Ascend Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: (ASCEND) BGP 4 ?





On Tue, 25 Nov 1997, Joe  Shaw wrote:

> > I would strongly suggest you reconsider and find some other box for that
> > function on your network. The Max just isn't powerful enough to do
> > anything useful, and even if BGP was implemented in the Max (don't hold
> > your breath), I wouldn't be willing to trust it.  OSPF doesn't even work
> > right yet. 
> > 
> > Multihop BGP to a real router may be an option, or it may not.
> 
> I'm wondering,
> 	Everyone who complains of not getting OSPF to work seems to have
> weird problems.  I'm wondering, are they putting all their maxen on
> network 0?  I've decided that since we got our new Cisco 7200 that I was
> going to put all our maxen on their own little switched part of the
> network and see if that would fix things.  Fortunately, I've decided to
> take that one step further.  I'm going to put all the of the MAXen that
> run OSPF and have to do stacking across multiple chasis (3) onto their own
> stub network. 


 One problem that we have had with attempting to run Max's in their on
stub areas is that they wont incorporate routes external such as routes
fed in by radius, Those are treated as external routes and are not
advertised. 

 Maybe this can be fixed by implementing NSSA's?


 Then, I'm going to put the others that run OSPF on their
> own stub network and see if it doesn't decrease the amount of reboots on
> the MAXen (the only problem I have with OSPF).  I am fairly certain that
> all the shortcomings of the MAXen regarding OSPF have to do with it's lack
> of processor power.

 Agreed, As I have stated several times. The I960 just isn't the most
robust chip on the market. But then again I don't believe the Max 
was desgined to be a full ABR.


  I think that the MAXen just get's so many LSAs that
> it starts re-running the SPF algorithm so many time that it just eats up
> processor and memmory and then just dies.  


 That in conjunction with doing all call setup's , 3 Pri's worth of 
traffic, and doing simple caculations of the LSA's it's gotta
tax the CPU somewhat. Now with the offload of moving the ppp setup down
to the DSP's on the digital modem's cards, Those boxes are a lot less
taxes, As you will see in the improved throughput with 56K cards.


I'm hoping that by reworking
> our OSPF network into stub areas that we'll improve the performance of the
> MAXen and increase their time between reboots.
> 
> BTW, is there still no way to get CPU and memmory consumption information
> on these units?

 NOPE. But, I have heard rumors that maybe soon this will be able on
the TNT. Kevin can you confirm this? 


Jason Nealis
Director Internet Operations
Network Access
Erols Internet


> 
> Joe Shaw - jshaw@insync.net 
> NetAdmin - Insync Internet Services
> Learn more and you will never starve.
> 
> ++ Ascend Users Mailing List ++
> To unsubscribe:	send unsubscribe to ascend-users-request@bungi.com
> To get FAQ'd:	<http://www.nealis.net/ascend/faq>
> 

++ Ascend Users Mailing List ++
To unsubscribe:	send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd:	<http://www.nealis.net/ascend/faq>


Follow-Ups: References: