In the distant year of 2004, the head of our technical department was fortunate to be invited to the launch of the first Wi-Fi network in Russia. It was launched at Nizhny Novgorod University by Cisco and Intel, which had opened a research and development center there in 2000, employing over a thousand engineers and, notably, acquiring a decent building for this purpose. At that time, according to these two 'pioneers of production', this was almost the only functioning corporate wireless network. Today, such claims of 'uniqueness' would likely provoke disputes, but back then, it was truly groundbreaking.
This was, in fact, Wi-Fi of the IEEE 802.11g standard. Of course, the presentation emphasized the very principle of being able to connect to a network without wires, and here the creators of the technology were not lying, but when it came to speed and range, there was a lot left unsaid and gaps in information. Well, Wi-Fi G is just 'G', as it was named, and that’s what you got. It would be misleading to say it was seriously used in critical areas within organizations.
A real step forward was the introduction of the 802.11n standard, which became the foundational point for most networks in operation today. History has shown that N300 type equipment is still used by many, and for numerous users, it is sufficient. At least it had been, until the 2.4GHz band turned into a graveyard of signals. With the arrival of 5GHz and the 11AC standard, things improved a bit, but apparently not for long. One of the key issues is the stability and speed of the link is still a concern.
Considering the combined issues and advantages, until recently, we advised all our clients to connect via cable whenever possible. This was justified, as 802.11n (which recently received the name 'Wi-Fi 4') did not provide the speed and stability that gigabit Ethernet offered. Naturally, with proper installation and choice of cable, on which one should never economize: only good copper and only category 5e or 6. Currently, we strive to use only category 6 and above, and soon it will be clear why.
Let’s clarify one more thing. Yesterday, we could insist that the client stick to a wired connection, but today that’s no longer the case. The paradigm of our surrounding world has changed. A quarter, if not half, of devices are gadgets, another quarter are ultrabooks that lack Ethernet ports (typically high-end models and mid-range ones that migrate from office to office), leaving only about 30-40% as stationary workstations. Consequently, the question, "Why is our office Wi-Fi so slow?" is becoming more frequent. And we are looking for solutions. We are trying different options.
This was just a prelude, and the core of the story is that one of my clients, after replacing the core network equipment and connecting to the "right" provider via proper fiber optics, wanted to improve their wireless network, built on Wi-Fi 4 standard equipment (we’ll refer to it differently going forward). Over many years, their access points have partially broken down, leading to numerous dead zones, and those that remain no longer meet the capabilities of most client devices in use. By "wanted," in such cases, one must understand the presence of financial resources and administrative will — without these, it’s just a topic for discussion over a cup of tea. For obvious reasons, I won’t disclose the client’s "name," I will just say that it’s a private gymnasium occupying a four-story building.
An educational institution is a rather complex structure where one component is tied to another, and the issue of access to local and internet networks plays an increasingly crucial role. Consequently, their IT needs have been constantly growing and continue to grow. For instance, the administration wants to conduct live broadcasts of all the events happening at the gymnasium, stream lessons for sick students and those temporarily receiving remote education, and hold group online seminars and staff meetings with the participation of remote teachers from other branches of the gymnasium. Additionally, the servers house a centralized archive of methodological materials for lessons that require the fastest possible access through a web interface of the intranet and simply via network drives. As the cherry on top, public access must be provided for visitors, as parents want to share photos and videos of their children on social media directly from the assembly hall during performances of their adorable kids.
What we had at the outset:
~ 15-20% of the gone points H~E on the N300 standard and perforated coverage as a result.
~ 10% of points with 'gastritis' — they are somewhat alive, but they need to be rebooted periodically.
~ a rather relative 'central management'; for the last 2-3 years, the points have lived and managed themselves. Some license was not renewed when the IT administration changed, and that's what happened.
So seven years ago, when the building was commissioned, it was a cutting-edge network with the latest technology, but what happened was inevitable: component aging, overheating due to dust, electricity surges, impact of a 'ball hitting a point' and so on.
The number of clients and computerization of the school has only grown each year. Laptop cabinets were installed in classrooms, and students began using their phones for educational purposes as well.
