X-Git-Url: https://gerrit.opnfv.org/gerrit/gitweb?a=blobdiff_plain;f=src%2Fceph%2Fdoc%2Fdev%2Fosd_internals%2Ferasure_coding.rst;fp=src%2Fceph%2Fdoc%2Fdev%2Fosd_internals%2Ferasure_coding.rst;h=7263cc35b67152194dcde112d1143f3467d1b20c;hb=812ff6ca9fcd3e629e49d4328905f33eee8ca3f5;hp=0000000000000000000000000000000000000000;hpb=15280273faafb77777eab341909a3f495cf248d9;p=stor4nfv.git diff --git a/src/ceph/doc/dev/osd_internals/erasure_coding.rst b/src/ceph/doc/dev/osd_internals/erasure_coding.rst new file mode 100644 index 0000000..7263cc3 --- /dev/null +++ b/src/ceph/doc/dev/osd_internals/erasure_coding.rst @@ -0,0 +1,82 @@ +============================== +Erasure Coded Placement Groups +============================== + +Glossary +-------- + +*chunk* + when the encoding function is called, it returns chunks of the same + size. Data chunks which can be concatenated to reconstruct the original + object and coding chunks which can be used to rebuild a lost chunk. + +*chunk rank* + the index of a chunk when returned by the encoding function. The + rank of the first chunk is 0, the rank of the second chunk is 1 + etc. + +*stripe* + when an object is too large to be encoded with a single call, + each set of chunks created by a call to the encoding function is + called a stripe. + +*shard|strip* + an ordered sequence of chunks of the same rank from the same + object. For a given placement group, each OSD contains shards of + the same rank. When dealing with objects that are encoded with a + single operation, *chunk* is sometime used instead of *shard* + because the shard is made of a single chunk. The *chunks* in a + *shard* are ordered according to the rank of the stripe they belong + to. + +*K* + the number of data *chunks*, i.e. the number of *chunks* in which the + original object is divided. For instance if *K* = 2 a 10KB object + will be divided into *K* objects of 5KB each. + +*M* + the number of coding *chunks*, i.e. the number of additional *chunks* + computed by the encoding functions. If there are 2 coding *chunks*, + it means 2 OSDs can be out without losing data. + +*N* + the number of data *chunks* plus the number of coding *chunks*, + i.e. *K+M*. + +*rate* + the proportion of the *chunks* that contains useful information, i.e. *K/N*. + For instance, for *K* = 9 and *M* = 3 (i.e. *K+M* = *N* = 12) the rate is + *K* = 9 / *N* = 12 = 0.75, i.e. 75% of the chunks contain useful information. + +The definitions are illustrated as follows (PG stands for placement group): +:: + + OSD 40 OSD 33 + +-------------------------+ +-------------------------+ + | shard 0 - PG 10 | | shard 1 - PG 10 | + |+------ object O -------+| |+------ object O -------+| + ||+---------------------+|| ||+---------------------+|| + stripe||| chunk 0 ||| ||| chunk 1 ||| ... + 0 ||| stripe 0 ||| ||| stripe 0 ||| + ||+---------------------+|| ||+---------------------+|| + ||+---------------------+|| ||+---------------------+|| + stripe||| chunk 0 ||| ||| chunk 1 ||| ... + 1 ||| stripe 1 ||| ||| stripe 1 ||| + ||+---------------------+|| ||+---------------------+|| + ||+---------------------+|| ||+---------------------+|| + stripe||| chunk 0 ||| ||| chunk 1 ||| ... + 2 ||| stripe 2 ||| ||| stripe 2 ||| + ||+---------------------+|| ||+---------------------+|| + |+-----------------------+| |+-----------------------+| + | ... | | ... | + +-------------------------+ +-------------------------+ + +Table of content +---------------- + +.. toctree:: + :maxdepth: 1 + + Developer notes + Jerasure plugin + High level design document