1   

Directory Layout •  Separate files per database •  Aggressive preallocation •  Files contain one or more extents 2   -rw------- 1 ben ben 64M May 1 19:14 test.0! -rw------- 1 ben ben 128M May 1 19:14 test.1! -rw------- 1 ben ben 256M May 1 18:25 test.2! -rw------- 1 ben ben 512M May 1 19:14 test.3! -rw------- 1 ben ben 1.0G May 1 19:14 test.4! -rw------- 1 ben ben 2.0G May 1 18:58 test.5! -rw------- 1 ben ben 16M May 1 19:14 test.ns! 

Memory Mapping STACK! …! LIBS! …! test.ns! test.0! test.1! …! ! …! HEAP! MONGOD! NULL! 0x7fffffffffff   0x0   {  …  }   Disk   Document   Process  Virtual  Memory   

Data Structures •  DiskLoc •  Stores file number and offset of data on disk •  Record *r = mmap base + DiskLoc.offset! •  Max offset is 2^31 (2GB)! •  NamespaceDetails •  Stores collection metadata! •  Extent! •  Stores contiguous blocks within a namespace •  Max extent size is 2GB   •  Record! •  Holds a BSON document or B-tree bucket •  DeletedRecord overwrites a Record! •  Includes Padding 

Namespace Details •  Holds metadata about a collection or index •  Stored in 1KB buckets in .ns file •  .ns file fixed size of 16MB •  Maintains document count •  Contains heads of linked lists firstExtent   lastExtent   _indexes[]   stats   freeList[]   NamespaceDetails   

Extent Structure Extent   length   xNext   xPrev   firstRecord   lastRecord   Extent   length   xNext   xPrev   firstRecord   lastRecord   

Extents >  db.foo.validate(  {  full  :  true  }  ).extents.forEach(                      function(z){  print(  z.loc  +  "\t\t"  +  z.size  );  }  )   0:3000    20480   0:12000    81920   0:26000    327680   0:76000    1310720   0:1da000  5242880   0:76a000  6291456   0:d6a000  7553024   0:16de000  9064448   0:1f83000  10878976   0:29e3000  13058048   1:2000    15671296   1:ef4000  18808832   1:29e4000  22573056   

Index Extents >  db.system.namespaces.find()   {  "name"  :  "test.foo"  }   {  "name"  :  "test.system.indexes"  }   {  "name"  :  "test.foo.$_id_"  }     >  db["foo.$_id_"].validate(  {  full  :  true  }  ).extents.forEach(                      function(z){  print(  z.loc  +  "\t\t"  +  z.size  );  }  )   0:9000    36864   0:1b6000  147456   0:6da000  589824   0:149e000  2359296   1:20e4000  9437184   

Extents and Records Extent   length   xNext   xPrev   firstRecord   lastRecord   Data  Record   length   rNext   rPrev   Document   {        _id:  “foo”,      ...     }   

Extents and Records Extent   length   xNext   xPrev   firstRecord   lastRecord   Data  Record   length   rNext   rPrev   Document   {        _id:  “foo”,      ...     }   

Extents and Records Extent   length   xNext   xPrev   firstRecord   lastRecord   Data  Record   length   rNext   rPrev   Document   {        _id:  “foo”,      ...     }   Data  Record   length   rNext   rPrev   Document   {        _id:  “foo”,      ...     }   

BSON Format {  hello:  “world”  }   \x16\x00\x00\x00 \x02hello\x00 ! \x06\x00\x00\x00 world\x00\x00! Doc  Length   Value  Type   Value  Length   

Index Extents Extent   length   xNext   xPrev   firstRecord   lastRecord   Index  Record   Bucket   parent   numKeys     length   rNext   rPrev   Index  Record   Bucket   parent   numKeys   K   length   rNext   rPrev         {  Document  }   

Index Extents Extent   length   xNext   xPrev   firstRecord   lastRecord   Index  Record   Bucket   parent   numKeys     length   rNext   rPrev   Index  Record   Bucket   parent   numKeys   K   length   rNext   rPrev         {  Document  }   4   9   1   3   5   6   8   A   B   

Journaling •  Write ahead logging •  Operations written to journal before memory mapped regions •  Private view •  Shared view •  Once journal written, data safe unless hardware problem •  By default, journal flushed every 100ms, 100mb of writes, or on write concern of j=true •  User configurable with --journalCommitInterval 

•  Section  contains  single  group  commit   •  Applied  all-­‐or-­‐nothing   Journal Format JHeader   JSectHeader  [LSN  3]   DurOp   DurOp   DurOp   JSectFooter   JSectHeader  [LSN  7]   DurOp   DurOp   DurOp   JSectFooter   …   Op_DbContext   length   offset   fileNo   data[length]   length   offset   fileNo   data[length]   length   offset   fileNo   data[length]   Write  Operation   Set  database  context  for   subsequent  operations   

Journal Performance •  On 99.9% read systems, no impact •  Write performance degraded 5-30% when journal on same drive •  Separate drive as low as 3% 

Journal Admin •  Journal stored in /dbpath/journal folder •  If faster, three 1gb files may be preallocated •  Can symlink to a different spindle •  --journalCommitInterval* (2ms - 300ms) •  When to journal •  Single node: required for data integrity •  Replica set: at least 1 node •  All nodes: removes possible need to resync 

Fragmentation •  Files may become fragmented over time if documents change size •  Free lists also contribute to fragmentation •  2.0 reduced scanning to reasonable amounts •  2.2 will change allocation strategy •  Need to re-write free list to do online compaction 

Compaction •  1.8 and previous: repairDatabase •  2.0+ : compact command •  Currently resets paddingFactor, but can be changed. •  Index (re)generation is now concurrent, so compaction can be N times faster •  Generally causes some extra allocation •  Does not delete or truncate files 

Planned Changes •  Split data and indexes into different files •  Indexes could by symlinked to a different drive (SSD) •  Improved allocation strategy 

Download  MongoDB   http://www.mongodb.org/downloads     Ben  Becker   [email protected]