Believe me, if the 'bare metal' Mikrotik x86 were still being developed and was a 64 bit variation, it would be the weapon of choice. Mikrotik chose to point their developer dollars at the virtual world.
That said, benchmarks have shown certain optimized linux variants to run at about 98% of bare metal speed in virtual environments. Paravirtualized drivers (which the CHR uses) run pretty much at bare metal capacity. Sure, purpose built Cisco or Xtreme or Juniper or what have you can (and probably do) go faster. If the time comes that I need them, I'll get them (although Cisco will be a very tough sell for me). I was a skeptic too, until I deployed a few CHR. And, a pretty "loaded" CHR platform can be built for less money than a CCR1072, so that's a bonus. 6 Gig ethernet and 6 10G Fiber ports for about $2k. As for the ER Infinity, I have a brand new one, in the box I will never deploy. It's probably fast as lightning, but felt totally alien when I was configuring it. Plus, Sonar has no integration for it. :) YMMV, JP On Wed, 2 May 2018 19:24:23 -0600, Tim Densmore wrote > I've been told that by other folks as well. It's hard for me to > grasp how adding abstraction would make something faster than its > bare metal counterpart. Given that, though, it leads me to wonder > how CHR compares to csr1000v or XRv in that respect. I've used both > quite a bit for labbing. I've never moved a single bit of actual > traffic with them though. > > So far, it doesn't sound like many people have used the ER infinity. > > On 5/2/2018 3:55 PM, J Portman wrote: > > However, a CHR in VMWARE vastly outperforms the CCR in bgp > > convergence. From boot to full tables in under 20 seconds. > > _______________________________________________ > Ubnt_users mailing list > [email protected] > http://lists.wispa.org/mailman/listinfo/ubnt_users Joe Portman Alamo Broadband Inc. www.alamobroadband.com _______________________________________________ Ubnt_users mailing list [email protected] http://lists.wispa.org/mailman/listinfo/ubnt_users
