Teradata does have it strengths but these strengths are also its weakness.
Its strength is that it's software is tied in with NCR hardware. Its
weakness is that it's tied in with NCR hardware. The data is actually stored
with a specific CPU, they call it AMP. It's like it's partitioned by cpu.
So it can be really fast and powerful on all AMP queries but it can get
really bogged down quickly on single keyed reads. You have to make sure that
the queries are tuned as single amp. To do this you need join indexes across
all of your tables. Before we did this are single keyed reads bringing back
one row was locking up the whole system. So reads can be locking and queries
will queue up behind each other.
I don't know much about its programming but it hasn't been around that long.
Oracle being a much bigger company is way more flexible in the manipulation
of its data. I don't know if you can do in-line sub queries yet in teradata.
Try storing xml. There is no way you can do something like stragg from
asktom. Our company actually put an oracle database in front of its teradata
database. The oracle database does the tricky grunt work so teradata can do
a fast etl.
But really, I'm sure there are great teradata data warehouses as is there
are great oracle datawarehouses. But you are tied with NCR using teradata.
Look at all the combinations you can do with oracle.
--
http://www.freelists.org/webpage/oracle-l
Received on Fri Oct 13 2006 - 07:56:26 CDT
