Robert Story wrote:
On Thu, 22 Dec 2005 09:40:45 -0500 Glenn wrote:
GM> I don't know at this time if the module that
GM> tracks reservations will be our own, or one of the stock modules from
GM> openSER. If it's our own, we can likely keep the list of registrations
GM> in a sorted order; if it's not, we can't depend on the registrations
GM> being properly sorted.
GM>
GM> Any suggestions on what approach I should take?
It really depends on whether or not your data is sorted, if it's sorted
properly, and if the fifo allows access to individual entries or just the
whole list.
I've been doing some digging and it turns out the usrloc module tracks
most of the information I need. I can write custom commands to get the
information in the format I need. Unfortunately, the current way the
module stores the data is in an unsorted linked list. I've got a custom
command that lists just the address of record (my index value, its a
(usually) short value like [EMAIL PROTECTED]), but not in
sorted order. Looks like I'll have to extend my command. On that note,
I'm not sure I know what you meant by "sorted correctly." I'm happy to
RTFM, if someone will point me at the FM. :-)
Once I have the indexes in sorted order, is there anything I need to do
to tell the cache that it's already sorted?
They do have a command that allows you to get an individual entry, but
it's not overly useful as it's missing just about everything we really
need, so I'll be writing a custom command to make up for it.
GM> While the FAQ entries
GM> for mib2c are helpful, and point me to the iterate or mfd config files
GM> to deal with tables that have external data, I was hoping someone could
GM> point out in advance whether or not the standard approaches will scale
GM> to the 10,000+ row case.
Is your system severely memory limited? If so, caching all the data could be an
issue and iterate is the way to go. But it will have *terrible* performance.
If memory is not so limited, MFD will allow you to work with indexes only,
until the rest of the data is needed.
This is a server box, not an embedded device, so while memory isn't
free, it's not nearly as scarce a resource as it might be. I'm going to
go ahead and work with the MFD approach (populate the index before the
data) for now. Performance is always a concern for us, so hopefully the
memory trade off will be acceptable.
Thanks,
Glenn McAllister
SOMA Networks, Inc.
-------------------------------------------------------
This SF.net email is sponsored by: Splunk Inc. Do you grep through log files
for problems? Stop! Download the new AJAX search engine that makes
searching your log files as easy as surfing the web. DOWNLOAD SPLUNK!
http://ads.osdn.com/?ad_id=7637&alloc_id=16865&op=click
_______________________________________________
Net-snmp-coders mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/net-snmp-coders
