I would like to open a technical discussion based on the experience many ISPs have had using MikroTik over several years in production networks.
There is no doubt that MikroTik has been a key piece in the growth of many small and medium ISPs thanks to its flexibility, low cost, and RouterOS. However, the landscape today is different.
Current scenario
Many ISPs are no longer operating small networks. We are now dealing with architectures with:
30,000 to 50,000 PPPoE users
High and constant 24/7 traffic
Intensive use of firewall, mangle, and QoS
Need for real high availability
Integration with FTTH OLTs and mass authentication systems
The emerging issue
At this scale, some points become critical:
CPU saturation on PPPoE concentrators under high load
Limitations in processing large-scale firewall/mangle rules
Complex QoS affecting overall performance
Heavy reliance on manual design to scale properly
Lack of a true carrier-grade focus in the base architecture
Open question
The discussion I want to raise is:
Should MikroTik develop a product line specifically designed for large-scale ISPs (30K–100K PPPoE users), with a true carrier-grade focus?
Some ideas that could be key:
True hardware offloading of PPPoE sessions
Architecture optimized for millions of packets per second
Better multi-core utilization in RouterOS
Active-active redundancy at core level
Factory-designed for massive ISP environments, not just general flexibility
Personal opinion
It’s not that MikroTik is insufficient, but rather that the market has evolved rapidly.
Many ISPs are now at a stage where they are no longer looking for something that “just works,” but rather carrier-grade stability without so much manual optimization.
I would like to hear the opinions of other ISP engineers:
Have you reached the limits of MikroTik in large networks? What solutions are you using to scale PPPoE at high volume? Do you think MikroTik should move more strongly into the carrier-grade world?
It's the only solution. I see that the big phone companies handle PPPoE without problems. I understand that they should make the equipment more robust since their customers have different needs nowadays.
We terminate our PPPoE on euh ... slightly bigger boxes but we also provide IPoE now.
Now I'm not from that department and don't have the nifty details, but we have > million subscribers
What subscriber termination platforms (BNG/BRAS) are you currently using? Approximately how many PPPoE subscribers do you handle per device?
I would also like to know whether you are still using PPPoE or if you have migrated, either partially or completely, to IPoE/DHCP.
I have noticed that many of the largest telecom operators and ISPs in the industry continue to rely on PPPoE. In your case, how do you manage authentication and subscriber administration for such a large customer base, ranging from hundreds of thousands to potentially millions of subscribers?
Are you using FreeRADIUS, a proprietary authentication platform, or another specific solution for subscriber authentication, authorization, accounting (AAA), and customer management?
We use very big Nokia boxes (eg. SR7750 & SR7705) and have quite some deployed across the country. We also have a engineering partnerships with Nokia for the past 20 years so we are not just a customer.
Cannot give exact numbers.
I also don't know the exact PPPoE <> IPoE ratio's at this moment. For sure IPoE is rapidly getting adopted.
The AAA looks tightly integrated with lots of other systems, running some large databases in the backend. We have many many "in house" developed things running for the past decades.
But this equipment is completely in a different league then Mikrotik both on hardware and software.
There is Nokia FP (network processor) silicon in these boxes providing massive performance.
Current scenario
The emerging issue
Open question
Personal opinion
Have you reached the limits of MikroTik in large networks?
but we also provide IPoE now.