Jul 01, 2026
Our first PPPOE Set Up in Uganda: Luxenetworks
How We Set Up PPPoE for a Client: A Luxenetworks Walkthrough
At Luxenetworks, we get a lot of calls that start the same way: "My internet was working fine, then the ISP switched us to a new connection type, and now nothing works." More often than not, the culprit is PPPoE (Point-to-Point Protocol over Ethernet). Last week, we handled exactly this kind of job at a 50-unit apartment complex whose ISP had just migrated the property onto a PPPoE-based connection. Here's how we approached it, the equipment we used, and the steps we took, in case it helps you understand what a proper PPPoE setup actually involves at scale.
First, What Is PPPoE and Why Does It Matter?
PPPoE is a networking protocol that many ISPs, especially DSL and fibre providers, use to authenticate and manage customer connections. Instead of your router just grabbing an IP address automatically (like with DHCP), PPPoE requires your router to "dial in" using a username and password supplied by the ISP, much like old-school dial-up internet, just running over Ethernet instead of a phone line.
The upside for ISPs is better control over billing, session management, and security. The downside for customers is that if it's not configured correctly, the connection simply won't come up: no internet, no clear error message, just a blinking light and a frustrated household.
The Property and the Situation
This job was for a 50-unit apartment complex. The building had just been switched over to a new connection by their ISP, and the property manager reached out after residents across multiple units started reporting the same issue: Wi-Fi showing as connected, but no actual internet access. With that many units relying on one shared connection point, even a small misconfiguration at the network core cascades into a building-wide outage, so we prioritized the visit.
They called us with three symptoms:
The core router showed a physical link to the ISP's line but no internet access
The ISP-provided PPPoE username and password weren't being accepted
Wi-Fi devices connecting through the access points on different floors could see the local network but had no external connectivity
This is a textbook PPPoE misconfiguration, so we scheduled a visit.
The Equipment We Used
For a property of this size, we relied on a compact but capable equipment stack:
MikroTik RB951: our core router, chosen for its RouterOS flexibility, reliable PPPoE handling, and the ability to manage NAT and firewall rules for the whole building from a single point
Managed switch: sitting between the RB951 and the rest of the building, distributing wired connections out to each access point
Tenda F6 wireless routers (x2): repurposed as dedicated Wi-Fi access points to extend coverage across the property, rather than acting as independent routers
ISP-provided line: the incoming connection requiring PPPoE authentication
Cat5e/Cat6 patch cabling: connecting the ISP termination point, RB951, switch, and each Tenda F6 in the chain
This combination gave us a single, centrally managed PPPoE session at the RB951, with the switch and Tenda F6s doing what they do best: distributing that connection cleanly across a larger property without introducing conflicting routers or duplicate DHCP servers.
Step 1: Confirming the Physical Layer First
Before touching any settings, we always rule out physical and cabling issues. With this setup, the chain ran: ISP line in, then the RB951 WAN port, then the managed switch, then the Tenda F6 access points on different floors, then resident devices. We checked that:
The ISP's incoming line was active and delivering a stable signal
The cable running from the ISP termination point to the RB951's WAN port was properly seated and undamaged
The RB951's WAN port link light was active
The cable from the RB951's LAN port into the managed switch was solid, and the switch itself was passing traffic (link lights active on every relevant port)
The cabling running from the switch out to each Tenda F6 access point was intact
It's tempting to jump straight into software configuration, but a good chunk of "PPPoE won't connect" calls turn out to be a loose cable, a faulty switch port, or a bad patch lead, and in a multi-floor property, tracing that down first saves a lot of guesswork later. In this case, the physical layer was clean end to end, so we moved on.
Step 2: Gathering the Correct PPPoE Credentials
This is where most self-installs go wrong. PPPoE credentials are not the same as your Wi-Fi password, and they're often formatted in ways that trip people up: extra characters, case sensitivity, or a required domain suffix (like username@isp.net instead of just username).
We contacted the ISP's provisioning line to confirm the exact credentials issued to the account, and verified there was no realm/domain suffix required for this particular provider. Small detail, but it's a common point of failure.
Step 3: Configuring PPPoE on the RB951
An important decision in a multi-device, multi-floor setup like this is choosing exactly one device to handle the PPPoE dial-up. You never want two devices both trying to authenticate the same session, especially on a property serving 50 units. We chose the RB951 as the PPPoE client, since it's the device sitting closest to the ISP line and has the routing horsepower to handle NAT and firewall duties for the entire building.