What is pleasing:
Seven years ago, we, along with our colleagues, also carried out the installation of the wired network, and since the client gave us a blank check, the cables and connectors were of the same Cat6 quality and proper thickness — no cutting corners. As a result, after seven years, the majority of the cabling infrastructure remained in more than acceptable condition.
All that seems necessary is to choose this wireless part of the network. This is where many controversial points arise: from the approach to selecting the standard to the brand and budgeting.
Depending on the current moment, the decision regarding the choice of standard can be obvious or not obvious. Obvious when the old standard has become common, and the new one is just on the horizon. Not obvious when the new one is already being implemented but still occupies a small share.
In this case, the new standard is IEEE 802.11ax, and the old one is IEEE 802.11ac, which are referred to as Wi-Fi 6 and Wi-Fi 5, respectively. Of course, networking equipment for the latest standard is always more expensive, but the temptation to save was outweighed by one argument: when we installed Wi-Fi 4, it was not cheap either, but we enjoyed many years of operation with almost no modernization costs, and at the maximum speeds at the time of implementation.
I won’t explain here why the 6th standard of wireless communication is better than the 5th; many specialized articles have been written on this topic. Perhaps the only thing that needs to be understood is that the airwaves are shared among all users, and we cannot lay additional airwaves, and each new generation of the wireless standard allows for more efficient use of the airwaves, meaning it enables more users to operate at higher speeds.
The next important point is the choice of vendor. The first thought was H~E – it worked well, so let's choose something from H~E/A~a.
We make a request for A~a with AC and AX. This will give us A~a N~s AP-5~5.
We receive: Ar~ AP-5~5 — with AX – 63 thousand rubles (November 2019) and A~a N~s AP-3~~ with AC – 52 thousand rubles (November 2019). We need such points for the site (4 floors with 10 x 15 pieces = a minimum of 40-50). In total: 2.6 million rubles if we take 11AC at RRC prices. Regarding 11ax, we can only say that AX is not cheap and postpone it for later. And we haven't even reached the cost of the controller and licenses yet!
What has happened over the past 7 years? The exchange rate has increased! Back in 2013, branded points also cost around 600-800 dollars, but the exchange rate was different. The gymnasium, although private, earns income in rubles. This is where the cognitive dissonance and rethinking occurred during discussions with the client.
Everyone knows the concept of overpaying for a brand.In this case, this option is clearly evident. For the client, choosing a brand means one thing: if you're not knowledgeable, buy from the most popular ones, and you certainly won’t go wrong, if you can afford it. For us, selling an expensive product is also great — we will earn more. There remains the risk that the client will 'skip' to someone who dares to offer something cheaper, given that both we and the client are in 2020, not 2013: a crisis is behind us, a new one is approaching, and we need to think clearly.
So what do we do? Convince the client to forget about AX? But what if they really want AX?
Then we look for alternatives!
Fortunately, the IT market is dynamic: something is constantly dying and something new is emerging. Sometimes, newcomers, in an attempt to charm the public, offer the same or similar specifications as top-tier brands, but at lower prices. Undoubtedly, there’s a risk of a lottery, roulette, and even 'Russian roulette' with a shot at losses. But this risk can be relatively minimized if we approach the filtering with thorough testing before purchase.
What are the chances of finding a gold ring in a pile of last year's leaves? The answer is 50/50% — either you’ll find it, or you won’t — probably not. But sometimes you do find it.
We as integrators are invited to all conferences. I mean all kinds: from telephony and intercom systems to access control and Wi-Fi. Sometimes we attend. Besides marketing, in 1 out of 100 cases, there’s some valuable insight.
Just last summer at a mixed vendor conference, there was a certain Taiwanese company called EnGenius. It's unclear who they are. All that remains in my memory a year later is that the brand sounds like a mouse manufacturer and they claimed to have ready-to-go Wi-Fi 6, also known as AX. I miraculously recalled this when looking at a Genius mouse.
I visited their website. I dug up a presentation from that conference mailing. When studying the slides, it struck me that EnGenius allegedly claims to be a contract manufacturer of networking devices (specifically, access points and controllers) for brands like Cisco, Dell, Extreme, Fortinet, Zyxel, and others. If we are to believe the Taiwanese, they produce wireless equipment under their own brand with the same technologies in those same factories.
