: Scalable HFSC
Subsystems (2)
cover : Scalable HFSC
These 4 patches for 2.4/2.6 make HFSC more scalable by using rbtree's instead of sorted lists and an O(1) algorithm for adjusting childrens vt-offset. With these patches HFSC with 1024 active sibling classes performs only slightly worse (CPU-wise) than HTB. Each change has been well tested in tcsim, so I think nothing should break.
Regards
Patrick
Quoted 5 lines · On Sat, 14 Aug 2004 22:00:09 +0200 Patrick McHardy <[email protected]> wrote:
These 4 patches for 2.4/2.6 make HFSC more scalable by using rbtree's instead of sorted lists and an O(1) algorithm for adjusting childrens vt-offset. With these patches HFSC with 1024 active sibling classes performs only slightly worse (CPU-wise) than HTB. Each change has been well tested in tcsim, so I think nothing should break.
All applied, thanks Patrick.