Shell Programming and Scripting

Suppose that I am performing some operation on an sql database. Lets say process of Searching and then if a value is found, updating it... Now, when I have some millions of records on which the operation has to be performed... Does it help to spawn multiple processes each executing the same script(assuming that I coded the script in php)? Will that make a difference in speed of execution?

Quite the reverse, a database will serialise access and as you access records will add them to the list of records touched by the current transaction.

Rather than doing searches and updates, you should let the database do the work with an "update .... where ...." statement. Let the database do all the searches internally and update internally.

Oops... Sorry... I meant to say that... What I was asking is that if I spawn new processes that execute this script, will the job get done faster? The computer I am referring to is a full fledged server which only I am using currently...

i would not expect so if the transactions touch the same records.

Oh... For example consider the following:

Table 1:
121::::ABCDE::::::0.2
230:::::ASDS:::::::0.6
xxx::::::jaksd:::::::0.7
xxx::::::dasdd:::::::0.3
xxx::::::asaaa:::::::0.1
xxx::::::ddddd:::::::0.5

Table 2:
121::::ABCDE
230:::::ASDS
343::::::jaksd
545::::::dasdd
656::::::asaaa
344::::::ddddd


What my script is doing is takes the second column of Table 1 and searches for a key in the table 2 and then when it finds a relvant key it updates table 1 with that value... So basically, I'm trying to merge the two tables with the only difference is that the number of rows in both the tables are not the same due to some corruption in table 2... What is the fastest way to achieve this when I have to perform millions of such operations?

Can you provide some more information ?

Which database are you using ?
and what about the status of the columns being used for searching and updating.

Are the fields indexed, if so obviously search is going to be faster.

I have tried this in Informix and not in Oracle.
When you have to do an update on a table : column based on a value from table2 : column2, there is an option in Informix to do parallelized search where you have to specify only the number of threads in which the search has to run.

And the approach is much faster on the indexed columns.

for n number of rows from 1 .. n
Based on ' m ' threads number of records would be split as
1 .. k, k+1 .. k<n> , and k <m> .. n

and the results are internally manipulated.

Would this help you ?

Thank you so much for your concern... I'm using a MySQL database... As for the status of the columns, I not quite sure what you meant, but they are of the following types:

Table_1
Column_1: int
Column_2: varchar(18)
Column_3: decimal

Table_2
Column_1: int
Column_2: varchar(18)

I had a couple of text files from which i used a php script + regular expressions to grab the text I want and then inserted them into the database's table_1 and table_2. Hope I was able to convey my requirement...