Hi,
any news about evolution of bgp multicore ?
thanks
Hi,
any news about evolution of bgp multicore ?
thanks
+1
absolutely necessary!
+2.
I have a case just happen today, suspend because of the upstream have a big BGP update at 2014-01-17 10:50 GMT. ~ 10k - 15k route update received from my Cisco router, and pass along to my new CCR-1036. at the sametime, some of my subnet lost connection, and suspect because losing part of the routing table at CCR.
When performing the full bgp table scan every 30 seconds, one of the CORE will run at 100% for more than 20 seconds. I guess because of this, some of the routing table get lost because of timeout.
I’m running 3 upstream with full tables, and around 20 peers connecting to Internet Exchange… My ROS is running the latest 6.7 already, and HW firmware 3.10.
I hope the BGP can work accross different core of CPU, and speed up the table scan…
+3
We’ve been hoping for such an update for a while. It seems that every time a large topology change is passed to the CCR, the router goes nuts due to CPU maxing out. As a result, the routing tables end up hosed and other services start failing quickly. In our deployment, we’ve had to reboot the device to resolve the issue.
This is a serious issue that needs addressed as it really hinders the ability to utilize the CCR series to their full potential.
If Mikrotik is pushing out devices with 36 cores, why isn’t EVERY service RouterOS has multi-core enabled?
+1
Waiting for the “new routing engine”. Full BPG table loading time is hardly acceptable, not what you’d expect from that much CPU power. And finding routes (“/ip route print where …”) takes way too much time on a full table.
Please, MT, give us at least /some/ information, some of us will even beta test it… hint hint ![]()
I’m also waiting for BGP routing functions to work on every core. Had to reboot router few times already because of stuck routes / long table rebuilding time. I even started searching if anyone already put clean Linux on CCR…
+1 from me!!!
Very urgent !!!
So what is a ‘table scan’? Have you opened a ticket with Mikrotik on this and what did they say?
+1
Waiting 3 years for “new routing”
The routing table scan is standard normal behavior of BGP connection, as it need to compare what had been received against it own internal routing table, and determine what is the best path.
I did not raise the ticket, as this is expected for single threaded program. that why everyone asking for the multicore optimization on BGP routing protocol..
I did not raise the ticket, as this is expected for single threaded program. that why everyone asking for the multicore optimization on BGP routing protocol.
This could perhaps be on one site, but on the other side I am really thinking,
if the entire RouterOS system is running on a many core cpu likes the Tilera
hardware platform it could be that the entire RouterOS is benefit from those
action and function.
Also non Tilera platforms should using in my eyes multicore cpu´s more
like a x_86_64 Bit RouterOS with multicore cpu and much more
RAM support would be a really gain and profit for us all.