Than it's PHDBase problem. It's not surprising because PHDBase hasn't been supported for years. I think it's time for you to look for some alternative solutions.

>I tried with those parameters in effect. I got the same result. Once it reached 600,000 records, I have this memory problem. The only way I could see to index that one million record table would be to use PHDBase per record indexation which will take a little bit longer.