Switches Juniper

Daniel Roesen dr at cluenet.de
Wed Oct 19 17:04:33 CEST 2011


On Wed, Oct 19, 2011 at 08:28:42PM +0800, Mark Tinka wrote:
> Unlike Cisco newer Juniper platforms, tunneling was not an 
> integrated forwarding process in the Juniper systems, i.e., 
> neither as part of the line card (until now) nor the 
> software, hence the need for a Tunnel PIC to support 
> tunneling.

Not really correct. At least in the older M-series, the "Tunnel PIC" was
nothing more than a $10k I2C dongle which made the PFE put the fabric
ports of that PIC port into loopback mode. No real intelligence on that
PIC at all.

At least that's the explanation I got many moons ago, together with some
hard evidence. PIC identities can change when talking nicely to them. :)

I just can't remember wether the loopback happens on the PIC (PFE
interface chip) or internally on the PFE - IIRC the latter, so I2C
"yeah, customer paid for it!" ID was technically enough. :) Word
is, that JNPR also tried to avoid people oversubbing the PFE by blocking
the PIC port with that "Tunnel PIC". :>

Best regards,
Daniel

-- 
CLUE-RIPE -- Jabber: dr at cluenet.de -- dr at IRCnet -- PGP: 0xA85C8AA0


More information about the ipv6-ops mailing list