[tahoe-dev] [tahoe-lafs] #757: there isn't a doc that says "which operations are efficient"

tahoe-lafs trac at allmydata.org
Fri Jan 29 09:10:57 UTC 2010


#757: there isn't a doc that says "which operations are efficient"
------------------------------------+---------------------------------------
 Reporter:  zooko                   |           Owner:  kevan    
     Type:  enhancement             |          Status:  assigned 
 Priority:  major                   |       Milestone:  undecided
Component:  documentation           |         Version:  1.4.1    
 Keywords:  performance docs large  |   Launchpad_bug:           
------------------------------------+---------------------------------------

Comment(by warner):

 I'd break out some operations that people might (or might not) expect to
 cost proportional to the size of the data they touch, especially when
 those operations actually cost proportional to the size of the overall
 file. So things like:

  * publishing A-byte immutable file
  * downloading B bytes (possibly all) of an A-byte immutable file
  * initial publish of an A-byte mutable file
  * modify B bytes of an existing A-byte mutable file
  * insert/delete B bytes in the middle of an existing A-byte mutable file
  * downloading B bytes (possibly all) of an A-byte mutable file
  * adding one directory entry to an existing A-entry directory
  * listing an A-entry directory

 The only surprises for our current formats is that any mutable-file
 operation costs O(A) (i.e. proportional to filesize, even if you're only
 modifying or downloading a single byte), likewise for directories. Each
 directory entry tends to use about 300-330 bytes. And of course mutable-
 file creation requires a relatively slow (1 or 2 second) RSA keypair
 generation.

-- 
Ticket URL: <http://allmydata.org/trac/tahoe/ticket/757#comment:10>
tahoe-lafs <http://allmydata.org>
secure decentralized file storage grid


More information about the tahoe-dev mailing list