Hi everyone,
my first question concerns the the size of a table with a bytea row. In
the documentation it says something like 4 Bytes + 1 Byte for each
escaped octet sequence per row. For example an insertion into a table
storing just one column with bytea data looks like this:
insert into test values (''\\003''); ---- this allows 185 inserts per
page until a new one is needed
insert into test values (''\\003\\123\\123\\111''); ---- this yields
exactly the same nr. of rows per page!
insert into test values (''\\003\\123\\123\\111\\001''); ---- this one
finally needs more pages!
How can that be, if the system internally allocates ONE BYTE per
octet...
Does it acually take 4 Bytes?
My second question is more generall:
My dbms (acutally it is supposed to be a decision support system, so I
do not really need rollbacks and transactions etc. --- can those
features be turned off to enhance performance?) needs to store 1
Billion rows in a single table (I know that I could use horizontal
partitioning) and I wonder if postgres is powerfull enough to handle
such large tables? If anyone has experience with tables this size and
could give me a hint which system to use (DB2, Sybase, Oracle,
Informix, Postgres) this would be great...
THX
tschak