feat: implement fixed bucket table distribution (#35092)

Author: localvar

docs/zh/05-basic/02-insert.md

@@ -146,11 +146,19 @@ delete from meters where ts < '2021-10-01 10:40:00.100' ;
 通过查询系统表 `INS_DISK_USAGE`，可以查看数据库的压缩率和磁盘空间，参见 [文档](../../reference/taos-sql/database/#查看数据库的磁盘空间占用)。
 
 ```sql
-select * from  INFORMATION_SCHEMA.INS_DISK_USAGE where db_name = 'db_name';
+select * from INFORMATION_SCHEMA.INS_DISK_USAGE where db_name = 'db_name';
 ```
 
 通过如下命令，可以查看表的表的压缩率及具体分布情况，参见 [文档](../../reference/taos-sql/show/#show-table-distributed)。
 
 ```sql
 show table distributed table_name;
 ```
+
+上面的 `show` 命令是按照数据块可能的数据条数动态计算的等分区间显示的分布情况，也可使用下面的语句查询系统表 `INS_TABLE_FIXED_DISTRIBUTED` 按照固定区间查看分布情况，后者更适合多表低频等单个数据块中数据条数较少的场景。
+
+注意：目前，下面的 SQL 语句仅支持使用 `SELECT *`，不支持指定具体的查询字段，且 `WHERE` 条件中也只能包含 `与` 关系的 `db_name` 和 `table_name` 条件。
+
+```sql
+select * from INFORMATION_SCHEMA.INS_TABLE_FIXED_DISTRIBUTED where db_name = 'db_name' and table_name = 'table_name';
+```

include/common/systable.h

@@ -48,6 +48,7 @@ extern "C" {
 #define TSDB_INS_TABLE_COLS                "ins_columns"
 #define TSDB_INS_TABLE_VC_COLS             "ins_virtual_child_columns"
 #define TSDB_INS_TABLE_TABLE_DISTRIBUTED   "ins_table_distributed"
+#define TSDB_INS_TABLE_TABLE_FIXED_DISTRIBUTED "ins_table_fixed_distributed"
 #define TSDB_INS_TABLE_USERS               "ins_users"
 #define TSDB_INS_TABLE_USERS_FULL          "ins_users_full"
 #define TSDB_INS_TABLE_TOKENS              "ins_tokens"

include/common/tcommon.h

@@ -307,6 +307,7 @@ typedef struct STableBlockDistInfo {
   uint32_t numOfSttRows;
   uint32_t numOfVgroups;
   int32_t  blockRowsHisto[20];
+  int32_t  blockRowsHistoFixed[8];  // buckets: ≤64, ≤128, ≤256, ≤512, ≤1024, ≤2048, ≤4096, >4096
 } STableBlockDistInfo;
 
 int32_t tSerializeBlockDistInfo(void* buf, int32_t bufLen, const STableBlockDistInfo* pInfo);

source/common/src/systable.c

@@ -290,6 +290,34 @@ static const SSysDbTableSchema userTblDistSchema[] = {
     {.name = "seek_header_time", .bytes = 4, .type = TSDB_DATA_TYPE_INT, .sysInfo = true},
 };
 
