Index¶

CrossDB Index is used to implement database Primary Key and accelerate query.

Index Type¶

HASH Index

• HASH index has highest O(1) performance. It's optimized a lot to achieve best performance.
• HASH index can dynamically scale with table rows number to get high performance.
• HASH index only supports exact match.

RBTREE Index

• Use flag CROSS_RBTREE
• RBTREE index is almost the same with popular RDBMS BTree, which is core index engine of RDBMS.
• RBTREE index can support exact match, range match and leftmost prefix match with multiple-column.

Guide¶

• Default index is HASH type.
• If one query matches both HASH index and RBTREE index, then Hash index is selected in most cases.
• You can create HASH index for high performance query and RBTREE index for normal performance queries.
• Index is not free, it occupies space and all inexes may be updated during INSERT/UPDATE/DELET row.
• If index is unique, please set CROSS_UNIQUE.
• For unique index, either insert replace update will check row is unique.
• You can do any query on any fields combination without index.
• Create index only when the query calls frequency is high and requires performance.

Example¶

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18

#define CHECK(ret,str)  \
    if (ret < 0) {  printf (str": %s\n", cross_errMsg(ret)); return -1; }

// Create hash index
ret = cross_dbIdxCreate (hRtTbl, "idx_nexthop", "nexthop", 0);
CHECK (ret, "Failed to create index");

// Create rbtree index
ret = cross_dbIdxCreate (hRtTbl, "idx_nexthop", "nexthop", 0);
CHECK (ret, "Failed to create index");

// Create unique hash index
ret = cross_dbIdxCreate (hRtTbl, "idx_route", "vrf,ipAddr", CROSS_UNIQUE);
CHECK (ret, "Failed to create index");

// Create unique rbtree index
ret = cross_dbIdxCreate (hRtTbl, "idx_route", "vrf,ipAddr", CROSS_UNIQUE|CROSS_RBTREE);
CHECK (ret, "Failed to create index");

Comments