[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: Advice on restricting SIZE of InetAddress in INDEX?
- To: bwijnen@lucent.com
- Subject: RE: Advice on restricting SIZE of InetAddress in INDEX?
- From: Bill Fenner <fenner@research.att.com>
- Date: Tue, 11 Jul 2000 14:21:00 -0700
- Cc: mibs@ops.ietf.org
- Delivery-date: Tue, 11 Jul 2000 14:21:08 -0700
- Envelope-to: mibs-data@psg.com
- Versions: dmail (solaris) 2.2g/makemail 2.9a
>> Include the max SIZE that will fit (e.g. I have two
>> InetAddresses in an INDEX so make them SIZE(1..120))? Or, since I
>> only expect to see IPv4 and IPv6 addresses, make them SIZE(4|16|20)?
>
>Well, two times 120 would make 240 subIDs and that is over 128.
>So that is certainly not valid. Maybe something like 32 might be a good
>size, or maybe 48, that leaves you some room for new protocols in the future.
Oops, when I did the first calculation my brain misfired and I used
256 instead of 128 as the max OID len. So, I guess the choices are:
1. Use the max that will fit (in this case, SIZE(1..57), since
msdpRequestsEntry is .1.3.6.1.3.92.1.2.1 and there are 3 other values
in the INDEX)?
2. Pick some arbitrary nice-looking number, like 32 or 48?
3. Limit the length to the values that I expect
#3 is probably out, since it doesn't allow for *any* extensibility.
#2 is prettier (and allows a little room for relocating the MIB lower
in the tree, if required) but allows for less extensibility.
#1 gives the most room. (Although requires looking ahead, since if the
DRAFT-MSDP-MIB were under mib-2 it'd have to be SIZE(1..56))
Bill