[P4-discuss] (Help) Questions about blog "Improving Network Monitoring and Management with Programmable Data Planes"

Libin Liu liulibinsdu at gmail.com
Fri Jan 13 01:57:07 EST 2017


Thanks a lot, Nate. This really helps me. Thanks, Andy and Nate.

Best Regards,
Libin

2017-01-13 14:48 GMT+08:00 Nate Foster <jnfoster at cs.cornell.edu>:

> That's correct. P4's constructs for describing parsers, actions, tables,
> and controls specify the core packet-processing functionality of the
> switch.
>
> In P4-14, some of the information you mention may be available as metadata
> on some targets, but this is not mandated by the language specification.
>
> P4-16 offers support for "architectures" and "extern functions." In
> principle, one could define an architecture that exposes this information
> and makes it available to the rest of the program through extern functions
> with standard names. Then, any target that implements this architecture
> would need to supply implementations for these functions.
>
> -N
>
> On Thu, Jan 12, 2017 at 10:40 AM, Libin Liu <liulibinsdu at gmail.com> wrote:
>
>> Thanks, Andy. I don't understand that.
>> You mean the P4 language can not guarantee those information, that is the
>> barefoot switch ASIC which can already provide those information, e.g.,
>> link utilization, queue depth, link latency? Thanks a lot.
>>
>> Best Regards,
>> Libin
>>
>> 在 2017年1月13日,02:19,Andy Fingerhut <andy.fingerhut at gmail.com> 写道:
>>
>> My understanding is that whether a particular P4 target architecture
>> supports visibility into such information about the queue depths and link
>> utilizations can vary from P4 target to P4 target.  It is not guaranteed by
>> the language to be available for all of them (nor even are the presence of
>> queues guaranteed by the language, e.g. programmable NIC cards might not
>> have any).
>>
>> I don't have access to details about current P4 capable switch ASICs, but
>> I would guess that Barefoot's Tofino ASIC family gives such visibility.
>>
>> Andy
>>
>> On Thu, Jan 12, 2017 at 5:20 AM, Libin Liu <liulibinsdu at gmail.com> wrote:
>>
>>> Dear P4 developers,
>>>
>>> Hope this email finds you in the best of your health and spirits.
>>> I am very interested in this blog. Can P4 switch already directly
>>> measure the link utilization, link latency, and switch queue? If so, where
>>> can I get the open source code about the module of measuring link
>>> utilization, link latency, and switch queue? Thanks a lot for your help.
>>>
>>> Best Regards,
>>> Libin
>>>
>>> _______________________________________________
>>> P4-discuss mailing list
>>> P4-discuss at lists.p4.org
>>> http://lists.p4.org/mailman/listinfo/p4-discuss_lists.p4.org
>>>
>>
>>
>> _______________________________________________
>> P4-discuss mailing list
>> P4-discuss at lists.p4.org
>> http://lists.p4.org/mailman/listinfo/p4-discuss_lists.p4.org
>>
>
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.p4.org/pipermail/p4-discuss_lists.p4.org/attachments/20170113/70d18155/attachment-0002.html>


More information about the P4-discuss mailing list