I want to know what makes it possible for ERPS to converge under 50 ms? it's a hardware-specific feature that provides this fast convergence or it is about the much more efficient algorithm? I read the ERPS Wikipedia and my sense is that the ERPS algorithm is much more centralized unlike the RSTP/MSTP and this helps faster convergence.

  • Have you read the G.8032? – Zac67 Oct 3 at 10:01
  • @Zac67, Actually no. I did take a look at it but have no time to read it carefully. I just understand RPL...... All I want to know is whether different and better logics in ERSP's algorithm (or APS as G.8032 says) make it so fast or it's something else? – A.A Oct 3 at 12:40
  • 2
    Different? Yes. Faster? Mostly. Better? Under ideal circumstances, M/RSTP can also react very quickly, so "better" depends on your network and requirements. – Zac67 Oct 3 at 12:46
  • I didn't have tested APS, but have tested RSTP with simple to more complicated topologies with 4 to 13 nodes and didn't reach convergence time shorter than 5 to 6 seconds.... and that would be amazing if ERPS could shorten this period to something around 50 ms. And this hits me and makes me ask: what does APS have that M/RSTP doesn't? If it is an algorithmic matter, why do not all switches support APS as M/RSTP? – A.A Oct 3 at 13:11
  • 1
    For a larger network, you should limit the L2 segment sizes and use L3 links in between. That way, spanning trees stay small, convergence happens quickly and L3 links fail over and load-balance better anyway. – Zac67 Oct 3 at 18:05

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.