C + + by writing HBase thrift2 efficiency is slow, has a similar Java BufferedMutator operation?

Is the efficient buffer Java interface BufferedMutator is mutate into memory each time, only to write. The value of the buffer before writing, or flush (),
C + + is to call thrift2 interface, how to achieve efficient insertion, or disable automatic flush table??

 

Void HBaseDriver: : putMultiple (string& TableName, string& RowKey, unordered_map & lt; String, string> & QualifierVal string& Family) {
CurrentPutCount +=1; 
//STD: : vector Puts; 
TPut put; 
STD: : vector CVS; 
Put the __set_row (rowKey); 
If (BOOL_TEST) {
Start=clock (); 
BOOL_TEST=false; 
} 
Unordered_map & lt; String, string> : : iterator it; 
Try {
It=qualifierVal. The begin (); 
While (it!=qualifierVal. End ()) {
TColumnValue CV; 
CV. __set_family (family); 
CV. __set_qualifier (it - & gt; First); 
CV. __set_value (it - & gt; Second); 
/* if (it - & gt; The first=="feature") {//if the key is' feature '
Assert (it - & gt; Second, length ()==BASE64_FEATURE_LEN); 
} */
CVS. Insert (CVS) end (), CV); 
The put. __set_durability (TDurability: : SKIP_WAL);//skit write ahead log for speed up 
Put the __set_columnValues (CVS); 
Puts. Insert (puts the end (), put); 
It++; 
} 
If (this - & gt; This currentPutCount % - & gt; PUT_BATCH_SIZE==0) {
Put_client_ - & gt; PutMultiple (tableName, puts); 
finish=clock(); 
Duration=(finish - start)/CLOCKS_PER_SEC; 
Printf ("=====================================% f seconds cost for hbase put. \ n ", duration). 
//long end=CURRENT_TIME (); 
Time Cost://LOG (string (" "+ to_string (end - start) +" ms ")); 
Puts the clear (); 
} 
CVS. The clear (); 
{} the catch (TException Tex) 
The LOG (string (Tex. What ())); 
} 
} 

CodePudding user response:

Can not through c + + to write? The message queue for Java to write, if the timeliness requirements can use bulkload isn't high (minutes)