I am new to packetfence but I have been developing network management
applications in perl for the last 8 years. In these applications I have had
to deal with scale in excess of 300k nodes (switches, routers, aps, etc) so
performance is always on my mind.

Here are a few initial questions I have come up with.


   1. Why are we using Net::SNMP and not the net-snmp SNMP.pm module.
   SNMP.pm is XS and in my applications has giving me a 300% increase in
   performance vs Net::SNMP. Net::SNMP is definitely easier to use and has
   better documentation, but SNMP.pm is extremely stable and uses hardly any
   cpu, and is just faster. A helper module would make device modules easier
   to develop.
   2. Have we ever considered multiplexing snmp requests (this is done very
   efficiently in SNMP.pm)
   3. Why is one of the goals to move away from snmptrapd? You can do
   direct perl integration. In some of the other applications I have written,
   I take the trap messages from snmptrapd, build them into a hash, convert
   them into json using JSON::XS and store them in a redis array. I then have
   another application processing those traps. It takes the work off of
   snmptrapd but utilizes the fact that it is written in c and can
   encode/decode packets faster and more efficiently.
   4. Have we considered using Redis instead of memcached? It is faster,
   supports more datatypes, can be persistent, and supports replication.
   5. Any consideration on moving from mysql to postgresql or supporting
   both?
   6. Can someone please outline the challenges in implementing and
   active/active setup? From what I can think of we would  need an external
   loadbalancer, external database (could replicate as well), and some way to
   sync configuration.


Thank you for any information you are able to provide. I am very excited
about contributing to this project.

Mike
------------------------------------------------------------------------------
HPCC Systems Open Source Big Data Platform from LexisNexis Risk Solutions
Find What Matters Most in Your Big Data with HPCC Systems
Open Source. Fast. Scalable. Simple. Ideal for Dirty Data.
Leverages Graph Analysis for Fast Processing & Easy Data Exploration
http://p.sf.net/sfu/hpccsystems
_______________________________________________
PacketFence-devel mailing list
PacketFence-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/packetfence-devel

Reply via email to