[c-nsp] "New" IOS release time frame, when bug is identified

Phil Mayers p.mayers at imperial.ac.uk
Tue May 19 06:32:13 EDT 2015


On 15/05/15 19:24, Mark Tinka wrote:
>
>
> On 15/May/15 18:36, Christian Kratzer wrote:
>>
>>
>> this is the time for:
>>
>>      <favourite-rant-about-the-virtues-of-open-source-hardware/>
>>
>> immediately followed by:
>>
>>      <favourite-rant-about-the-availablity-of-open-source-hardware/>
>
> I think open source hardware is abound.

How open *is* the whitebox stuff? I'm under the impression the 
Broadcom-based stuff uses binary blobs compiled into the kernel, and 
that coding to it directly entails NDAs and/or SDK licensing?

Writing a routing protocol is one thing. Writing the code to correctly 
program the FIB on multiple platforms... Hardware is messy.

> The issue with open source solutions has tended to be more with software
> refinement, than the hardware itself.

Maybe. Merchant silicon is deeply impressive these days, but it's still 
very... thin... in a few areas. Buffer sizes are a classic example. 
Expensive RAM is expensive, after all...

> This is where the Cisco's and Juniper's lick the rest, because
> developing software is easy. But developing software that works
> reliably, now there is a story we can all tell.

And there's not enough market for a Linux-alike in terms of 
control/forwarding plane, because there are far fewer network devices 
and network engineers than there are server/desktop and sysadmins/users.


More information about the cisco-nsp mailing list