Upgrading Switches in 24/7 environment

[Marsrock2021]

Just wondering what process people use for updating switches that are in use 24/7.

Do you disable patching ?

Skip patches?

Deploy previous patch?

Test a patch in not critical devices and leave for a few weeks to test?

Or not use meraki for 24/7 environment 

[cmr]

@Marsrock2021 we have used Meraki across 9 24/7 sites for the last few years. 

One of the great benefits of Meraki is that the upgrades don't take long (~2 minutes per MS/MX and less per MR).  The MRs have a great mode where they can upgrade in a way to minimise disruption by not upgrading neighbouring APs at the same time.  For MSs we just pick a quiet time and for MXs we have HA pairs so clients don't notice.

If you compare this to our Cisco IOS switch stacks that take over 10 minutes to restart when upgrading and when we had WLCs where all APs went down at once, I'd take Meraki any day.

[cmr]

@Marsrock2021 adding the the above, we absolutely do apply updates at less critical sites first and also start with more simple config a, i.e. upgrade L2 switches before L3 switches and individual switches before stacks etc.

[PhilipDAth]

If it was a true 24x7 environment, I wouldn't be using Meraki.

Like @cmr says, we have used Cisco Meraki in 24x7 manufacturing envirnments.  We typically ask for a maintenance window every 6 months.  Nearly always you get a shift being cancelled, a public holiday of some kind, internal factory machine maintenance, or something like that, and you can schedule in the upgrades during that window.

Note you can not disable Cisco Meraki firmware updates.

[BlakeRichardson]

Its hard to promise 24/7 as no matter what vendor you use all hardware has firmware upgrades not just for reliability but also security. 

Yes big companies use co-location and have data spread across multiple data centres but if its just a single server 24/7 is almost impossible to provide. 

[sebas]

We've had Cisco all day long and an architecture which made it possible to update every single component with little to none downtime. And we were able to have specific devices to update to see how things went.

In the office all is replaced by Meraki. 

Seems ok but i really do miss some enterprise worthy things; 

- Changing configs on vlans without interrupting other vlan traffic (On MS for example)

- Changing ospf config without the possibility to determine the moment of clearing the proces by yourself.

- No LACP on MX's

- being able to update 1 or 2 devices without having to "clone" the network to specific subsets

- being in control of which version you exactly want. You only have stable, release candidate and beta and no choices in between.

- Advanced logging from the gui so you don't have to contact support to have some logs which you would normally see yourself (Like the exact reason why a failover was triggered on the MX)

- etc etc

So for beginner network admins or system admins things seem sufficient and graphically logic, but I am more flexible with Cisco.