With confirmed credentials in hand, we logged into the RB951 via WinBox and:
Created a new PPPoE client interface bound to the WAN-facing Ethernet port (ether1), rather than leaving it on a plain DHCP client
Entered the username and password exactly as provided by the ISP, double-checking for trailing spaces, a surprisingly common issue when credentials are copy-pasted from an email
Set the MTU to 1492, the standard value for PPPoE, since it accounts for the protocol's overhead compared to a normal 1500-byte Ethernet frame
Set "Add Default Route" and "Use Peer DNS" so the RB951 would automatically pick up routing and DNS information from the ISP once connected
Configured NAT masquerading on the PPPoE interface so devices on the LAN side could share the single public IP
Enabled the connection and confirmed the PPPoE interface came up with a "running" status and a valid public IP address
Step 4: Setting the Switch and Tenda F6s to Their Proper Roles
With the RB951 handling PPPoE and routing, everything downstream just needed to pass traffic correctly across the building:
The managed switch was configured to carry traffic cleanly from the RB951's LAN port out to every Tenda F6 access point and any wired connections on the property
Each Tenda F6 was set to Access Point mode rather than its default router mode, with DHCP disabled on both units. This is a critical step, because if a Tenda F6 is left in router mode, it will try to hand out its own IP addresses and NAT traffic, creating a double-NAT situation that causes exactly the kind of "connected but no internet" symptom residents were seeing
Both F6s were connected to the switch via their LAN ports (not WAN), configured with static management IPs on the same subnet as the RB951, and set to the same Wi-Fi SSID and password so residents could roam between coverage areas seamlessly as they moved around the property
Step 5: Verifying the Connection End to End
With the PPPoE session up on the RB951 and the F6s reconfigured as access points, we ran through our standard checks:
Confirmed the RB951's PPPoE interface held a stable public IP with no repeated drops
Pinged an external IP from the RB951 to confirm outbound connectivity
Resolved a domain name to confirm DNS was working correctly (thanks to "Use Peer DNS" pulling the ISP's DNS servers automatically)
Tested speeds on a wired device through the switch and on Wi-Fi through each Tenda F6
Walked the property between coverage areas with a phone to confirm seamless roaming on the shared SSID, with internet access holding throughout
Spot-checked connectivity with a few residents on different floors to confirm the fix had resolved the outage building-wide, not just near the core router
Everything came back clean.
Step 6: Locking In Reliability
Getting PPPoE to connect once isn't the whole job. We wanted to make sure it stayed connected. So we also:
Enabled the RB951's built-in PPPoE keep-alive behaviour so a brief ISP-side blip wouldn't require a manual reboot or an on-site visit
Double-checked that DHCP was fully disabled on both Tenda F6s, so there was no risk of them silently re-enabling and causing IP conflicts across the building
Checked firmware/RouterOS versions on the RB951 and the F6s and applied available updates, since outdated firmware is a common cause of intermittent PPPoE drops and Wi-Fi instability
Documented the full topology and working configuration (RB951 PPPoE settings, switch layout, and F6 access point settings) securely for the property manager, in case a device ever needs to be replaced or the network expanded to cover more of the building
Common PPPoE Pitfalls We See Again and Again
If you're attempting a PPPoE setup yourself, especially with more than one networking device on site, here are the mistakes we run into most often:
Letting more than one device try to handle PPPoE. If your main router and a secondary access point (like a Tenda F6) both attempt to dial the PPPoE session, or both run DHCP and NAT, you end up with conflicts and double-NAT issues that are painful to diagnose.
Leaving access points in router mode. The Tenda F6 is a capable router in its own right, but when it's meant to just extend Wi-Fi, it needs to be switched into access point mode with DHCP turned off. Otherwise it'll hand out its own conflicting IP addresses.
Mistyped or copy-pasted credentials with hidden characters. Always type PPPoE credentials manually if pasting isn't working reliably.
Ignoring MTU settings. An incorrect MTU on the PPPoE interface can cause some websites to load while others time out, a confusing, hard-to-diagnose symptom.
No keep-alive configured on the dialing device. Without it, the connection drops and needs manual intervention, often at the worst possible time.
Assuming the switch and cabling are fine without checking link lights first. Physical issues on a switch port masquerade as configuration issues constantly.
