Kouhei Sutou
null+****@clear*****
Fri Dec 27 16:44:48 JST 2013
Kouhei Sutou 2013-12-27 16:44:48 +0900 (Fri, 27 Dec 2013) New Revision: ae5c34ffd9c9a629cf8bda57082f38380d7d9650 https://github.com/droonga/droonga.org/commit/ae5c34ffd9c9a629cf8bda57082f38380d7d9650 Message: catalog: use TOC and subsections Modified files: reference/catalog/index.md Modified: reference/catalog/index.md (+65 -46) =================================================================== --- reference/catalog/index.md 2013-12-27 16:39:30 +0900 (e89e554) +++ reference/catalog/index.md 2013-12-27 16:44:48 +0900 (5f94cd8) @@ -9,6 +9,9 @@ catalog. This documentation describes about catalog. + * TOC +{:toc} + ## How to manage So far, you need to write catalog and share it to all the nodes @@ -22,67 +25,76 @@ automatically. This section describes terms in catalog. -Catalog -: Catalog is a series of data which represents the resources in the - network. +### Catalog + +Catalog is a series of data which represents the resources in the +network. + +### Zone + +Zone is a set of farms. Each farm in a zone are expected to close to +each other, like in the same host, in the same switch, in the same +network. + +### Farm + +Farm is a Droonga Engine instance. Droonga Engine is implemented as +a [Fluentd][] plugin, fluent-plugin-droonga. + +A `fluentd` process can have multiple Droonga Engines. If you add one +or more `match` entries with type `droonga` into `fluentd.conf`, a +`fluentd` process instantiates one or more Droonga Engines. -Zone -: Zone is a set of farms. Each farm in a zone are expected to close to - each other, like in the same host, in the same switch, in the same - network. +### Dataset -Farm -: Farm is a Droonga Engine instance. Droonga Engine is implemented as - a [Fluentd][] plugin, fluent-plugin-droonga. +Dataset is a set of logical tables. A logical table must belong to +only one dataset. - A `fluentd` process can have multiple Droonga Engines. If you add - one or more `match` entries with type `droonga` into `fluentd.conf`, - a `fluentd` process instantiates one or more Droonga Engines. +Each dataset must have an unique name in the same Droonga network. -Dataset -: Dataset is a set of logical tables. A logical table must belong to - only one dataset. +### Logical table - Each dataset must have an unique name in the same Droonga network. +Logical table consists of one or more partitioned physical tables. +Logical table doesn't have physical records. It returns physical +records from physical tables. -Logical table -: Logical table consists of one or more partitioned physical tables. - Logical table doesn't have physical records. It returns physical - records from physical tables. +You can custom how to partition a logical table into one or more +physical tables. For example, you can custom partition key, the +number of partitions and so on. - You can custom how to partition a logical table into one or more - physical tables. For example, you can custom partition key, the - number of partitions and so on. +### Physical table -Physical table -: Physical table is a table in Groonga database. It stores physical - records of the table. +Physical table is a table in Groonga database. It stores physical +records of the table. -Ring -: Ring is a series of partition sets. Dataset must have one - ring. Dataset creates logical tables on the ring. +### Ring - Droonga Engine replicates each record in a logical table into one or - more partition sets. +Ring is a series of partition sets. Dataset must have one +ring. Dataset creates logical tables on the ring. -Partition set -: Partition set is a set of partitions. A partition set stores all - records in all logical tables in the same Droonga network. In other - words, dataset is partitioned in a partition set. +Droonga Engine replicates each record in a logical table into one or +more partition sets. - A partition set is a replication of other partition set. +### Partition set - Droonga Engine may support partitioning in one or more partition - sets in the future. It will be useful to use different partition - size for old data and new data. Normally, old data are smaller and - new data are bigger. It is reasonable that you use larger partition - size for bigger data. +Partition set is a set of partitions. A partition set stores all +records in all logical tables in the same Droonga network. In other +words, dataset is partitioned in a partition set. -Partition -: Partition is a Groonga database. It has zero or more physical - tables. +A partition set is a replication of other partition set. - Note: Handler plugins in Droonga Engine work on a partition. +Droonga Engine may support partitioning in one or more partition +sets in the future. It will be useful to use different partition +size for old data and new data. Normally, old data are smaller and +new data are bigger. It is reasonable that you use larger partition +size for bigger data. + +### Partition + +Partition is a Groonga database. It has zero or more physical +tables. + +Note: Handler plugins in Droonga Engine work on a partition. ## Example @@ -100,6 +112,7 @@ Here is a `catalog.json` for the above case: ~~~json { + "version": 1, "effective_date": "2013-06-05T00:05:51Z", "zones": ["localhost:23003/farm0", "localhost:23003/farm1"], "farms": { @@ -146,6 +159,12 @@ Here is a `catalog.json` for the above case: ## Parameters +Here are descriptions about parameters in `catalog.json`. + +### `version` + +It is the format version of the catalog file. + ### `effective_date` A date string representing the day the **catalog** becomes effective. -------------- next part -------------- HTML����������������������������...Download