registerSnapshotEngine

Syntax

registerSnapshotEngine(dbName, tableName, keyColumnName)

Arguments

dbName is a string indicating the name of a distributed database.

tableName is a string indicating the name of a table.

keyColumnName is a string indicating the name of a column.

Details

Register a snapshot engine for a distributed table. After registering the snapshot engine, we can get the most recent row of each group specified by keyColumnName with the following SQL statement:

select [HINT_SNAPSHOT] * from loadTable(dbName,tableName).

Please note that this command can be executed in single node mode. If used in a cluster, we cannot specify where conditions in the SQL statement.

Only one snapshot engine can be registered for a distributed table. After a system restart, we need to register the snapshot engine again.

Use command unregisterSnapshotEngine to unregister a snapshot engine.

Examples

Create a partitioned table:

$ db1=database("",VALUE,2018.09.01..2018.09.30)
$ db2=database("",VALUE,`AAPL`MSFT`MS`C)
$ db=database("dfs://compoDB",COMPO,[db1,db2])
$ t=table(1:0,`date`sym`val,[DATE,SYMBOL,DOUBLE])
$ pt=db.createPartitionedTable(t,`pt,`date`sym);

$ registerSnapshotEngine("dfs://compoDB","pt","sym");

$ def writeData(batch){
$     pt=loadTable("dfs://compoDB","pt")
$     tmp=table(batch:0,`date`sym`val,[DATE,SYMBOL,DOUBLE])
$     dates=take(2018.09.01..2018.09.30,batch)
$     syms=take(`AAPL`MSFT`MS`C,batch)
$     vals=rand(100.0,batch)
$     insert into tmp values(dates,syms,vals)
$     pt.append!(tmp)
$ }

$ writeData(1000);

select [HINT_SNAPSHOT] * from loadTable("dfs://compoDB","pt");

date	sym	val
2018.09.29	AAPL	24.945753
2018.09.29	MS	14.034453
2018.09.30	C	3.89175
2018.09.30	MSFT	17.720025

$ writeData(1000);
$ select [HINT_SNAPSHOT] * from loadTable("dfs://compoDB","pt");

date	sym	val
2018.09.29	AAPL	86.296883
2018.09.29	MS	48.17885
2018.09.30	C	83.7821
2018.09.30	MSFT	44.415456