Iain,
There are a couple of reasons you might not want to use internal
storage. The first is that its wiped out when Kannel restarts, so if
there is a problem, or you want to add a new SMSC into the
configuration any outstanding DLRs are lost.
The second would be if you had 2 Kannel boxes running and there was
the possibility of a DLR coming back into the other Kannel box from
the one that sent it. You would need to have a single database that
both boxes shared.
Its quite possible that people have phones switched off in meetings,
on flights, go on an underground train etc. So whilst most messages
are delivered quickly its possible with a large volume of messages to
see outstanding DLRs for quite some time.
Regards
Ben
On 13 Mar 2007, at 11:54, Iain Dooley wrote:
Thats correct. Kannel will always call the URL you provide to send
you a DLR. The storage method is irrelevant in that respect.
okay, so i'm sorry if this sounds really bleeding obvious, and if
it does it's because i've never dealt with astronomical volumes of
messages but:
in a machine with (say) 1gb of RAM, and a dlr-url, and the fact
that most smsc's don't tell you anything except the 'final' state
of a message (and the fact that your messages generally get
delivered pretty quickly as long as your smsc isn't complete
shite)....
why does anyone use mysql or postgresql dlr-storage? i mean, does
anyone, at any time, have more than, say, 100mb of dlr-storage?
mysql/postres, they can handle TERABYTES of data, but, couldn't you
handle a dlr-queue for a _telco_ with a peasant's budget worth of ram?
cheers
iain
