[pytables-users] Performance of table.flush()

[r***@gmail.com]

Hello,

I have in my project a table with a completely sorted index, which is used
by a web application.
In one type of http request, one row is added to the table and the table is
immediately flushed.
While profiling my application, I have noticed that the time spent in the
method table.flush() increases with the size of the table.
Most of the time of the flush method seems to be spent a method called
_read().
Is this expected?

On my old laptop, I could see in my profiling data that when the table has
about 850 rows, the flush method is executed in about 180ms.
However after a while, when the table has about 1450 rows, the execution
time has increased to about 300ms!
While these figures will likely be smaller on the server, I am still
worried of the time it will take to process the request when the table
reaches an expected size of ~100,000 rows...

Could there be anything I may have done wrongly that would show this
behavior? Or maybe there is a setting I can tune to improve the figures?

Thanks for your help,
Romain