Wrapping Up
For this 50-unit property, the whole process, from diagnosis to a fully stable, building-wide connection, took under a few hours once we were on site. PPPoE isn't inherently complicated, but it does require getting several small details right: correct credentials, correct connection type, sensible MTU, and a reliable reconnect policy. At scale, it also means making sure every downstream device (switch, access points) is configured to complement the core router rather than compete with it.
If you're dealing with a similar situation, a new ISP connection that just won't come online, whether it's a single home or a full apartment complex, it's often faster and less frustrating to have someone experienced take a look rather than guessing through router menus. That's exactly the kind of job our team at Luxenetworks handles regularly, and we're always happy to help get your connection stable and secure.
Jul 24, 2026
Why Safaricom and MTN Keep Losing to a Guy With a MikroTik and a Ladder
Why Your Neighborhood Wi-Fi Guy Is Beating Safaricom and MTN at Their Own Game
Walk through almost any estate, market, or trading center in Kenya or Uganda and you'll find a small router mounted on a rooftop, a hand-painted sign advertising "Wi-Fi 500/= per day," and a shopkeeper who'll top up your voucher on the spot. Meanwhile, Safaricom's BLive/BLAZE public Wi-Fi and MTN's public hotspot sit quietly in malls and airports, rarely mentioned, rarely used compared to the volume these grassroots networks pull in.
This isn't an accident, and it isn't because the big telcos lack the money or technology to do better. It comes down to incentives, economics, and geography four things in particular.
1. Big Telcos Are Protecting a Bigger Business
Safaricom and MTN make the bulk of their high-margin revenue from cellular data bundles. That creates a built-in conflict of interest with public Wi-Fi:
If they offered dirt-cheap, genuinely unlimited hotspot access everywhere, people would simply stop buying daily data bundles.
To avoid cannibalizing that core revenue, telco hotspot products tend to be capped, throttled, or priced in a way that doesn't seriously undercut mobile data.
Independent ISPs and street-level Wi-Fi vendors have no cellular network to protect. Selling bandwidth is the entire business, so they're free to price as aggressively as the market will bear which turns out to be very aggressive indeed.
2. The Wholesale Bandwidth Arbitrage Model
Local operators run on a simple, repeatable loop:
Buy wholesale bulk. They lease a fixed fiber connection say 50–200 Mbps at business rates from a backhaul or wholesale carrier.
Oversubscribe the neighborhood. Cheap hardware (MikroTik routers, directional outdoor access points) blasts that connection across an estate, market, or boda stage.
Sell micro-vouchers. Access goes for KES 10–20 or UGX 500–1,000 for a few hours of high-cap or unlimited use.
Because internet usage is bursty not everyone is streaming HD video at the same second the operator can comfortably oversubscribe the line to 100+ concurrent users. That keeps prices low for customers while still generating a healthy margin for the operator. It's the same logic airlines use when overbooking seats, just applied to bandwidth.
Compare that to street vendors selling hourly access for $0.19–$0.31, against budget ISPs charging $9–$12 a month for entry-level plans a pricing structure built around what a cost-sensitive customer can actually spend right now, not around a monthly subscription commitment.
3. Placement: Corporate Coverage vs. Targeted Proximity
Where the access point sits determines who actually uses it.
Telco Hotspots (Safaricom / MTN)
Local Neighborhood ISPs
Typical locations
Malls, airports, city centers, official shops, university centers
Residential estates, informal settlements, local shops, markets, boda stages
Hardware
High-end corporate APs, tighter range limits
Long-range outdoor APs on rooftops, masts, utility poles
Sign-in
Splash pages, OTPs, app logins, SIM-based checks
Instant M-Pesa/MoMo STK push, or a paper voucher bought from the shop next door
Telco hotspots go where people are passing through. Local ISPs go where people are staying put home, work, the corner shop, the place they spend hours every day. That single difference in deployment philosophy explains a huge share of the usage gap.
4. Trust and a Grassroots Reseller Ecosystem
Local ISPs function less like companies and more like community franchises:
They partner with corner shops, cyber cafés, barber shops, and local youth to resell vouchers for a small commission distribution that's dense, personal, and everywhere.
When something breaks, customers message a WhatsApp group or call "the guy" who physically climbs up and fixes the AP not a corporate call center queue with hold music and a ticket number.
Pricing is often flexible in practice: a known customer can get credit, a discount, or a personal favor. A faceless telco billing system can't do that.
That relationship layer builds a kind of trust and stickiness that no splash-page login screen can replicate.
