Pass the start date + time and end date + time to the following script. It looks at the alert log file and obtains the no of archive logs that have been applied during media recovery in that time interval.

#!/usr/bin/perl -w

BEGIN { push @INC,"/home/user/TimeDate-1.16/lib/" }
use Date::Parse;
$start=str2time(@ARGV[0]);
chomp(@ARGV[1]);
$end=str2time(@ARGV[1]);

$log='path to the log file';
open(ALERTLOGFILE,$log) or die "Not able to open ".$log;
$progress=-1;
$count=0;
while(<ALERTLOGFILE>){

if(/^[A-Z][a-z]{2}\s[A-Z][a-z]{2}/){
  $time=str2time($_);

  if(($time >= $start) && ($time <= $end)){
   $progress=1;
   }else{
     $progress=-1;
   }
}
if(/^Media Recovery Log/){
  if($progress == 1){
    $count+=1;
  }
}
}
print $count ."\n";

REGEXP_REPLACE(target_name,’(([\.]|[:])[a-z,A-Z,0-9]*)|(-[0-9]{3})’,”)

It was written to transform host entries in the table of the format “aaa-bbb-001.ccc.ddd.eeeeeee.fff” to aaa-bbb . The underlying requirement was to get one distinct entry for all RAC nodes in a cluster.

1. Is index always the best way.. Should I really use the index or just do a full table scan.

2. Use bind variables everywhere. Parse once execute many times.

3. Should i partition the table ? What type of partition ? (range,list, hash etc)

4. What can be done in SQL is better done in SQL, never try with PL/SQL.

5. All “drop table” and “truncate table” commands work better after good backup of that table. (Storage is pretty cheap these days!)

6. Classify a query as I/O intensive or CPU intensive. Run a CPU intensive query in identified windows when there is high CPU idle time. (Same with I/O intensive query)

7. Never create any standalone functions/procedures. Package them based on functional dependency of each of them.

8. Table with millions of rows will not respond as fast as the table with 1 million rows.

9. Hardware scalability is not the solution. Design for Application Scalability.

10. Constraints are better enforced at the Database end than the Application end.