My customer is using mikrotik for terminating l2tp channels. About 30 CCR.
Typical scheme is:
Aggregator CCR: CPE:
vlan ---- bridge ---- vpls---- l2tpl-----------l2tp----vpls----bridge-----ethernet
On each CCR can be from 100 to 200 clients approximately.
It has such kind of problems:
- on some of devices there are very strange situations - after rebooting there are no traffic on physical interfaces. In is possible to manage device only via console.
TechSupport said, that they were able to repeat this problem on lab, but cannot say, when this issue will be solved. Their advice is to disable/enable IP addresses on interfaces.
Our solution is - to disable-enable physical interfaces after restart this equipment.
- At this time there are some new issues:
on one of CCRs - CCR1036-8G-2S+ when it was about 550 active l2tp sessions, customers (from CPE side) complain about service degradation (lost packages, no traffic, etc.)
on CCR1036-8G-2S+ it was no problem with CPU load, no problems with bandwidth on interfaces.
when number of l2tp-sessions was more than 600, CCR1036-8G-2S+ begun to lost management traffic (ssh), the console started be delays. It was necessary to reboot it.
The solution at this time is moving part of l2tp clients from one mikrotik to other.
We send information to mikrotik techsupport, but this time it is no answer.
My opinion - it is scaling problems. But unfortunately, I was not able to find any official detail scaling information, so it is very difficult to plan network with this equipment for ISP or SP.
Other problem - a reaction of Mikrotik tech support for ISP issues. Yes I like the equipment of this vendor, but unfortunately the quality of tech support organization, scaling information is not enough for ISP.