Design It Once, Design It A Half Dozen Times…

For the experienced WiFi engineers reading this (if you give it a second or two), what I am about to share is going to maybe make some cringe. Or shout an expletive or two. Look, I get it…this one was a learning experience, and we all have them a time or two in our careers. Honestly, sometimes we need the painful projects to really drill things home…and for me, this was a doozy. So I figured I will share it for a few reasons. One, it reminds me what I’ve learned, and to ensure I don’t repeat any of my stumbles. Two, perhaps me sharing this will help others who are starting out. I promise it’s mostly going to have you going “well, duh!”, but it’s still worth sharing. Forgive me for the humorous depictions of the project floor plan–they aren’t real, but the project actually did happen.

I honestly rather enjoy designing a WiFi deployment. Of all the aspects of the CWNE, design was the favorite for me. However, as much as I enjoy designing, I can’t say that designing a site six or seven times is my idea of a fun time. To clarify, this one wasn’t like, intentional iterations of a design as things evolved, per se. I’ve had those. One site I worked on last year had several evolutions to it. Some came to be when hardware got changed up, while other changes were necessitated because floor plans changed in the middle of my design efforts. Those aren’t bad…they certainly wouldn’t make me pull my hair out (which, as a fun aside, is probably a good reason I have short hair!).

Before I dive into the painful experience, it’s worth putting this out there-this is repeated in the CWDP book as well (or, it was in the version I studied for, but I have to assume this would remain). For any project you are designing, make sure you have customer requirements. As in, what do they need WiFi for? Where do they need it? How much coverage? How many devices, and how capable are those devices? In addition to the requirements, it’s absolutely critical to have the best possible floor plans you can get. I am not revealing a secret with that…and I also know that I, and many other WiFi engineers, have done surveys with bad floor plans-whether they were outdated, poorly drawn, too small or anything else…a bad floor plan is, in theory, better than none at all.

So, imagine working on a building where you have a floor plan…and you know stuff is going to go into the floor plan…but that’s all you know. Customer is insistent on getting a deisgn done ASAP, in order to procure gear or budget for the eventual procurement, and of course to run cables before the site is finished (this was an all-new site). It’s really hard to put out a good design where you are not working off of the end layout, isn’t it? Yes, yes it is.

first iteration
Here’s the first (blank) floor plan

Iteration #1

Given the empty floor plans, customer said they didn’t have end layouts, so I should approximate. Seemed this would be a budgetary ballpark, so I could go a bit aggressive and, once we had some clarity, we could go back and update the design to be more accurate. That seemed fair, and while I knew I was making some assumptions, customer was understanding that basing the first draft off of assumptions was a risk, and could/would mean the ultimate design would be different.

It’s worth pointing out here that I had also requested customer requirements/use cases/equipment info, but as with a populated floor plan, there wasn’t much information to be had. Without their requirements, I was designing with our overall corporate standards in mind-coverage everywhere, 5ghz, etc. Makes sense, right?

Made my assumptions, created design #1, got quotes…nothing crazy here, outside of not getting info that I should have managed.

iteration 4
First stab at it, they let me guess on where the furniture would go (and on how crazy it would look)

Iteration #2

Is anyone here surprised that the assumptions I made about furniture and shelving layouts ended up being wrong? I can’t say I was. I did have to laugh though. I was tackling two new sites for this build, and I had to start with assumptions for each. I assumed for site 1 that the furniture would go one way…and for site 2, it would actually go the opposite way.

How is that funny? Because after design 1 was done, we got some feedback or info from the customer. Site 1, the furniture was going the opposite direction I had expected/guessed. Ditto for site 2. Just can’t make this stuff up.

So, back into Ekahau I went, and I moved around walls and shelves and the like, to be in line with the actual intended build. I say intended, because…well…stay tuned.

iteration 7
heard I guessed the layout wrong. At least this far in, we finally had some answers…right?

