devices already holding an address carry on; new devices and anything that reboots or whose lease expires can’t get one, so the fault turns up over the following days.
The same reading applies whichever way the move goes. Juniper, Huawei, Meraki and FortiSwitch configurations have their own silent defaults, and they don’t match Cisco’s or each other’s.
Four layers, each with its own way to come unstuck
Most failed swaps trace back to one detail on one layer that nobody looked at. We work through all four on every job, from the copper up.
Physical
The box fits the rack. The rest needs checking.
- Optics
- Transceivers coded for one vendor often won’t come up in another vendor’s switch. An uplink that doesn’t light stalls the whole window.
- Power over Ethernet
- Count the draw, not the ports. Access points, phones and cameras can need more than the new switch’s power budget, or a higher PoE class than its ports supply.
- Uplinks and stacking
- Port speeds, uplink modules and stack cabling differ between models. The link to the core has to match at both ends.
Switching
Where most of the silent defaults live.
- VLANs and trunks
- Every VLAN ID and name, and exactly which VLANs each trunk carries, stated explicitly on the new platform.
- Spanning tree
- One mode and a root bridge chosen on purpose, set before old and new switches are joined.
- Link aggregation
- LACP on one end and a static bundle on the other won’t form cleanly. Both ends have to agree, including links to servers, storage and the firewall.
- Jumbo frames
- Storage and backup traffic that relies on a larger MTU needs it end to end, or it slows down without ever failing outright.
Routing
How traffic leaves each VLAN, and the building.
- Gateways
- Which device routes between VLANs, and whether that stays where it is.
- Routes
- Every static and dynamic route carried across, including the ones added years ago for a supplier link that nobody remembers.
- DHCP relay
- Helper addresses rebuilt for each VLAN that gets its addresses from a central server.
- The internet handoff
- PPPoE credentials, a VLAN tag the provider expects on the WAN, and upstream equipment that can hold on to the old router’s MAC address for a while after the swap.
Management
So the new device can be looked after.
- Access
- Admin accounts, RADIUS or TACACS+ where they’re used, and which networks are allowed to reach the management interface.
- Visibility
- SNMP, syslog and NTP pointed where they were, so your monitoring sees the new device instead of reporting the old one as down.
- Firmware
- A current, supported release chosen and loaded before the window, not upgraded halfway through it.
Same vendor, or a different one
Like for like
A newer model from the same vendor speaks mostly the same language, but not quite. Interface names change with new hardware and stacking, commands get renamed or retired between software generations, and defaults shift. We rebuild the configuration line by line rather than pasting the old one in.
Between vendors
Moving between vendors means restating each VLAN, trunk, route and port setting in the new platform’s terms, and handling the gaps on purpose: a feature with no direct equivalent, a default that runs the other way, port numbering that starts at zero instead of one.
When the management model changes too
Meraki switches and routers are configured in a cloud dashboard rather than at a command line. FortiSwitches are usually managed from the FortiGate in front of them, so the switching configuration lives in the firewall. Moving on to or off platforms like these changes where the configuration lives and who can see it, and the cutover plan accounts for that.
Vendors we work with: Fortinet, Palo Alto Networks, Cisco, Meraki, SonicWall, WatchGuard, Sophos, Check Point, Juniper and Huawei.
The cutover window
The window is for moving leads and testing. By the time it opens, the new equipment already holds its full configuration, so nobody is typing commands while the client waits.
-
Ready before it opens
New equipment configured and checked line by line against the port map. The window itself is agreed with you and the client well ahead.
-
Racked and patched
Someone on site installs the equipment and moves each lead to the port the map gives it. We’re connected remotely throughout and work through it with them.
-
Settled
Uplinks, spanning tree and link aggregation confirmed stable before testing anything that depends on them.
-
Tested
Every VLAN gets an address and reaches its gateway, the others and the internet. Phones land on the voice VLAN, access points and cameras power up, printers and servers answer, and monitoring sees the new device.
-
Signed off, or rolled back
The cutover plan sets out the way back as plainly as the way forward, so a failed test that can’t be fixed inside the window ends with the old setup back in place.
What the client ends up with
A network that works the way it did, on equipment documented as it is.
Configured equipment
Running a configuration rebuilt for the new platform, not carried over line for line.
As-built documentation, including a port map
Every port, its VLAN, its mode and what’s connected to it, recorded after cutover so it matches reality.
A cutover plan
The sequence, the tests and the way back, written before the window and kept with the documentation.
| New port | Was | VLAN | Mode | Connected to |
|---|---|---|---|---|
ge-0/0/0 | Gi1/0/3 | 10 Staff | Access | Desk 14 |
ge-0/0/12 | Gi1/0/1 | 10 Staff, 20 Voice | Access with voice | Reception phone and PC |
ge-0/0/16 | Gi1/0/5 | 30 Wi-Fi, 31 Guest | Trunk | Access point, level 1 |
ge-0/0/23 | Gi1/0/8 | 40 Devices | Access | Printer, copy room |
xe-0/1/0 | Te1/1/1 | 10, 20, 30, 31, 40, native 99 | Trunk | Core switch |
Fixed price, under your brand
Delivered remotely as a white-label project, under your brand or as your named partner. We quote and invoice you at a fixed price for the agreed scope, and you keep the client and the ongoing service.