This year I have worked on a number of projects and most of them had no planning while others had very little. The planning phase of a project is the most critical part of a project. Planning can not only make or break a project, but your reputation as well. If a project doesn’t go well, next time you put in a change request you might receive a lot more push back from your peers or management. How detail should you get in your planning? I prefer to get very detailed and before any purchases are made.
As an example, this might or might not have actually happened, let us take a Data Center upgrade from 6500s to Nexus. When the project is started it maybe at the end of the fiscal year, there maybe some monies left to spend, and Cisco may also have some great deals on Nexus bundles because it is also the end of their fiscal year. So you buy a couple Nexus 7010 bundles and a Nexus 5k/2k bundle. No planning is done, you just know you need that gear. Now you go to actually install the gear and do the planning that should already be done. After planning you determine you still need the following items: 1) redundant supervisors because the 7k bundles only came with one each, 2) additional modules for the 5ks because you don’t have enough switchports, 3) copper and fiber patch leads because you need a lot more fiber and want to replace the existing copper, 4) copper and fiber infrastructure cabling because you are putting switches in new locations, 5) cage nuts because these are all server racks, and 6) screws because you will need a lot of them. Oh, and when you purchased the Nexus 7010s you were going to put them both in the same rack, so can the floor handle the current and future weight of the rack with two Nexus 7010s? What about power? Is there enough power to each of the racks, especially the rack with the Nexus 7010s? Did you budget for all of these additional parts? Probably not since you didn’t do any of the planning ahead of time. In this example, you have to go to management and ask for about $80k just to put the new network into production with significant compromises. All because it wasn’t planned for ahead of time. Now you have a black mark against you with management.
When do you do your planning?
In the beginning is a good start; the very beginning. Before you make any purchases you should know the exact equipment you need, where you are going to put it, the power you need for it, cabling, optics, SFPs, cage nuts, screws, how and where devices are connecting, your L1/L2/L3 designs, where network services connect,etc. How are you going to do your cut or migration to the new network? What are your back out plans? Estimated amount of time to complete the cut or migration
What planning must be done before making a purchase?
Everything physical that you will need should be identified and documented before you make the purchase. Likewise you should have most of your logical design worked out before the purchase. Make sure you have all of the licenses you need for the features you are going to enable and use. If there are monies remaining at the end of a fiscal year, it is very tempting to just buy some equipment you know you will need, but make sure management understands that is not all you will need. Perhaps do some rough planning and have a part 1 and part 2 cost estimates so budgets can be put in place to finish the project. Also, make sure you put some padding into your budget for things that you may have missed. Even the best planning can sometimes overlook something little like cabling and if you don’t have the budget for it, and that could stop the project dead in the water.
Every project should have all of the planning completed before purchases are made or you start on execution. Depending on the size of the project will dictate how long the planning takes and what goes into it. My experience has been that the projects that go through the detailed planning go much smoother than those that just wing it.