[squid-users] Squid scales up tcp traffic to adsl users

[Ben Goz]

By the help of God.

Sorry I forgot to say that cache is disabled in this setup and the
increasing traffic rate on the clients side can cause a bottle neck as it
reaches the maximum
rates that the customers have.
Is there any option to limit the traffic?

‫בתאריך יום ו׳, 29 בדצמ׳ 2023 ב-13:15 מאת ‪‪Amos Jeffries‬‬‏ <‪
squid3 at treenet.co.nz‬‏>:‬

> This may be normal. A proxy cache like Squid moves objects closer to the
> clients, reduces upstream traffic and multiplexes transactions. All of
> which increase the traffic bandwidth efficiency. Allowing clients to
> receive their downloaded content faster, and thus users can browse through
> more pages faster.
>
> This type of user traffic increase can be just the result of users doing
> more.
> Notice that the WAN/eth0 traffic has decreased by ~30% despite this LAN
> increase.
> You can check the Hit Ratio in squid mgr:info report lines up with the
> increased efficiency.
>
> Cheers,
> Amos
>
>
> -------- Original message --------
> From: Ben Goz
> Date: Mon, 25 Dec 2023, 04:11
>
> Hi,
> This is basically the network topology that I'm using:
>  adsl <--> vrf <--> <eth1> [squid/icap machine] <eth0> <--> vrf <-->
> <internet>
>
> When traffic goes via squid I see that eth1 (The one closes to adsl users)
> is very high this is from sar output:
>
> Average:        IFACE   rxpck/s   txpck/s    rxkB/s    txkB/s   rxcmp/s
> txcmp/s  rxmcst/s   %ifutil
> Average:           lo   4191.83   4191.83  16976.00  16976.00      0.00
>    0.00      0.00      0.00
> Average:         eth0   2921.48   1224.57   3432.46    485.77      0.00
>    0.00      0.00      0.28
> Average:         eth1   7558.70  11544.74    920.91  *14447.44*      0.00
>      0.00      0.00      1.18
>
> When traffic doesn't go via squid this is the output:
>
> Average:        IFACE   rxpck/s   txpck/s    rxkB/s    txkB/s   rxcmp/s
> txcmp/s  rxmcst/s   %ifutil
> Average:           lo     19.10     19.10      2.25      2.25      0.00
>    0.00      0.00      0.00
> Average:         eth0   3666.40   2133.70   4608.70    409.59      0.00
>    0.00      0.00      0.38
> Average:         eth1   2213.40   3741.10    424.38   *4613.08*      0.00
>      0.00      0.00      0.38
>
> I'm can't tell for sure that this is related but I saw several times the
> kernel prints:
> TCP: out of memory -- consider tuning tcp_mem
>
> The squid version I'm using is:
> /usr/local/squid/sbin/squid -v
> Squid Cache: Version 6.5-VCS
> Service Name: squid
>
> This binary uses OpenSSL 3.0.2 15 Mar 2022. configure options:
>  '--with-large-files' '--with-openssl' '--enable-ssl' '--enable-ssl-crtd'
> '--enable-icap-client' '--enable-linux-netfilter' '--disable-ident-lookups'
>
> And I turned off persistence from client, icap and server sessions.
>
> What could be the problem?
>
> Thanks,
> Ben
>
>
> _______________________________________________
> squid-users mailing list
> squid-users at lists.squid-cache.org
> https://lists.squid-cache.org/listinfo/squid-users
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.squid-cache.org/pipermail/squid-users/attachments/20231231/458dd49b/attachment.htm>