Step 1: Get it to work. Mostly done. Some things need doing still, but it's looking good. Step 2: Get it to work *fast* Removed big chunks of CPU time for a simple clear-text password test In some cases, hashes are faster than trees (read: constantly changing data). In others, trees are faster than hashes (data that doesn't change). And caching pointers makes the code a little more complicated, but much faster. Right now it's about 20% faster than 1.x for the same test. Alan DeKok. -- http://deployingradius.com - The web site of the book http://deployingradius.com/blog/ - The blog
On Wed 18 Apr 2007, Alan DeKok wrote:
Step 1: Get it to work. Mostly done. Some things need doing still, but it's looking good.
Step 2: Get it to work *fast* Removed big chunks of CPU time for a simple clear-text password test
In some cases, hashes are faster than trees (read: constantly changing data). In others, trees are faster than hashes (data that doesn't change). And caching pointers makes the code a little more complicated, but much faster.
Right now it's about 20% faster than 1.x for the same test.
OK. Let me know when you have got it to a point where you think its _mostly_ stable and I will update one of my servers and see which bits fall off :-) Cheers -- Peter Nixon http://www.peternixon.net/ PGP Key: http://www.peternixon.net/public.asc
Peter Nixon wrote:
OK. Let me know when you have got it to a point where you think its _mostly_ stable and I will update one of my servers and see which bits fall off :-)
Uh.. yes. I found something else yesterday in my stress tests that was "Duh... of course that's valid, I should have remembered that." So far, I've run about 500k packets through the new code in a variety of situations. It's looking a lot better than it did 2 weeks ago. I'll give you another ping when I think it's ready for production use. Alan DeKok. -- http://deployingradius.com - The web site of the book http://deployingradius.com/blog/ - The blog
participants (2)
-
Alan DeKok -
Peter Nixon