AdBlock effectiveness

Is anyone better? Beyond Active Sources - the private blacklist contains only 17 items.

AdBlock is my favorite.

image

image

s0.2mdn.net
googleads.g.doubleclick.net
adclick.g.doubleclick.net
pagead2.googlesyndication.com
www.googleadservices.com
ookla-d.openx.net
cat.nl.eu.criteo.com
pix.eu.criteo.net
cat.fr.eu.criteo.com
criteo.net
criteo.com
clicktrack.pubmatic.com
static.criteo.net
i.seznam.cz
a.denik.cz
content.aimatch.com
a.iprima.cz

my experience is that i had to disable it on the router, as it completely degrades web pages performance mostly. I found adblock as web browser extension plugin much more effective, even if then i have to use it on every device separatelly.

I don’t find Adblock that effective. The Blacklist doesn’t seem to work especially concerning Google domains I want crippled. Overall it does improve the network speed and combined with Pi-Hole they definitely are doing some much needed filtering.

Doesn’t AdBlock seem effective to you? What is your percentage of blocked domains. What lists did you choose? Did you follow the recommendations of max 5 optional lists? How many of your own records have you added to blacklit?

I’m getting Blocked DNS Requests 2474 (56.13%)

This is my config for blocklist sources. I’ve added some blacklist and whitelist entries for various things on my network like a Roku and an Xbox. I’ve also whitelisted a few things that get blocked by overzealous sources from time to time.

I really think the Adblock has saved me some heartburn from not getting drive-by attacks on my family’s devices. I don’t really have any solid evidence for this except that our house hasn’t had any problems with that, where other folks I know have issues from time to time (maybe once a year).

1 Like

Your number of blocked domains (280 tousand) is more than 10x larger !!! than me (22 thousand). I have a reload of 7 sec - in your case it’s 1:40 min.

I think that in the discussions about AdBlock, I read that with a high number of blocked domains, AdBlock may not work correctly (it is also related to the available memory - 1Gb for me).

I used to see that really blocked Google domains were still showing ads to me. Adblock and router restarts didn’t help. Only after I reduced the number of blocked domains does AdBlock work without errors

I’ve got the 2GB model. Never the less, reload of 1:40 is still pretty small all things considered. I only update my lists once a week. I think anything more often than that is overkill.

I’ve not really read any discussion on Adblock memory usage and the number of domains. But it would be usage by kresd since it’s adding to the local domain lists.

ps -aux lists kresd as using 381MB. I expect there would be different numbers if it was using dnsmasq or bind. There also may be different effectiveness depending on the servers.

Effectiveness shouldn’t be different but I would guess bind and kresd are both used more for enterprise level environments and dnsmasq is not.

Could you please elaborate on that, what do you mean in detail (examples)?

Same to you … more details/examples please - thanks! :wink:

I can’t list the examples exactly now, they were domains added by me in the blacklist - I checked its existence in the Blocklist Query - for sure I restarted AdBlock or router and the mentioned ads were still permanently displayed.

googleads.g.doubleclick.net
adclick.g.doubleclick.net
pagead2.googlesyndication.com
www.googleadservices.com

Now, I don’t see such a problem, nor have I tested the limit total number of domains or number of domain lists from which the problem will occur.

EDIT: I’m not sure if it was related to the AdBlock version (if there was an update over time)

Uh, wow. The RPZ in current kresd isn’t really designed for this large lists. I’m glad that it’s still well usable for you. For each RRset in the RPZ, an anonymous Lua(JIT) function is created, based on what the line prescribes… but LuaJIT is generally pretty efficient, all things considered.

Should I move to using BIND?

I have 8 sources checked. One, “notracking” is listed as XL and probably consists of the bulk of the list. The rest are S and one M.

I don’t think you need to, if the performance seems OK. I suspect that BIND might be more memory-efficient for such large RPZ. At least in near term; we have longer-term design plans for a different approach to RPZ and other rule-lists.