| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
 |
 | |||
| Oracle FAQ | Your Portal to the Oracle Knowledge Grid | |
|
| |||
Home -> Community -> Mailing Lists -> Oracle-L -> OT: teradata question - does this seem odd to you
Anybody out there really know teradata and is this valid?
Our team is creating an app for data that resides on our teradata rdbms. We
got this (read below) from our teradata dba as our app has the potential to
drag them down. It reads like selects can lock tables and force sequential
access. My knowledge of teradata is that rowids (primary key) get assigned
to amps. We have to use views to access the data and I guess views have to
be created in locking or non-locking mode. Must be to guarantee that the
data isn't changed. Readers blocking writers and readers blocking other
readers. But this is supposed to be a massive datawarehouse, wouldn't just
any user query have the potential to force everybody else to querue up. Just
wanted to verify if my assumptions are correct. Hard to get the info off of
the net and from our teradata people.
Thanks
Mike
"The key is that the nature of a Single-AMP operation is that it uses only ONE (or few) AMP that is 'squeezed' between the regular stream of queries. If the query is an All-AMPs, ALL the AMPs are 'locked' for the query, restricting any other query (tactical or not) from executing at the same time, hence the other queries are queuing. At the moment, there are 3 queries in the sample submitted that are Single-AMP. But since the Views that they hit have LOCKING TABLE IN ACCESS, only a single query at a time can go through. Yes, the query will be fast, but at the detriment of all other queries on the platform. If the estimated 10 to 15K queries were to run with TABLE LOCKs Views, these queries could literally overtake the machine if they were to be given the High Performance Priority Group."
 |
 |