> Ah, it *is* the documentation which is obtuse :-) Your
> message made me check a bit more, and I finally found the
> "show interfaces mac-database" command.
>
> Now if the "Configuring MAC Address Accounting" documentation
> had been a bit more explicit about what it did, and had
> actuall
> > Is it just me, or is the JunOS documentation for Ethernet MAC address
> > accounting incredibly obtuse?
> >
> > I am connected to an IX on GigE shared L2 - so I would like to count
> > number of bytes and packets transmitted to and received from each BGP
> > peer, based on the MAC address of t
On Sat, Mar 17, 2007 at 09:41:03PM +0100, [EMAIL PROTECTED] wrote:
> Is it just me, or is the JunOS documentation for Ethernet MAC address
> accounting incredibly obtuse?
>
> I am connected to an IX on GigE shared L2 - so I would like to count
> number of bytes and packets transmitted to and recei
Is it just me, or is the JunOS documentation for Ethernet MAC address
accounting incredibly obtuse?
I am connected to an IX on GigE shared L2 - so I would like to count
number of bytes and packets transmitted to and received from each BGP
peer, based on the MAC address of the peer. This would seem
> That said, you could look into adding L3 and L4 inspection for
> (potentially) better hashing. Enabling these under [edit
> forwarding-options hash-key family-inet] will start considering
> factors like:
> Source IP address
> Destination IP address
> Protocol
> Source port number
> Destination
> >>I've heard that although the load balance option is known as
> >>"per-packet" but it behaves more like "per flow". Meaning packets
> >>would not be breaked up and merged on the other end. Am i right?
>
> * [EMAIL PROTECTED] [Fri 16 Mar 2007, 20:57 CET]:
> >It *is* per flow. The old IP 1 coul
On 3/17/07, David Ball <[EMAIL PROTECTED]> wrote:
>when doing the hashing. I haven't tried it yet personally, but it
> likely comes with a CPU hit of some kind.
>
> David
Hi David,
There is no CPU hit associated with the hash-key statements on M/T-series.
In fact hardware always uses hashin
> * [EMAIL PROTECTED] [Fri 16 Mar 2007, 20:57 CET]:
> >It *is* per flow. The old IP 1 could do per packet, current IP 2 cannot.
>
> That's a feature. Per-packet load-balancing can easily lead to packet
> reordering.
>
I dunno if I'd say 'easily', as the delta would have to be pretty
significan
>* [EMAIL PROTECTED] (Kanagaraj Krishna) [Fri 16 Mar 2007, 20:51 CET]:
>>I've heard that although the load balance option is known as
>>"per-packet" but it behaves more like "per flow". Meaning packets
>>would not be breaked up and merged on the other end. Am i right?
* [EMAIL PROTECTED] [Fri 16
9 matches
Mail list logo