Iteration #3

This was straight forward. I had to move the furniture and shelves in each of the sites to be in line with what I now knew to be the final layout. If only it would stay “final”…

iteration 5
Ignore the left side for this excercise…that remark actually happens in the next step.

Iteration #4

This was beginning to feel like my “My Cousin Vinny” design. After the 3rd change, I figured what else could possibly be wrong. Turns out, the customer I’d been working with? Never fully engaged the actual customer. So, we had a meeting to review my design…guess what? More changes.

Half of one building, they decided they didn’t need any WiFi coverage. And reviewing some of the other spaces, guess what? Yep, the furniture and shelving layout we thought was final in the prior go-round? It wasn’t. They had moved things and changed things and added things…basically making my 3rd design irrelevant.

iteration 6
Another couple of changes…including that one half of the floor wasn’t going to need WiFi

Iteration #5

After removing the WiFi where customer (the real one this time) said they would not need WiFi, I thought (oh, so foolishly) that we were close.

Yeah, no…still chugging away. This one wasn’t as bad as some others, but going through minor adjustments still adds up. This one, customer was requesting “value engineering”, which may be the least favorite pair of words I hear kicked around. But I get it, they wanted to keep the costs under control, and I get that (even if I disagree).

It’s just information that would have been useful….four iterations ago.

iteration 9

Iteration #6

It’s worth mentioning that, by like, iteration #4, the hardware for the project had long since been purchased.

That being said, this revision saw more value engineering. Apparently the customer had some “spare” 3802 or similar models and wanted to use those (even though a decent chunk of the designs called for external antennas). Also, at some point since the opening iteration, the furniture layout changed…again. In this iteration, they were all about going in and just some minor tweaks…well, sort of. More shelving ended up being added (worth saying this shelving was floor to ceiling, so it did matter more than your average shelf). But they wanted to know if they could just move around existing/planned APs, to fill in the newly created gaps.

It was right then and there that I knew this design process was going to hurt me.

iteration 8

Iteration #7

As I write this up, lucky #7 appears to be the winner…I think. Really, it’s like 7a and 7b. 7a is the “value engineered” version of a “design” that they had someone there cook up, just by sheer luck I guess. 7b was me taking that new floor plan and treating it as though, you know, someone with a CWDP and an ECSE would design it.

Anyone going to be shocked to learn that the customer version (7a) had coverage gaps a plenty? Where 7b, the design done by the designer, did not.

Guess which one is getting deployed?

Yep…not 7b….


See, there’s plenty to learn here, especially if you are doing design for the first time (and no, this was not my first rodeo).

  • Ensure you have the final layout before you start going at a design. Even if they tweak it a bit (like, add a shelf or two, but 95% of what you had was right from the start), that’s important. If they cannot get you accurate floor plans, do not start a design, because more likely than not, ANYTHING you cook up will be wrong (and thus, a waste of time and money).
  • Ensure you have the actual customer requirements…not just a project manager saying they had no requirements. As in…should the building be entirely covered? Are certain areas more critical or less critical? What are the most important devices being supported? What are the most and least capable devices?
  • Are there any other constraints you have to deal with. Like, has the overall budget been burned up, so you don’t have a lot of funds to work with? Are there certain deadlines that need to be adhered to?

In the end, iteration 7a is what the customer really wanted and felt they needed. That is fine, because that is our end goal, is it not? To deliver the customer the design which they are requesting us to put forth…nothing more, nothing less (OK, if you can give them more great…but nothing less.). The trouble here was, we thought we got the requirements, but we didn’t. We thought we got the layout (several times) but we didn’t. 6+ designs were created that did not need to be, all because I didn’t stand firm and decline to design without a final, confirmed and blessed layout, matched up with a confirmed list of requirements.  Some of this was on the customer (we asked, they didn’t deliver the right stuff, etc). but at the end of the day, it was my boo-boo, because I didn’t push back hard enough and thus, I wasted a bunch of time on quite a few designs that ended up never needing to be drawn up.

With some luck, this will be the one and only time I go through a process like that, with that many revisions which could have been avoided.



Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s