The beauty is that you can shove Pi in it of course.
Don’t fall for the trap that they recommend an expensive Pi 5: I am running Pi-hole on a Pi 2 but you can basically run this on obsolete hardware, whether that’s a Pi or a PC/laptop
66.6% of all traffic is blocked with no functional impact on anything that I do
Okay. I’m convinced.
Misleading statement. It doesn’t block “traffic”, it blocks DNS requests… you don’t know how much traffic this corresponds to.
Correct. The payload of DNS requests is tiny compared to, say requesting a webpage. So there might not be a huge decrease of bandwidth usage reduction. However, having 66.6% less DNS requests is still a win. The router/gateway doesn’t have to work that hard because of the dropped requests.
It isn’t so much about the payload of the DNS requests, but about the content that would have been loaded if the DNS request hadn’t been blocked.
If you load a page that has 100kB of useful information, but 1MB of banner ads and trackers … you’ve blocked a lot more than 66%. But if you block 1MB of banner ads on a page that hosts a 200MB video, you’ve blocked a lot less.
Also a 66% blocked percentage seems very high. I have installed pihole on 2 networks, and I’m seeing 1.7% on my own network, but I do run uBlock on almost everything which catches most stuff before it reaches the pihole, and 25% on the other network.
I run a handful of instances across different networks, 1.7% is suspiciously low, you should make sure you’ve got the right lists. I like HageZi’s
Of course, because ads have zero bandwidth. /s
Are you an idiot?
As per the article
on my own network a whopping 66.6% of all traffic is blocked
I stated it’s actually 66.6% DNS requests being blocked, not the raw bandwidth utilization. Raw bandwidth savings (by not downloading the non-zero ads) would be much lesser.
Can’t we be nicer on the internet?
No, raw bandwidth savings would likely be very significant. You do realize that for many webpages the ads are most of the bandwidth? On my network (I have capped internet so this is important) if I run dns ad blocking my total bandwidth is 40% less.
I’m not sure whether it makes sense trying to discuss with you but let’s try…
You couldn’t know how much traffic you saved because you didn’t load the ad. The ad could be 1KB, 1MB or 1GB, but because you didn’t load it you wouldn’t know it’s size. Without knowing it’s size, you wouldn’t be able to calculate the savings.
As mentioned somewhere is in the thread you would have to directly compare two machines visiting the same pages and even then it’s probably only approximate because both machines might get served different ads.
I’ve been thinking of setting one up for a while, if I have a home server would I be better off hosting it on that or as a separate device? What are the alternatives to a raspberry pi? They’ve shot up in price over the years.
If you have a server running, I wouldn’t buy more hardware. They have good example documentation for just such a configuration:
https://docs.pi-hole.net/docker/
If your server already has those ports bound (specifically the DNS port 53) you are going to have to get creative; otherwise it’ll work well!
Worst case, a cheapo pi 3 will do the job. At one point I had it running on a pi zero, so hardware requirements are pretty low.
If your using docker and the ports are bound you can just use the network mode host so the container gets it’s own ip. It’s how I have adguard running on my unraid server
edit: Sorry I mixed up the details as @starkzarn@infosec.pub pointed out. It’s a macvlan configuration. My intention was to point out it’s possible. Here’s some documentation https://docs.docker.com/engine/network/drivers/macvlan/
That’s not how that works.
network_mode: host
shares the network namespace with the container host, so it doesn’t do any NAT, it only exists on the host’s IP. It would be akin to running a natively installed app, rather than in a container.macvlan
networking is what gives a container its own IP on the logical network, without the layer of NAT that the defaultbridge
mode networking that docker typically does.
phrasing
I’m reasonably certain the name was intentional because of the way it could be phrased.