Efficiency of ADO, adostream, dbexpress and tdatasetprovider

On the intermediate layer server, package dataset data. The client receives the data using dataset. There are two formats:

(1) tdatasetprovider. Data Format

(2) The format obtained by ADO _ stream is called adostream format, which is also divided into adtg format and XML format.

The following tests the efficiency of ADO and dbexpress database engines in generating different formats.

The test code is as follows:

Function adoreaddata (adataset: tcustomadodataset; aformat: tpersistformat = pfadtg): variant; var vstream: _ stream; vfilter: variant; begin vstream: = costream. create; vfilter: = adataset. recordset. filter; try // remove filter adataset first. recordset. filter: = ''; If aformat = pfadtg then begin vstream. type _: = adtypebinary; adodb_tlb. _ recordset (adataset. recordset ). save (vstream, adpersistadtg); vstream. position: = 0; Result: = vstream. read (vstream. size); End else begin vstream. type _: = adtypetext; olevariant (adataset. recordset ). save (vstream, adpersistxml); vstream. position: = 0; Result: = vstream. readtext (vstream. size); end; finally vstream: = nil; adataset. recordset. filter: = vfilter; end;

 

Procedure tform4.button1click (Sender: tobject); const cnst_repeattimes = 10; var I: integer; A: tdatetime; vdata: variant; begin // no image // sqldataset1.commandtext: = 'select * From cssystemlogs'; // adodataset1.commandtext: = 'select * From cssystemlogs'; // sqldataset1.commandtext: = 'select * From csmachines '; adodataset1.commandtext: = 'select * From csmachines '; A: = now; for I: = 1 to cnst_repeattimes do begin sqlconnection1.open; sqldataset1.open; vdata: = datasetprovider1.data; sqldataset1.close; sqlconnection1.close; end; edit1.text: = floattostr (round (now-a) * 86400000); edit1_1.text: = inttostr (vararrayhighbound (vdata, 1); a: = now; for I: = 1 to export do begin adoconnection1.open; adodataset1.open; vdata: = allow; adodataset1.close; adoconnection1.close; end; edit2.text: = floattostr (round (now-a) * 86400000); edit2_1.text: = inttostr (vararrayhighbound (vdata, 1); a: = now; for I: = 1 to export do begin adoconnection1.open; adodataset1.open; vdata: = adoreaddata (adodataset1, pfadtg ); adodataset1.close; adoconnection1.close; end; edit3.text: = floattostr (round (now-a) * 86400000); edit3_1.text: = inttostr (vararrayhighbound (vdata, 1);: = now; for I: = 1 to cnst_repeattimes do begin adoconnection1.open; adodataset1.open; vdata: = adoreaddata (adodataset1, pfxml); adodataset1.close; adoconnection1.close; end; edit4.text: = floattostr (round (now-a) * 86400000); edit4_1.text: = inttostr (length (widestring (vdata) * 2); end;

Test results:

1. The data table contains the image field, which is packaged 10 times in a row.

Serial number	Midas dbexpress	Midas ado	ADO Stream adtg	ADO Stream XML
Packaging time (MS ):
1	468	610	594	108672
2	484	469	578	108985
3	468	454	546	109023
4	484	454	562	109434
5	485	453	562	109828
Returned data size (in bytes ):
	2603267	2603112	2603998	11186048

The fastest speed is "MIDAS Ado", and the smallest data is "MIDAS Ado ". The efficiency of "ADO Stream XML" is 241 times lower, and the data size is 4 times lower.

2. There is no image field in the data table, and it is packed 10 times in a row.

Serial number	Midas dbexpress	Midas ado	ADO Stream adtg	ADO Stream XML
Packaging time (MS ):
1	469	562	250	469
2	375	562	219	516
3	375	547	234	500
4	391	547	234	500
5	375	562	219	500
Returned data size (in bytes ):
	177732	156805	182433	534026

The fastest speed is "ADO Stream adtg", and the smallest data is "MIDAS Ado ".