Documentation/technical/multi-pack-index.txt

   1 Multi-Pack-Index (MIDX) Design Notes
   2 ====================================
   3
   4 The Git object directory contains a 'pack' directory containing
   5 packfiles (with suffix ".pack") and pack-indexes (with suffix
   6 ".idx"). The pack-indexes provide a way to lookup objects and
   7 navigate to their offset within the pack, but these must come
   8 in pairs with the packfiles. This pairing depends on the file
   9 names, as the pack-index differs only in suffix with its pack-
  10 file. While the pack-indexes provide fast lookup per packfile,
  11 this performance degrades as the number of packfiles increases,
  12 because abbreviations need to inspect every packfile and we are
  13 more likely to have a miss on our most-recently-used packfile.
  14 For some large repositories, repacking into a single packfile
  15 is not feasible due to storage space or excessive repack times.
  16
  17 The multi-pack-index (MIDX for short) stores a list of objects
  18 and their offsets into multiple packfiles. It contains:
  19
  20 * A list of packfile names.
  21 * A sorted list of object IDs.
  22 * A list of metadata for the ith object ID including:
  23 ** A value j referring to the jth packfile.
  24 ** An offset within the jth packfile for the object.
  25 * If large offsets are required, we use another list of large
  26   offsets similar to version 2 pack-indexes.
  27
  28 Thus, we can provide O(log N) lookup time for any number
  29 of packfiles.
  30
  31 Design Details
  32 --------------
  33
  34 - The MIDX is stored in a file named 'multi-pack-index' in the
  35   .git/objects/pack directory. This could be stored in the pack
  36   directory of an alternate. It refers only to packfiles in that
  37   same directory.
  38
  39 - The core.multiPackIndex config setting must be on (which is the
  40   default) to consume MIDX files.  Setting it to `false` prevents
  41   Git from reading a MIDX file, even if one exists.
  42
  43 - The file format includes parameters for the object ID hash
  44   function, so a future change of hash algorithm does not require
  45   a change in format.
  46
  47 - The MIDX keeps only one record per object ID. If an object appears
  48   in multiple packfiles, then the MIDX selects the copy in the
  49   preferred packfile, otherwise selecting from the most-recently
  50   modified packfile.
  51
  52 - If there exist packfiles in the pack directory not registered in
  53   the MIDX, then those packfiles are loaded into the `packed_git`
  54   list and `packed_git_mru` cache.
  55
  56 - The pack-indexes (.idx files) remain in the pack directory so we
  57   can delete the MIDX file, set core.midx to false, or downgrade
  58   without any loss of information.
  59
  60 - The MIDX file format uses a chunk-based approach (similar to the
  61   commit-graph file) that allows optional data to be added.
  62
  63 Future Work
  64 -----------
  65
  66 - The multi-pack-index allows many packfiles, especially in a context
  67   where repacking is expensive (such as a very large repo), or
  68   unexpected maintenance time is unacceptable (such as a high-demand
  69   build machine). However, the multi-pack-index needs to be rewritten
  70   in full every time. We can extend the format to be incremental, so
  71   writes are fast. By storing a small "tip" multi-pack-index that
  72   points to large "base" MIDX files, we can keep writes fast while
  73   still reducing the number of binary searches required for object
  74   lookups.
  75
  76 - If the multi-pack-index is extended to store a "stable object order"
  77   (a function Order(hash) = integer that is constant for a given hash,
  78   even as the multi-pack-index is updated) then MIDX bitmaps could be
  79   updated independently of the MIDX.
  80
  81 - Packfiles can be marked as "special" using empty files that share
  82   the initial name but replace ".pack" with ".keep" or ".promisor".
  83   We can add an optional chunk of data to the multi-pack-index that
  84   records flags of information about the packfiles. This allows new
  85   states, such as 'repacked' or 'redeltified', that can help with
  86   pack maintenance in a multi-pack environment. It may also be
  87   helpful to organize packfiles by object type (commit, tree, blob,
  88   etc.) and use this metadata to help that maintenance.
  89
  90 - The partial clone feature records special "promisor" packs that
  91   may point to objects that are not stored locally, but available
  92   on request to a server. The multi-pack-index does not currently
  93   track these promisor packs.
  94
  95 Related Links
  96 -------------
  97 [0] https://bugs.chromium.org/p/git/issues/detail?id=6
  98     Chromium work item for: Multi-Pack Index (MIDX)
  99
 100 [1] https://lore.kernel.org/git/20180107181459.222909-1-dstolee@microsoft.com/
 101     An earlier RFC for the multi-pack-index feature
 102
 103 [2] https://lore.kernel.org/git/alpine.DEB.2.20.1803091557510.23109@alexmv-linux/
 104     Git Merge 2018 Contributor's summit notes (includes discussion of MIDX)