RE: 64 bit counters in MPLS MIBs

["Harrington, David" <dbh@enterasys.com>]

Hi,

The business reality is that many SNMP consumers continue to utilize tools that only support SNMPv1. In those cases, they cannot retrieve the counters if they are defined only as Counter64s. They can retrieve HC counters.

It doesn't help the MPLS community to define mibs they cannot use effectively in many real-world environments, so only supporting the Counter64s seems counter-productive from the standpoint of meeting customers' needs. Supporting HCs, of course, seems counter-productive to the community producing standards, and the desire to have customers migrate to tools that do support SNMPv3.

I suggest that the best approach is to design mibs to support both approaches, but to indicate that the Counter64 approach is preferred and the HC approach is not, but may be supported by some vendors for backwards compatibility. Failure to provide standardized HC counters would likely lead to proprietary mib modules containing the HC objects (or worse alternatives), which would hurt interoperability. 

The SMI has features designed for this purpose. The IETF/IESG could require that HC counters in a standards-track mib module be given a status of deprecated (preferably with a reason given in the mib module for that status), and the compliance statement for the mib module could have the Counter64s in the basic compliance statement and the HCs in a separate compliance statement (or optional group). 

Vendors meeting the demands of customers for SNMPv1 capability can support the standardized deprecated HC objects. When enough customers move to SNMPv3, vendors can choose to stop supporting them, and still claim compliance.

my $.02
dbh

> -----Original Message-----
> From: Wijnen, Bert (Bert) [mailto:bwijnen@lucent.com]
> Sent: Monday, January 06, 2003 12:08 PM
> To: Mreview (E-mail)
> Subject: FW: 64 bit counters in MPLS MIBs
> 
> 
> I had suggested to MPLS MIB people that at places
> where they use parallel 64 and 32 bit counter (the
> 32-bit values fo those systems that do not support 
> 64bit), that it might be better to use just 64 bit
> and forget the 32 bit counters.
> 
> This is what I am getting back. Any opinions/input
> or comments?
> 
> I understand that we will NOT change the description
> of counter64 of course.
> 
> Thanks,
> Bert 
> 
> -----Original Message-----
> From: Thomas D. Nadeau [mailto:tnadeau@cisco.com]
> Sent: maandag 6 januari 2003 16:11
> To: Wijnen, Bert (Bert)
> Cc: Joan Cucchiara x302; Cheenu Srinivasan
> Subject: 64 bit counters in MPLS MIBs
> 
> 
> 
> 	I have a question RE: 64 bit counter changes
> in the MIBs.  During the MIB review, you (Bert) had
> suggested that we delete the counters called
> HC and change the type of the existing counters
> from 32 bits to 64 bits.   My question is when we
> support these on platforms that indeed do not
> support 64 bit counters, can we add something
> to the description that allows for this? I have been
> discussing this change with some developers
> here and have been getting serious push back
> based on the fact that on platforms that do not
> natively support 64 bit counters one must emulate
> them at the process level. On the surface this seems
> reasonable. However, when you look at the number of
> emulated counters in just say the LSR MIB which has
> per label counters (there can be 100,000s of labels),
> this starts looking very unpalatable very quickly.
> 
> 	So I think we either need to continue
> supporting the 32 bit versions and/or change
> the description of Counter64 to allow for a variation.
> 
> 	--Tom
> 
> 
>