The Twist: Telcos Are Starting to Notice
This gap hasn't gone unnoticed. Safaricom has reportedly been developing a tokenized, pay-as-you-go home internet and public Wi-Fi product, with tokens priced as low as KES 15–100, aimed directly at undercutting the informal vendors and budget ISPs (Poa!, Mawingu, Vilcom, and others) that currently dominate low-income areas. It's a tacit admission that the micro-pricing, hyper-local model works and that beating it requires playing by the same rules the local guys already wrote.
Whether a company the size of Safaricom can actually replicate the "your neighbor fixes your router" trust factor at scale is the real open question. Infrastructure and pricing can be copied; a personal relationship with the shopkeeper down the road is much harder to manufacture from a head office.
Bottom Line
Big telcos built public Wi-Fi as a branding perk or a way to soak up excess bundle allowance for people on the move. Local ISPs and street vendors built it as their entire livelihood a high-volume, low-margin utility engineered specifically for cost-sensitive customers in high-density areas, sold by people the customer already knows and trusts. Until the telcos are willing to compete on price, placement, and relationship all at once, the neighborhood Wi-Fi guy is going to keep winning.
May 22, 2026
Choosing A Tech Course the Right Way
Navigating the Maze: Why Choosing a Tech Course Feels So Hard (And How to Fix It)
So, you have decided to break into tech. You open a course platform, type in "coding," and suddenly you are staring at 50,000 results.
Python, UX design, cloud architecture, cybersecurity, data science: the options are endless. Very quickly, excitement turns into total paralysis.
Choosing a tech course is uniquely overwhelming. Here is an honest look at why this choice feels so incredibly hard, and a simple framework to help you cut through the noise.
Why the Choice Feels Impossible
If you are struggling to pick a path, you aren't lazy or indecisive. The tech education landscape is specifically designed to confuse beginners for three main reasons:
1. The Hype Machine
Every bootcamp advertisement promises that their specific language or tool is the golden ticket to a six-figure job. One week "AI is replacing everything," the next week "cybersecurity has a massive worker shortage." It is hard to choose when the industry's marketing changes every month.
2. The Illusion of Choice
In tech, there are ten different ways to solve the exact same problem. You can build a website using WordPress, React, Python, or No-Code tools. As a beginner, it is impossible to know which ecosystem is worth your time and money.
3. Fear of the "Wrong" Choice
Tech moves fast. It is completely normal to worry that by the time you spend six months finishing a course, the tool you learned will be obsolete or replaced by artificial intelligence.
The Reality Check: Where the Difficulty Actually Lies
The difficulty of a tech course isn't just about the coding syntax. It depends heavily on how you think. Before you swipe your credit card, match your natural brain type to the actual day-to-day work:
[ Your Brain Type ]
│
├──► Logic & Math Heavy ───► Data Science / AI / Backend
│
├──► Visual & Creative ────► UI/UX Design / Frontend Development
│
└──► Process & Systems ────► DevOps / Cybersecurity
The Logic Trap (Data and Backend): If you hate math, statistics, or abstract problem-solving, a Python Data Science course will feel like a mountain.
The Creative Trap (UI/UX and Frontend): If you hate tweaking colors, layouts, and thinking about human psychology, frontend development will frustrate you quickly.
The Chaos Trap (Cybersecurity and DevOps): If you like predictable, structured tasks, these fields will overwhelm you. They require hunting for hidden errors in massive systems.
How to Choose Without the Regret
To beat the analysis paralysis, stop looking for the "perfect" course. Instead, follow this three-step filters framework:
Step 1: Pick a Bucket, Not a Language
Do you want to build things people look at (Frontend/Design), handle the invisible logic behind the scenes (Backend/Cloud), or find stories in numbers (Data Science)? Pick one bucket first.
Step 2: Date the Field Before You Marry It
Never buy a $1,000 bootcamp or a long university course on day one. Go to YouTube. Spend 5 hours watching free, absolute-beginner tutorials. Try to build a tiny project. If you want to throw your laptop out the window after two hours of HTML, you just saved yourself thousands of dollars.
Step 3: Look for "Building," Not "Watching"
When you are ready to buy a course, look at the syllabus. If it is 40 hours of a lecturer talking over slides, skip it. The only way to learn tech is by breaking things. Choose courses structured around building real projects for a portfolio.
The Bottom Line
There is no single "right" course. The secret of the tech industry is that most skills overlap. Learning the basics of logic in a JavaScript course will help you if you later switch to Python.
Pick one introductory course today, commit to it for two weeks, and see how it feels. Action cures anxiety every single time.
