RE: cam timer tuning

Well,

They have the 7500 series for you catalyst 6509 and Nexus converts...

Nexus is a high latency (in my world sorry), high power, high cost switch...

We have all seen the networkworld.com calculation of just the power savings by choosing Arista... Are the F2 and M2 line cards closer or better than Arista? Debateable... but I am of the school most networks are WAY over engineered and way UNDER managed on a day to day basis.

In my experience and in depth weekends trying everything google can think of - Brocade is actually the 10gb iscsi or raw throughput switch of choice.

My whole 2c was there are WAY better options out there before we discuss the whole 3750x vs 4948E debate again :)

Check this link - number 10,001 or so from the master Dr. Welcher -

http://www.netcraftsmen.net/component/content/article/69-data-center/1275.html

Nexus F1 and M1 were a joke and not serious. When I was subcontracting they actually got my general contractor fired for oversubscribing the switch and not knowing the shared and dedicated 10gb ports rule :( the F2/M2 cards are a lot more competitive with Arista and Juniper's raw performance. Will Cisco have a few weird features some Satcomm backbone needs to segregate control frames? Perhaps... but THAT possibility is the niche - not the rule.

thanks

-----Original Message-----
From: Marko Milivojevic [mailto:markom_at_ipexpert.com]
Sent: Saturday, December 29, 2012 12:45 AM
To: Joseph L. Brunner
Cc: me you; Carlos G Mendioroz; Naufal Jamal; ccielab_at_groupstudy.com
Subject: Re: cam timer tuning

While I love Arista, theirs is a very niche market... Even they say so themselves :-)

--
Marko Milivojevic - CCIE #18427 (SP R&S) Senior CCIE Instructor - IPexpert
On Fri, Dec 28, 2012 at 9:33 PM, Joseph L. Brunner <joe_at_affirmedsystems.com> wrote:
> Dude do yourself a favor and switch to Arista.
>
>
>
> Itbs a great product with great people behind it.
>
>
>
> I would say without breaking any NDAbs 60% of all US Equities, Futures 
> and options trade on Arista driven networks. 20% of the rest of the 
> volume moves over Juniper EX series. Cisco, Extreme, BNT and all the 
> others have the crumbs.
>
>
>
> Arista is used in many sophisticated networks and its rock solid.
>
>
>
> -Joe
>
>
>
>
>
> From: me you [mailto:anunda19_at_gmail.com]
> Sent: Saturday, December 29, 2012 12:29 AM
> To: Carlos G Mendioroz
> Cc: Marko Milivojevic; Joseph L. Brunner; Naufal Jamal; 
> ccielab_at_groupstudy.com
>
>
> Subject: Re: cam timer tuning
>
>
>
> All I know is the 4948's suck. Give me a 3750 any day, I have been 
> told that the 4948's are better becasue they do IPv6 software instead 
> of in hardward like 3750. But I don't know for sure I would love to 
> see someone with some documentation on that.  but I see all kinds of 
> problems with the 4948's. if you change a intergrated motherboard with 
> NIC on an end device you have to clear ARP, but not with a 3750. Heard 
> the 4948E are better, but from the text that was provided it looked 
> like a 4948E because the "media" command was not provided for for the 
> G1/49. In Fact the whole topology looks like the same B/S being pushed 
> by my company. if it is you know me
>
>
>
> Rob
>
>
>
> On Fri, Dec 28, 2012 at 5:52 PM, Carlos G Mendioroz <tron_at_huapi.ba.ar>
> wrote:
>
> Hmm, go back to the first (rather cryptic) message:
>
>
>
> whenever we bring up a new
> trunk link between Nexus 7K and 4948 we see unicast flooding caused 
> due to TCN's.
>
>
>
> So in this particular case, we (well, he) knows that a TC has 
> happened, and that it causes the flooding...
>
> -Carlos
>
> Marko Milivojevic @ 28/12/2012 19:48 -0300 dixit:
>
>
>
> TC and event correlation are not cause and effect. The issue is that a 
> host went down and someone was still transmitting to it. And can do so 
> for 4 more hours :-)
>
> --
> Marko Milivojevic - CCIE #18427 (SP R&S) Senior CCIE Instructor - 
> IPexpert
>
> On Fri, Dec 28, 2012 at 1:43 PM, Carlos G Mendioroz <tron_at_huapi.ba.ar>
> wrote:
>
> That won't cut this, because the event happens *when there is a TC*.
> No way you will get ARP down to 15 seconds or so.
> Nah, identify your silent destinations and ping them from the DG :)
>
> -Carlos
>
> Marko Milivojevic @ 28/12/2012 15:38 -0300 dixit:
>
>
> On Fri, Dec 28, 2012 at 10:05 AM, Joseph L. Brunner 
> <joe_at_affirmedsystems.com> wrote:
>
>
> You should unify all timers at 600 seconds.
>
> That number is usually good to me.
>
>
>
> It's hard to change ARp timers in the entire domain. I was thinking 
> more along the lines of dramatically increasing the aging timers in 
> CAM tables to match the ARP :-)
>
> --
> Marko Milivojevic - CCIE #18427 (SP R&S) Senior CCIE Instructor - 
> IPexpert
>
>
> Blogs and organic groups at http://www.ccie.net
>
> ______________________________________________________________________
> _ Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>
> --
> Carlos G Mendioroz  <tron_at_huapi.ba.ar>  LW7 EQI  Argentina
>
>
> Blogs and organic groups at http://www.ccie.net
>
> ______________________________________________________________________
> _ Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
>
>
>
>
>
>
>
> --
> Carlos G Mendioroz  <tron_at_huapi.ba.ar>  LW7 EQI  Argentina
>
>
> Blogs and organic groups at http://www.ccie.net
>
> ______________________________________________________________________
> _ Subscription information may be found at:
> http://www.groupstudy.com/list/CCIELab.html
Blogs and organic groups at http://www.ccie.net