Hello Martin,
Thanks for the useful tip. On cross checking the trace statements of both
inner and left join queries, found out that most of the time taken by the
left join query is spent at the following stage. This instance has happened
for at least 20 times and each is around 40-60 seconds long while executing
a single left/right join query.
<<<-------- start quote - sample trace left join query --------->>>
45983604 | X_2138=[64851] :=
bat.append(X_2128=:bat[:oid,:str][64851],X_2137=:bat[:oid,:str][64851],true);
46443428 | X_3402=[64851] :=
bat.append(X_3390=:bat[:oid,:str][64851],X_3401=:bat[:oid,:str][64851],true);
53520133 | X_1820=[64851] :=
bat.append(X_1808=:bat[:oid,:str][64851],X_1819=:bat[:oid,:str][64851],true);
64750174 | X_1864=[64851] :=
bat.append(X_1854=:bat[:oid,:str][64851],X_1863=:bat[:oid,:str][64851],true);
<<<-------- end quote - sample trace left join query --------->>>
To my understanding, I think the discrepancy is because of the different
approaches on deltas in executing both the queries. Not sure if I'm right.
I would also like to know if this large left/inner join discrepancy is a
default Monet behaviour or this is just because of some ignorant
mistake/misconfigurations of mine (considering my heavy data load).
Thanks & Regards,
Vijayakrishna.P.
Mobile : 9500402305.