+static const SSysDbTableSchema userTblFixedDistSchema[] = {
+    {.name = "db_name", .bytes = SYSTABLE_SCH_DB_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
+    {.name = "table_name", .bytes = SYSTABLE_SCH_TABLE_NAME_LEN, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = false},
+    {.name = "vgroup_id", .bytes = 4, .type = TSDB_DATA_TYPE_INT, .sysInfo = true},
+    {.name = "total_blocks", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "total_size", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "average_size", .bytes = 8, .type = TSDB_DATA_TYPE_DOUBLE, .sysInfo = false},
+    {.name = "compression_ratio", .bytes = 8, .type = TSDB_DATA_TYPE_DOUBLE, .sysInfo = false},
+    {.name = "block_rows", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "min_rows", .bytes = 4, .type = TSDB_DATA_TYPE_INT, .sysInfo = false},
+    {.name = "max_rows", .bytes = 4, .type = TSDB_DATA_TYPE_INT, .sysInfo = false},
+    {.name = "avg_rows", .bytes = 8, .type = TSDB_DATA_TYPE_DOUBLE, .sysInfo = false},
+    {.name = "in_mem_rows", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "stt_rows", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "total_tables", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "total_filesets", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "total_vgroups", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "row_size", .bytes = 4, .type = TSDB_DATA_TYPE_INT, .sysInfo = true},
+    {.name = "block_dist_64", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "block_dist_128", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "block_dist_256", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "block_dist_512", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "block_dist_1024", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "block_dist_2048", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "block_dist_4096", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+    {.name = "block_dist_other", .bytes = 8, .type = TSDB_DATA_TYPE_BIGINT, .sysInfo = false},
+};
+
 static const SSysDbTableSchema userUsersSchema[] = {
     {.name = "name", .bytes = TSDB_USER_LEN + VARSTR_HEADER_SIZE, .type = TSDB_DATA_TYPE_VARCHAR, .sysInfo = true},
     {.name = "super", .bytes = 1, .type = TSDB_DATA_TYPE_TINYINT, .sysInfo = true},
@@ -777,6 +805,7 @@ static const SSysTableMeta infosMeta[] = {
     {TSDB_INS_TABLE_TAGS, userTagsSchema, tListLen(userTagsSchema), false, PRIV_CAT_BASIC},
     {TSDB_INS_TABLE_COLS, userColsSchema, tListLen(userColsSchema), false, PRIV_CAT_BASIC},
     {TSDB_INS_TABLE_VC_COLS, userVctbColsSchema, tListLen(userVctbColsSchema), false, PRIV_CAT_BASIC},
+    {TSDB_INS_TABLE_TABLE_FIXED_DISTRIBUTED, userTblFixedDistSchema, tListLen(userTblFixedDistSchema), false, PRIV_CAT_BASIC},
     // {TSDB_INS_TABLE_TABLE_DISTRIBUTED, userTblDistSchema, tListLen(userTblDistSchema), PRIV_CAT_PRIVILEGED},
     {TSDB_INS_TABLE_USERS, userUsersSchema, tListLen(userUsersSchema), true, PRIV_CAT_SECURITY},
     {TSDB_INS_TABLE_USERS_FULL, userUsersFullSchema, tListLen(userUsersFullSchema), true, PRIV_CAT_SECURITY},

source/dnode/vnode/src/tsdb/tsdbRead2.c

@@ -6916,6 +6916,16 @@ int32_t tsdbGetFileBlocksDistInfo2(STsdbReader* pReader, STableBlockDistInfo* pT
       int32_t bucketIndex = getBucketIndex(pTableBlockInfo->defMinRows, bucketRange, numOfRows, numOfBuckets);
       pTableBlockInfo->blockRowsHisto[bucketIndex]++;
 
+      // fixed-size row-count histogram
+      if (numOfRows <= 64)        pTableBlockInfo->blockRowsHistoFixed[0]++;
+      else if (numOfRows <= 128)   pTableBlockInfo->blockRowsHistoFixed[1]++;
+      else if (numOfRows <= 256)   pTableBlockInfo->blockRowsHistoFixed[2]++;
+      else if (numOfRows <= 512)   pTableBlockInfo->blockRowsHistoFixed[3]++;
+      else if (numOfRows <= 1024)  pTableBlockInfo->blockRowsHistoFixed[4]++;
+      else if (numOfRows <= 2048)  pTableBlockInfo->blockRowsHistoFixed[5]++;
+      else if (numOfRows <= 4096)  pTableBlockInfo->blockRowsHistoFixed[6]++;
+      else                         pTableBlockInfo->blockRowsHistoFixed[7]++;
+
       hasNext = blockIteratorNext(&pStatus->blockIter);
     } else {
       code = initForFirstBlockInFile(pReader, pBlockIter);