[ofa-general] RE: QoS for iSER

Yevgeny Kliteynik kliteyn at dev.mellanox.co.il
Sun Nov 18 07:21:24 PST 2007


Kanevsky, Arkady wrote:
> I think providing QoS per vertical (that is ULP) is less prevelant
> as providing QoS per horizontal, that is application, which uses
> miltiple ULPs.

No problem, there's a way to define it per application too.
Administrator will just have to define qos-level, and set of
match-rules for this application that point to the defined
qos-level.
The cons here is that administrator will have to really
understand what ULPs and which ports this application is using,
but I can't think of any way to simplify this task.
Perhaps in the future we will have examples for common applications.

-- Yevgeny

> Arkady Kanevsky                       email: arkady at netapp.com
> Network Appliance Inc.               phone: 781-768-5395
> 1601 Trapelo Rd. - Suite 16.        Fax: 781-895-1195
> Waltham, MA 02451                   central phone: 781-768-5300
>  
> 
>> -----Original Message-----
>> From: Yevgeny Kliteynik [mailto:kliteyn at dev.mellanox.co.il] 
>> Sent: Sunday, November 18, 2007 8:19 AM
>> To: Sean Hefty
>> Cc: gdror at mellanox.co.il; 'Hal Rosenstock'; 
>> general at lists.openfabrics.org
>> Subject: Re: [ofa-general] RE: QoS for iSER
>>
>> Sean Hefty wrote:
>>>>> And as you've mentioned, some rules may overlap. For instance, if 
>>>>> the rule for all the RDS traffic will appear before the 
>> iSER rule, 
>>>>> then iSER requests will be caught by the RDS rule.
>>>> That doesn't sound so good but I don't see a good alternative here 
>>>> other than for this case to put the iSER rule first. The other 
>>>> fallback is the more detailed configuration but RDS falls into the 
>>>> generic range category which is problematic in terms of this (and 
>>>> can't be differentiated by ServiceID unlike the other ULPs).
>>> I'm not overly familiar with the details of RDS, but event if the 
>>> active side uses a dynamic service ID, I would expect the 
>> passive side 
>>> to use something well known.
>> Couldn't agree more.
>> That's why I think that although there are cases where this 
>> simplified way of defining SLs per ULP plus target TCP port 
>> won't be useful, in many cases it would actually make the 
>> administrator's life easier.
>>
>> -- Yevgeny
>>
>>> - Sean
>>>
>> _______________________________________________
>> general mailing list
>> general at lists.openfabrics.org
>> http://lists.openfabrics.org/cgi-bin/mailman/listinfo/general
>>
>> To unsubscribe, please visit 
>> http://openib.org/mailman/listinfo/openib-general
>>
> 




More information about the general mailing list