← Back to branch summary

~bzr-pqm/bzr/bzr.dev

« back to all changes in this revision

Viewing changes to doc/developers/groupcompress-design.txt

  • Committer: Canonical.com Patch Queue Manager
  • Date: 2010-08-24 23:20:14 UTC
  • mfrom: (5365.5.29 2.3-btree-chk-leaf)
  • Revision ID: pqm@pqm.ubuntu.com-20100824232014-nu9owzel2zym2jk2
(jam) Use a custom C type for CHK index entries, saves memory

Show diffs side-by-side

added added

removed removed

Lines of Context:
doc/developers/groupcompress-design.txt
29
29
 
30
30
Reasonable sizes 'amount read' from remote machines to reconstruct an arbitrary
31
31
text: Reading 5MB for a 100K plain text is not a good trade off. Reading (say)
32
 
500K is probably acceptable. Reading ~100K is ideal. However, it's likely that
 
32
500K is probably acceptable. Reading ~100K is ideal. However, its likely that
33
33
some texts (e.g NEWS versions) can be stored for nearly-no space at all if we
34
34
are willing to have unbounded IO. Profiling to set a good heuristic will be
35
35
important. Also allowing users to choose to optimise for a server environment