On (2013-09-13 10:09 +0200), Adam Vitkovsky wrote: > Regarding the ISSU I heard there's a plan to run something like two codes on > the LC asics so that one can be upgraded while the other sill performs the > forwarding duties.
Interesting. I wonder if they do this in NPUs like EZ and nPower, where you'll have lot of PPE cores. I guess you could upgrade each PPE core one by one having very marginal effect on overall platform performance during the process. But even then, I suppose there are some controllers/dispatchers which are SPOF, and if they are offline, NPU can't do work. I also assume that on ASIC style devices having ASIC offline, would mean some ports can't do work, I don't think they have parallelism? But I'm not personally holding breath on seeing true ISSU, which I define as guaranteed <50ms stopping to forwarding (it's still whole lot of frames in 100GE) -- ++ytti _______________________________________________ cisco-nsp mailing list cisco-nsp@puck.nether.net https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
