Interesting opinion piece here:
What do you think?
I was talking to my students about this very thing today. It wasn't an original thought from me though. I think the first place I read it was in Devin's blog. I am usually initially opposed to automation in early generation hardware and software. It now seems that there is more and more information pointing in the same direction, which is that site surveys in most situations cost more than the extra equipment to blanket the entire area.
Unfortunately I don't have any experience comparing the two. I still use site surveys but that is because it is very difficult to install AP's in my situations (cement hotels) so I want to install as few AP's as possible to reduce my cost and the cost to my client. In a building with pretty drop ceilings? Boy, that would be nice. :)
I think at times, it (predictive analysis) can be seen as a bunch of speculation and hocus pocus or more like a blind date. WYSIWYG.
You always need to study your target. In the military we sent in the Long Range Surveillance (LRS) teams to get real-time intelligence and analysis of the objective....also we had the Calvary Scouts to help give us the information we needed to come in with the right defense and offence and the big GUNS. We should simply call it as the article says PREINSTALLATION PLANNING.
Some customers want a firm fixed price for installation. You can't under price this just to close the deal. It is not good business. This is when the "devil" in the details is desired. Educating them (clients) on the dynamics of RF and the ways to optimize the RF in their facilities is the key driver before going to do a Site Survey.
With thin APs you can throw them out there and let them grow. Much like the Chia pet. Or something like.. Wonder Twin Power Activate.. .Shape of Radio Waves?
As GT Hill hinted at each structure and site is different, especially concrete structures. Understanding the dynamics of RF in a Multi- Story building and how to optimize the RF propagation of Wifi and Cellular signals best to provide not only coverage, but capacity is EXPENSIVE. With applications like Voice over IP and Multimedia ...you get the point?
Typical losses is these structures range from -20 to -50 dB loss depending on the architect and building code requirements, since 9-11 it has gotten tougher on radio waves. There are rooms full of transfromers and generating equipment that cycles on an off above and below you .... if your spectrum analyzer skills is not at par, you could place an AP right next to the room with it causing major havoc on your users in those areas.
Slapping up APs and running CAT5e/CAT6/coax to these locations through concrete slab... is well , next to impossible. Dropped Ceilings are a cake walk. But wait, is it plenum space and you decide to use a non-plenum rated solution to SAVE MONEY?
There goes the firm fixed price and the customer looks for someone that is willing to hang some illegal FCC unapproved access points with a high-power AMPLIFIER to provide only coverage... NOT CAPACITY.
LET's JUST BLAST THROUGH the FCC won't care?
I am looking for an indoor mesh that provides high speed data connectivity on the ingress and egress (triple radios), so I can limit the amont of UTP on the floor to several key locations in a hybrid MESH /POE thin AP scenario. Also a POWER Over Fiber (POF) driven AP?
I could go on and on .. PREINSTALLATION PLANNING is critical.
Physically seeing where things need to go is the key. RF spot checking in key locations is what I would sell. RF Analysis around the elevators , Below Ground, in the Mechanical Rooms, Electrical Rooms anywhere there is heavy machinery is KEY.
It costs to have full coverage and being honest and up front with the customer is important, especially if they are looking at lots of usage on the WLAN.
Cisco is now saying 14 users or less to an AP? That's a lot of AP$ my friend. I am for PREINSTALLATION PLANNING and am looking for a tools that can show me predictively how all of the above can be overcome.
A HYBRID APPROACH is BEST....then again I can buy a couple of Chia Pets and watch the rest of SuperFriends? :)
I have had very good results performing predictive site surveys. I think that they provide a fantastic balance between the labor of an on-site survey and the unpredictability of the "put them in and then fill in the holes" approach. Let's not forget that the "fill in the holes" approach requires an experienced designer/installer to detect holes and determine the best place to put the APs to fill in the holes. Many customers don't have this expertise and don't want to hassle with having somebody come back time and again to tweak the design. If you're an experienced 802.11 engineer putting in a network for a company that doesn't mind the tweaking process, then "plan, install, tweak," might be best.
But many of the customers that I deal with want a single design that they can be reasonably sure will work right out of the box. Hire installers, have them come in and put the equipment up, go home, done. With a predictive site survey, I can provide an installation plan that ensures good coverage throughout the site, and all the customer has to do is hand it to their installers, who put the APs up on the wall. This is especially important when the customer is a large telecommunications company that is, itself, providing the designs to ITS customers (in other words, when the design has been sub-contracted out to you). Imagine a hypothetical company that has won a contract for putting wireless networks into a system of 250 hospitals or grocery stores throughout the country. There is no way that this installation could be reasonably completed with the "plan, install, tweak" method. "Tweak" would take too much labor and time, even if only a small percentage of the sites needed tweaking. I find that, in cases like this, predictive surveys provide the extra accuracy that is necessary to be able to provide a design that won't need tweaking.
I forget where I read it, but I remember seeing that enterprise LANs are typically about 30% acquisition cost and 70% management cost. I certainly understand the logic behind ditching the site survey to some degree, but where is the real benefit? If acquisition costs are such a relatively small percentage of your overall WLAN cost anyway, doesn't it make more sense to spend the cash up front so that management costs (troublshooting drops, low speeds, inconsistent roaming, etc.) can be reduced? Call me a cynic, but I am still distrustful of relying solely on a software utility to manage Wi-Fi channel selection and power output settings in a grid installation.
I liken today's automatic wireless systems to MS Windows plug and play. The concept was great initially, but the what you actually got was often less than advertised. Today it is rock solid. Perhaps in the future automatic wireless power output and channel settings will be as well, but in my experience it is more like the days of Windows 95b.
For clients that use their mission-critical applications over a wireless network, when they don't perform as expected you must deal with facts to prove or disprove it is the environment or the application. And if it is the environment, is it the wireless implementation, or something else. Site surveys give you the facts.
Whether folks use predictive site survey software, do the old fashioned site survey, or a combination of both - at least one of them needs to be done!
Earlier today, I was in a public library and found four 802.11g AP's via RF scan. They were on channels 1, 4, 8 and 11. That was my first clue something was wrong. Then I walked around and quickly found the AP's. All four mounted very close to each other in the middle of the library. Throughput was horrible on any AP I associated with. Didn't have a protocol analyzer NIC handy, but could only imagine the retransmissions in the air.
I'm sure the IT folks who "engineered" and installed this WLAN system just needed to get it working, and don't care to learn anything more to make it better.
Wonder if sites like this exist in many more places [around the globe!] than we CWNP folks care to imagine!?