In general, it turned out that Wi-Fi 6 is something that EnGenius has been working on for a long time for their 'senior' products. Moreover, they were among the first in the world to manufacture networking devices based on the Wi-Fi 6 standard (IEEE 802.11ax).
A year ago, this was just an interesting piece of information that quickly faded away, but now, with the issue of upgrading Wi-Fi in the gymnasium becoming urgent, it has gained attention.
Question No. 2. How much and where to get samples.
The first thing to compare is the economic efficiency. A quick evaluation of retail prices showed an astonishing effect. A point from EnGenius costs on average half of what a brand-class ‘A’ costs. So there’s a problem inside! Or is it again a factor of overpaying for the brand?
Samples are needed. Without in-depth testing, considering a product with such characteristics and at that price is, how should I put it, 'risky'. We call various companies in Moscow and St. Petersburg – no stock available, and those who do have it by chance are not giving it out for testing. Especially not the AX points.
But we are persistent! We write to Taiwan. For some reason, they respond from the Netherlands. It turns out, there are EnGenius representatives right in Russia. After some correspondence in broken English regarding vague clarifications, we get contacts for people in Russia. Apparently, there is a testing fund. The product exists, and we can get our hands on it.
After describing the task with an emphasis on choosing only AX points and signing a letter of guarantee, after one and a half weeks (from Samara!) we received a set of 4 different points, including an AX point and a PoE switch, which turned out to also be a network controller.
Taking into account all factors (price limits, required density, and the gymnasium's requests), we selected the EnGenius EWS377AP access points for testing and potential future installation.
What drew us to them: the claimed speed is up to 2400 Mbps on the 5 GHz band + 1148 Mbps on 2.4 GHz. This is incredible if we believe the figures.
The package included an 8-port gigabit switch-controller with PoE+.
For testing, it was certainly suitable, but it was obvious that traffic potentially generated by the AX point cannot be transmitted over a gigabit Ethernet port. In fact, the access point itself is equipped with a multi-gigabit interface of 2.5 Gbps. If anyone remembers, this standard was adopted back in 2016. and has now started to gain traction.
This feature of the access points fits perfectly into the customer's requirements, as after the network core upgrade, most of the ports for the gymnasium are indeed multi-gigabit over copper + part 10G SFP+.
Everything is good, but it raises the question of switch selection. In the case of EnGenius, if we're building a homogeneous network, there are only 8-port switches with 2.5G with PoE+ currently available. Initially, we planned to use a 48-port high-density switch or at least two 24-port switches with SFP uplinks to accommodate the increasing number of PoE+ ports. However, all of EnGenius's offerings are currently gigabit, like the eight-port model received.
The good news is that we can boast about our favorite cabling topic. The presence of Category 6 cables in the project, laid 'for the future' and capable of carrying 2.5 Gbps, significantly speeds up, reduces costs, and simplifies the task.
As we can see, at the time of laying the cabling system, there was no active equipment available for such speeds, which further confirms that one should not skimp on cables.
In the end, the picture is as follows: we are testing the system on their 8-port switch-controller, but in the future, we will likely opt for as floor switches. The details of the required number of ports will be shown by radio planning.
Step 1.
We are assembling a stand from the access points and the Switch Controller.
Entering the web interface.
The main page of the switch, which is also the controller.
Distributing the access points into groups.
Great! The entire network, including third-party devices, is laid out clearly! Convenient and accessible.
Step 2.
We are looking for the radio planning tool in the controller but cannot find it.
It turns out that EnGenius has radio planning, but it has been moved to the cloud and is called ezWiFiPlanner. We call customer support at EnGenius. They register us in their system and provide access.
So what do we see here.
The cloud system for Wi-Fi coverage planning turns out to be remarkably powerful. I would say it rivals similar products from Ekahau, but with one pleasant exception — this ezWiFiPlanner is completely free. That is, entirely. The downside is, of course, that it knows nothing beyond its own EnGenius access points.
