FreeSoft Labs All articles
Lab Benchmarks

One Tiny Pi, Five Screaming Devices: 30 Days of Free DNS Filtering on a $35 Raspberry Pi Zero 2 W

FreeSoft Labs
One Tiny Pi, Five Screaming Devices: 30 Days of Free DNS Filtering on a $35 Raspberry Pi Zero 2 W

Photo by Photo by Rahul Mishra on Unsplash on Unsplash

Most DNS filtering tests happen on hardware nobody owns. A spare mini PC, a VM on a beefy host, a container on a NAS. This time we went the other direction. The FreeSoft Labs bench got a Raspberry Pi Zero 2 W — the $15 board that fits in an Altoids tin — and we pointed every phone, laptop, tablet, Chromebook, and TV in a five-person household at it for 30 days.

The goal wasn't to prove ad blocking works. It does. The goal was to find out whether the cheapest credible Pi can survive the actual chaos of a family: three kids between 9 and 16, fourteen network devices, a smart TV that phones home constantly, and at least one urgent 10 p.m. need for something the filter probably shouldn't have blocked.

The Hardware, and Why We Kept It Absurdly Cheap

The board is a Pi Zero 2 W, which is a quad-core 1 GHz chip with 512 MB of RAM and Wi-Fi baked in. Our total build:

Call it $38 before shipping. Week one ran on Wi-Fi. Week two got a $9 USB Ethernet adapter, because Wi-Fi DNS lookups occasionally spiked to 200 ms and one kid's Chromebook would briefly lose name resolution during video calls. If you're going to do this for real, budget the adapter. Forty-seven bucks all-in is still less than a year of most commercial filtering subscriptions.

One housekeeping note: we moved logging to RAM on all three tools. The first SD card we burned through lasted three weeks on a default Pi-hole install, which is a good reminder that DNS servers write a lot.

Setup Friction: Three Very Different First Hours

Pi-hole was the fastest to a working state. The curl install script, a few prompts, and about ten minutes later we had a web UI on port 80 and default blocklists running. Anyone who has installed Pi-hole before will recognize the whole ritual.

AdGuard Home took about fifteen minutes, mostly because the setup wizard on port 3000 walks you through upstreams and admin credentials more politely. The web UI is more polished, and per-client settings are exposed without editing files.

Blocky was the odd one out. No web UI. A YAML config file, a binary or a container, and a list of upstream resolvers you write by hand. Our first working config took 90 minutes including a detour through the docs to figure out the caching block. If you manage infrastructure for a living, that's not a problem. If you're the family's unofficial IT department and you have a day job, it's a real cost.

Query Volume and What the Block Rates Actually Mean

Our household pushed 80,000 to 110,000 DNS queries a day. The smart TV alone was good for 20,000 of them, which surprised nobody who has watched a Roku idle on a network.

Block rates over the month:

It would be easy to declare AdGuard the winner. Don't. A higher block rate mostly means a bigger list, and a bigger list is a bigger surface for false positives. The number we cared about more was how often someone in the house had to ask us to unblock something legitimate.

The 10 p.m. Problem: False Positives and Family Diplomacy

This is the part no benchmark chart captures. Over 30 days, we logged requests to allowlist specific domains that a family member actually needed:

The worst one was a school portal. Our district's single sign-on sits on a domain that appears on more than one tracker list, so the 13-year-old got a blank page on a Sunday night while trying to submit an assignment. On Pi-hole, the fix took about two minutes in the web UI once we figured out which query to look at. On AdGuard Home, the kid had already installed the AdGuard client on her phone, so she could see the block and ask us to whitelist it from her side. On Blocky, it meant SSH, editing the YAML, and a service restart.

Other repeat offenders: a USPS tracking link that redirected through a flagged analytics domain, a Disney+ authentication flow on the living room TV, and a Roblox update endpoint that got caught by an aggressive list. None were catastrophic. All of them cost social capital.

Updates, Memory, and the Babysitting Tax

On the Pi Zero 2 W's 512 MB of RAM, all three tools fit comfortably, but the spread is worth knowing:

CPU stayed under 10 percent at our query volume for all three. Updates were the bigger differentiator. Pi-hole needed a manual pihole -up most weeks, and its gravity list updates ran nightly without drama. AdGuard Home updated cleanly with a one-liner in cron, and its filter list refresh was invisible. Blocky updated on our schedule, which is to say we pinned a version and touched it twice in 30 days — stable, but you own the maintenance.

So Which One Would We Deploy Again?

For a household with kids and a mix of devices, AdGuard Home gets our nod. The per-client views are the difference-maker: being able to see which device triggered a block, from that device, turns a 10 p.m. crisis into a 30-second fix. Pi-hole remains the community default for good reasons and is still the easiest first install, especially if you like the familiar dashboard. Blocky is a great fit for anyone who already manages config files and wants a tiny, predictable resolver with zero web surface — just don't hand it to the least technical adult in the house.

All three survived a month on a $38 board. The Pi didn't fall over. The family did, twice, over a school portal and a package. That's the benchmark that matters.

All Articles

Related Articles

The 500-Page Office Test: Three Free Wikis on a $60 Secondhand Mini PC

The 500-Page Office Test: Three Free Wikis on a $60 Secondhand Mini PC

Twenty 'Can You See My Screen?' Calls Later: Three Free Remote-Support Stacks on a $99 Mini PC

Twenty 'Can You See My Screen?' Calls Later: Three Free Remote-Support Stacks on a $99 Mini PC

The 200-Device Closet Test: Two Free Asset Trackers, One Cheap Dell, and 45 Days of Sticky Labels

The 200-Device Closet Test: Two Free Asset Trackers, One Cheap Dell, and 45 Days of Sticky Labels