Hi all
I want to know what is your cpu load on your Core Mikrotik.
I have AMD 1.4ghz with 512MB ram, HD 40gb, 6 NICs (all on 1 bridge).
This router manage 12mbps (real) of internet, have 320 pppoe connection (average), webproxy and a few rules of firewall.
The cpu average is 90-100%. I think that it 's high value.
What is your value?
single Xeon 2.6 ghz (512kb cache)
on average 730 users
firewall, shaping (for all)
averages 40% cpu load
741 users connected when i looked at the cpu graph
none, but i really shouldn’t have said anything earlier.
Because i investigated why it had such “high cpu load” and now i reduced it to a stable below 20%. The amount of pppoe users online got reduced to 700 in the fixing.
I am not kidding but actually pretty impressed. I didn’t know it had so wrong BIOS settings.
2,4ghz celeron,256mb ram one radio card.Normaly about 10% cpu load..Core router with trafic marking and about 1159 qos, 3.2 pentium with 512mb ram and hypertreading, load generaly 20-30%
Bandwidth averaging at night time is 15mbit+20mbit. Night time is when upload is more than download (peer2peer traffic is unrestricted).
Webproxy is not a good idea in a service provider scenario.
I can’t recommend having 700 users on a single pppoe terminator.
The optimal design to drive a low power unit is contradicting design for a high power pppoe router with lots of users. No matter how powerful the hardware is, the software is not designed for it.
Just to mention a problem as example; All queues (seem to) get reinstalled each time a user log in which in turn pose real problems (packet loss, tcp synchronisation issues).
About the cpus, code cache size is often better than gigahertz. Packet forwarding is about moving in a relativly small loop that benefits greatly from a large code cache.
AMD tends to make much better cpus than Intel but people tend to stick them in low cost, low end, no name motherboards. You are only as strong as your weakest link. This holds true for most non-Intel brands unfortunatly.
Since this is starting to go way off topic, to continue the discussion (if you can call it that), you may reach me at; lists@wm-access.no
Just to mention a problem as example; All queues (seem to) get reinstalled each time a user log in which in turn pose real problems (packet loss, tcp synchronisation issues).