[Send all replies to the list, please]
> From: [email protected] > Sent: Tuesday, August 04, 2009 1:26 AM > The explained "Nagios plugin" (Nagios is a well known Open Source > Monitoring System) is only an application for SNMP to collect performance > parameters for monitoring and historical graphing. I know what nagios is - I write nagios plugins myself. The nagios plugin spec is an interface spec only, so I know nothing about the details of your specific plugin. > Please, tried it just as > an example for any x-y-z SNMP application. This one is written in C and > uses SNMP v1,v2,v3 and with v3 we found that SNMPv3 performance is very > weak because unefficiant using Net-SNMP as the SNMP-Client compared to v2 > and v1. I've seen SNMPv3 billed as more secure, more flexible, even more useful than SNMPv1, but never faster! Even stripped bare, I expect it would be a little, albeit almost imperceptibly, slower. > We spend the time in this investigation because we need the > performance of this application for v3 simular as in v2. Easy - buy a faster computer! Got any tough problems? Yeah, it's a snarky answer, but it's the only one I've got. > The main cause for this v3 impact is unnecessary, because a very, very, > very time consuming step (simple strace) with the Net-SNMP-function > "init_snmp()" is needed to get v3/AuthPriv requests working. This step is > not necassary for v1/v2 requests, if you work with plain OID's. It has > nothing to do with Frankenstein experiments - the problem is related to > Net-SNMP and how the Net-SNMP library is structurizing functionality in its > functions ONLY. You are vivisecting the library and claiming it would be much better without this or that vital component - that is your Frankenstein experiment. Software can be made to go very fast indeed if it does nothing - behold the blur of glory that is the infinite loop! More seriously, if you want authentication, you'll need to spend time and RAM to run the hash algorithm - there is no workaround. If you want privacy, you must spend CPU cycles and RAM to run the ciphers - no workaround for that, either. You can get similar performance by using neither of those, but that pretty much defeats the point of moving to v3. It may be possible to tune up your plugin so that it runs much more efficiently than it does now, but to help with that we need details of how it works: Is it a wrapper for the command line tools? Is it long lived? Details, please. HTH, Mike ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Net-snmp-users mailing list [email protected] Please see the following page to unsubscribe or change other options: https://lists.sourceforge.net/lists/listinfo/net-snmp-users
