| Active Users is the most bogus of the generic performance metrics.
Users doing what and doing it how often????
100 users each executing transaction X every 10 seconds is the same as
1000 users each executing transaction X every 100 seconds. The database system
cares little what the terminal network looks like, it cares how many
transactions per second you are trying to do of what kind of transaction.
For a more real world example: A mail-order operation may take orders
through the mail or over the telephone. The ones coming in via mail are
entered into the computer by heads-down data entry clerks. The ones
coming in via phone involve lots of personal interactions and delays.
Although the transaction is identicical, a system could handle lots more
users entering information gathered over the telephone than users doing
the heads down data entry.
On another dimension, a system can handle lots more users doing
DebitCredit transactions than doing DECdecision/Access (or Teamdata) end
user information processing.
So, ORACLE or Rdb/VMS, there is no good answer for your customer unless
they clarify the question by asking asking what context it should be
evaluated in!
|
| Why doesn't your customer ask Oracle ?? You don't want to officially
(or unofficially) obligate yourself to estimating another vendor's
software performance, do you ?? If you guess wrong and find out
later that you needed more system than you thought to run Oracle,
Oracle will be sure to blame your sizing NOT their product.
-- gerry
|
| Better still, why don't you ask Oracle. They won't bite and will tell
you the same answers they will tell the customer - be they right or be
they wrong. Directing them to Oracle is giving up another little piece
of account involvement. If you remain their source of information you
at least retain the opportunity to put the Digital point of view. If
you teach them to go to Oracle, you loose that and eventually lot's
more.
-mark
|