Multiple Insert into Postgresql using Criteria is very slow - performance

Question

I am reading a big text file using Java. The file has 5.000.000 of rows and each one have 3 columns. The file size is 350 MB.
For each row, I read it, I create an object using Criteria on Maven and I store it into a Postgresql database with a session.saveOrUpdate(object) command.
In the database I have a table with a serial ID and three attributes where I store the three columns of the file.
At the beginning, the process run "fast" (35.000 registers in 30 min) but every time is slower and the time to finish grow exponentially. How can I improve the process??
I have tried to split the big file into several smaller files but it is almost slower.
Many thanks in advance!
PD: The code
public void process(){
File archivo = null;
FileReader fr = null;
BufferedReader br = null;
String linea;
String [] columna;
try{
archivo = new File ("/home/josealopez/Escritorio/file.txt");
fr = new FileReader (archivo);
br = new BufferedReader(fr);
while((linea=br.readLine())!=null){
columna = linea.split(";");
saveIntoBBDD(columna[0],columna[1],columna[2]);
}
}
catch(Exception e){
e.printStackTrace();
}
finally{
try{
if( null != fr ){
fr.close();
}
}
catch (Exception e2){
e2.printStackTrace();
}
}
}
#CommitAfter
public void saveIntoBBDD(String lon, String lat, String met){
Object b = new Object();
b.setLon(Double.parseDouble(lon));
b.setLat(Double.parseDouble(lat));
b.setMeters(Double.parseDouble(met));
session.saveOrUpdate(b);
}

Answer 1

You should focus on running this as a bulk process and line-based processing is your issue here. PostgreSQL has built-in command for bulk file loading, named COPY, that can deal with Comma Separated Files and Tab Separated Files. Of course, delimiter, quotations chars and many other settings are customizable.
Please, check official PostgreSQL documentation on DB population and also details of the COPY command.
In this answer I provided a small example of how I do similar kind of things.