A simple layout of a radio plan can be created in just a few minutes, as demonstrated in the video. Next, we outline the walls and windows, identify which walls are load-bearing, and where the gypsum board ceilings are located. We clarify with the client that we attach the points to the ceilings and as close as possible to the previous locations, moving them back and forth until they occupy their final positions.
Overall, I can confirm from my own experience that working with the EnGenius planner is quite easy and convenient; the libraries contain everything necessary. Changing network parameters is simple, and we can immediately see the results. I would like to note that in the cloud, one can save their projects and then export them to use as templates for other facilities. This is a plus, as I have experienced that many built-in software in controllers do not allow saving radio plans, even paid systems that do not permit exporting projects in simple PDF. What were we paying for then?
And so we get this coverage diagram of our facility.
This is the ground floor layout for the 5GHz frequency; the other floors have practically identical layouts.
In fact, that’s the whole solution.
As for choosing the installation locations for the access points, in our case, it would have been simpler to install new access points in the same places where the previous Wi-Fi 4 standard ones were located, and perhaps cover the assembly hall more densely. Essentially, we did just that, trying to minimize the work of re-routing cable paths to the points. However, given the real picture of the new radio plan and the list of requests/adjustments from the client derived from seven years of experience with the previous network, some cable ends still had to be rerouted to different locations, and some sections had to be re-laid through the trays. But overall, this can be considered a minimal upgrade.
In planning, I preferred, as they say in preference, to re-lay – the number of client devices and traffic volume will only grow, and I would like this network to last without the need for upgrades longer.
Step 3. We test and compare.
It's time to understand what the AX technology gives us. Especially since, besides the AX access points, we have the Wave2 and Wave1 with various antenna configurations. So we can afford to compare the results. For testing, we are using a Samsung S10 with stated support for AX (802.11 a/b/g/n/ac/ax 2.4G+5 GHz, HE80, MIMO, 1024-QAM).
Measurement on EWS360AP and EWS377AP.
Tests were conducted at a distance of 2-3 meters from the access point, which is a typical distance from the access point to a student in a classroom. With the latest Galaxy model belonging to one of our technicians, we managed to achieve nearly 640Mb/s wirelessly. This is indeed impressive.
The results are very interesting: ~320Mb/s on the EWS360AP(AC) access point compared to ~480MB/s on the EWS377AP(AX) over the local network. The gain is nearly 50%. Naturally, in real conditions the speeds will be lower, but the difference is absolutely clear.
Surprise where we least expected it!
Let's consider our tests to be quite positive. The remaining issue is managing the entire network within the framework of the operational project. The planned access points, EnGenius EWS377AP, definitely have a built-in web interface for configuration, but it makes sense to use it only for standalone setups, outside of a group. We have a different task – orchestrating an entire matrix of access points.
In the scope of a gymnasium, it is necessary to achieve seamless roaming according to IEEE 802.11k/r/v standards, and a separate guest network, possibly more than one. In principle, allows for a total of 16 SSIDs, each with its own group policies (for administration, accounting, teachers, students) – but this is only possible with centralized management.
During my time working with the EnGenius switch-controller, I have come to the realization that a PoE switch and a controller are essentially the same thing, and that no additional payments would be necessary. However, when we transitioned to compiling a specific specification, we discovered that the new 2.5GbE PoE+ switches from EnGenius do not have a built-in controller, as they are hybrid – local-cloud. It is expected that in the future we may switch from local to cloud controllers. This might be a global trend, but for now such an option would only cause panic for the client, hence a question was raised in the technical support about what other options are available.
In response, two options were proposed: installation of a free product. for computer or purchase of a hardware mini-controller with features and a web interface identical to ezMaster.
Let's summarize the platform selection issue in a table
SkyKey – mini-controller
ezMaster – software on the server
Maximum number of points in the array
100
1000+
Management
Through EnGenius Cloud or locally via web interface
Hardware requirements
The box-controller itself
Needs a host: PC or server and a virtual environment
System startup speed
Almost instantly – just plug it in and get to work
We need to figure out how to install, configure, and then, as always…
Certain fluctuations occurred, but here we decided to take the path of simplicity in the project. Plugged it in and it took off – Plug-and-fly!
