Wi-Fi Gone Wrong
Here's a look at an installation that had great Wi-Fi saturation, but in the wrong place.
In my last post, "Chaos in the Cabling," I wrote about common cabling problems that crop up when installations ignore building codes and best practices. Here's another example showing that Wi-Fi deployments have their issues, too.
An overreliance on heat mapping, used to pinpoint network activity and plan Wi-Fi coverage, is one issue. Another is failing to take basic construction principles into account. Radio frequency is able to penetrate many areas. Understanding that certain materials and construction methods may impact signal penetration is key, but then so is an understanding of what it is that you are trying to penetrate.
In the client installation I'm sharing today, a campus building is shaped like a flat U and the wireless access points (WAPs) are managed in the cloud.
The main two-story classroom building, of early 1960's construction, resembles a motel. It has no windows on the front, and just a door for access at each end of a room. Glass windows fill the backside walls. The WAPs sit outside the classrooms under an extended eave that creates a porch-like feeling with plenty of cover from the weather elements. These outdoor-rated WAPs are mounted on the concrete block structure at about a height of eight feet.
Complaints about the Wi-Fi network include slow speed, inability to get access as needed, and general unreliability. But before I could resolve any problems, I had to walk the campus and discover what connects, and where. This would help give me a sense of the issues at hand.
Because the WAPs are mounted to concrete block and outdoors, penetration into the classrooms is poor. However, because the backside walls have all those windows, the signal is excellent in the building's green space, or courtyard. Unfortunately, classes are held inside, not outside.
Since glass is easier to penetrate than concrete block, the campus needed more WAPs to create an improved coverage on the backsides of the main classroom building. The thick concrete walls between the classrooms makes indoor antennas a lesser choice.
Overall, the campus needed more WAPs to facilitate connectivity and provide a good blanket of coverage. We mapped the distances between buildings and took measurements of buildings on a basic floor plan, and determined where to add more WAPs.
In our legwork we also discovered other problems, and tackled those, too. These included:
- Power - Power over Ethernet (PoE) switches powered only three WAPs, and only one of the PoE switches connected to an uninterruptible power supply. Injectors connected to external-outdoor power receptacles powered the other antennas. Not only was this foolish but also a blatant code violation since the transformers negated use of the protective covers of the outdoor weather-resistant receptacles.
- Firewall - All of the WAPs used a cloud-based firewall to restrict access to certain websites. The onsite firewall had content filtering and other security services enabled, but administrators, teachers, and students were prevented from reaching legitimate websites.
- Bandwidth throttling - The Internet link underserved the entire campus, and with bandwidth throttling enabled on "Student" and "Guest" SSIDs, students weren't able to connect. The slow and sluggish experience, which delayed classroom instruction and inhibited interaction, was compounded by the two firewalls fighting with one another.
- Security - The domain server was misconfigured, the SSIDs were not in virtual LANs, and the group policy and permissions in Active Directory were loosely implemented. Overall, security was poor.
- Improper material - Indoor Cat5E plenum cabling was used, even though all cabling is in an outdoor environment subject to the weather-related and even physical disruptions since no conduit was used.
Again, these kind of installations put customer faith in the technology on the curb. Had the installing company paid attention to building codes, best practices, and common sense, these problems would never have popped up. Avoid these pitfalls, and don't rely on what vendors tell you -- test everything.