How would you plan a network migration?

You can usually measure the success of any project by the amount of planning, research and testing that’s been invested into the project. I’ve done dozens of network migrations, from complete forklifts and gradual side by side migrations and all of them required a significant amount of planning, research and testing prior to the actual execution. It was the planning, research and testing that I directly credit for the success for all those projects. Here are the general steps that I go through when migrating a network (replacing or upgrading the physical hardware or equipment);

1. Cleanup
2. Documentation
3. Research
4. Testing pre-migration
5. Execution
6. Testing post-migration
7. Turnover

Let’s go through each of those steps and I’ll explain what I’m talking about.

Step 1. Cleanup

This step is usually overlooked but can be the step that provides the biggest bang for the buck. We have a core switch with 240 ports, 110 of which are completely idle. Why not cleanup and remove those 110 ports and save yourself from having to worry about trying to migrate them. The same goes for the actual configuration. We have 8 port-channels of which only 5 are active. The other 3 port-channels have been decommissioned but no one ever cleaned up the configuration or cabling. Let’s clean up the configuration prior to any migration so we only need to worry about what’s actually in use.

Step 2. Documentation

I generally like to document all the switch ports, not just the uplinks and downlinks but also dumping the MAC/FDB and ARP tables and document what’s connected to every port. You’d be surprised how often this has proved very helpful either during the migration or post migration troubleshooting.

Step 3. Research

It’s really important to-do the research to understand what caveats you could run into. In most cases you won’t be the first person building a wheel, there will have been a bunch of other folks that have done this already and have discussed their issues, problems and experiences online somewhere. It’s equally important to understand how you should be configuring the new gear and how you’re going to reach the final goal. Let’s not forget the logistics of any implementation. Is there enough space, power, cooling… is the power 120V or 220V, do I have the proper PDU and UPS sized properly, do I need 5-15P or C14 power cords?

Step 4. Testing Pre-Migration

No one wants to jump off a cliff without knowing with a high degree of certainty that the parachute is going to open and work. This is the phase where you prove that all the planning and research is going to show real fruit. If you have a test plan, please make sure you execute it pre-migration. You’d be surprised how many times I run into people telling me that X or Y isn’t working after a network change – only to find out that X or Y had never worked for quite sometime.

Step 5. Execution

Here’s where the rubber meets the road.. whether it’s an overnight forklift or a side by side migration this is what you’ve been planning for. It’s time to get the job done.

Step 6. Testing Post-Migration

Let’s make sure that everything is still working properly… before the users start calling on Monday morning.

Step 7. Turnover

The final hurdle, documentation and the implementation of some type of monitoring and management solution.

Let me know what’s been your largest or most challenging upgrade or migration in the past few years.

Cheers!

Image Credit: sanja gjenero