In practical life, a specialized network device with a web interface that can be accessed from any client (including a smartphone) is always appealing due to potentially greater stability than installed software, especially for a very specific deployment via VMware. I have nothing against VMware; the virtual machine has its advantages, but it needs to be set up on the server for other virtualized tasks. This takes time. And we're already saving the client a decent amount of money.
The mini-controller’s limitation of 100 access points is not critical for the gymnasium – in our wildest fantasies, we won’t come close to the limit, and radio planning gives us less than half the load.
The discussion was settled by the magnetic mounting of this gadget. Snap! – it stuck. Everyone laughed and took it.
The schematic diagram and core of the network.
The overall connection diagram turned out this way. The total number of points decreased from 40 to 32.
Since the 'main' switch has 4 ports, and we needed 5, it was decided to connect the third floor through the second (half of the second floor is occupied by the assembly and sports halls, and there are significantly fewer clients there).
And the core of the system was chosen as Juniper EX2300-24T. The choice was between it, SG500X-24P, and AT-GS924MPX-50. But with very similar specifications, the device from Juniper is significantly more affordable and fits within the budget.
Summary of the experience gained.
It's too early to draw conclusions. Conclusions can only be made when the network is finally put into operation and has been running for at least six months.
Currently, impressions can be divided into 3 components.
Positive:
- The price for AX is more than reasonable. Essentially, choosing this vendor allowed us to stick with the idea of using Wi-Fi 6 in general. Looking at others who already have AX, it’s expensive and there are many complicated licenses. For example, I have great respect for the company A~d T~sis, but charging money for seamless roaming is absurd and outrageous in our time.
- I was very impressed with the cloud Wi-Fi planner. It’s done at a high level and for free.
- The interface of the controller is quite well designed, where the entire network is transparent and can be managed from a single screen.
- The appearance of the access points is neutral; they blend into the interior, and the brand name is barely visible.
- Despite the fact that we took a risk choosing something unknown, the network on the Engenius works. In fact, it works excellently. The signal is stable, it doesn’t fluctuate, and the access points don’t disconnect. Time will tell how they perform at a large event in the auditorium, but the entire office segment, which launched first, is operating very steadily.
- Roaming. It exists. I won’t argue that it actually works, as we did with another product once at the dawn of this phenomenon — it’s just that many have it nowadays, and any decent manufacturer should have it.
- Native support for Mesh networks and its setup presented no issues.
+band steering. Yes, it works. It transitions normally between frequencies.
Negative:
In my opinion, the ceiling mount is silly and not very reliable. Many cheaper products even use metal for the mounting base, which is firmly secured and easier to install. I’m not saying it doesn’t hold, but we have an ‘active crowd’ with mass sprints down the corridors and throwing objects, so we anticipate potential problems.
The cable input window on the 377th point is, to say the least, not very well thought out. The cable needs to be fed in from the ceiling at an angle, and if feeding power with a separate 12V pair and connector, it barely fits into this opening. The situation is worsened by the not-so-blunt edge of the metal back that can crush the cable.
Neutral-strange:
It looks a bit odd that in the older version of gigabit switches, there was a built-in local controller, but in the new versions, it’s absent.
In conclusion.
The choice of whether to rush and buy AX points today remains with the client. It's clear that this is a trend. If it's a trend, then there's no need to swim against the current; we should adopt it if requested.
You can evaluate the technical advantages and draw conclusions based on test results. A major question is what will connect to this Wi-Fi 6 and how to measure the speed. On old equipment, there’s no reason to do so. But on new devices, the gain is obvious if the incoming network to the point is also adequate.
The remaining question is, what about EnGenius? Overall, it seems more like a "Yes" than a "No." What impressed us was that the network was set up all at once without issues, and everything worked smoothly. However, we will be able to judge overall in about a year. For now, we’ll leave it at that, but there's nothing really negative to say.
The situation for now.
From the beginning to mid-March, we managed to deploy a pilot segment. Now, for obvious reasons, we can't continue the rollout of the segment, but the test results we obtained are more than encouraging.
Source: habr.com
