I've a problem with selecting data from 2 big mysql tables, you can find below the queries & tables details
Query 1:
It takes : 394 seconds.

SELECT SQL_CALC_FOUND_ROWS `table1`.`url`, `table1`.`Title`, `table1`.`Description`, `table2`.`col16`, `table2`.`col17`, `table2`.`col18`, `table2`.`col19`
FROM `table1`, `table2` 
WHERE `table1`.`url` = `table2`.`url` 
AND `topic` = 'Arts' 
LIMIT 0, 10

Query 2:
It takes : 0.09 seconds (which is MUCH BETTER, but the ids doesn't match in all rows, because some URLs are duplicated in table 1, but all urls are unique in table 2 . So, I can't use it)

SELECT SQL_CALC_FOUND_ROWS `table1`.`url`, `table1`.`Title`, `table1`.`Description`, `table2`.`col16`, `table2`.`col17`, `table2`.`col18`, `table2`.`col19`
FROM `table1`, `table2` 
WHERE `table1`.`id` = `table2`.`id` 
AND `topic` = 'Arts' 
LIMIT 0, 10

Table 1 "3,206,543 rows" :

`table1` (
  `id` int(11) unsigned NOT NULL auto_increment,
  `url` varchar(255) character set utf8 collate utf8_bin NOT NULL,
  `Title` varchar(1024) character set utf8 collate utf8_bin NOT NULL,
  `Description` varchar(4000) character set utf8 collate utf8_bin NOT NULL,
  `topic` varchar(1024) character set utf8 collate utf8_bin NOT NULL,
  PRIMARY KEY  (`id`),
  KEY `topic` (`topic`(333)),
  KEY `url` (`url`)
) ENGINE=MyISAM  DEFAULT CHARSET=utf8;

Table 2 "2,846,092 rows":

`table2` (
  `id` int(10) unsigned NOT NULL auto_increment,
  `url` varchar(255) NOT NULL,
  `title` varchar(255) default NULL,
  `description` varchar(1024) default NULL,
  `keywords` varchar(1024) default NULL,
  `col1` varchar(1024) default NULL,
  `col2` varchar(1024) default NULL,
  `col3` varchar(1024) default NULL,
  `col4` varchar(50) default NULL,
  `col5` mediumint(8) unsigned default NULL,
  `col6` int(11) unsigned default NULL,
  `col7` int(11) unsigned default NULL,
     |
     |
     |
  `col23` int(11) unsigned default NULL,
  PRIMARY KEY  (`id`),
  UNIQUE KEY `url` (`url`),
) ENGINE=MyISAM  DEFAULT CHARSET=utf8;

So, what should I do now ?

It's wrong way to combine tables in query by varchar column. It's possible, but it's step to kill performance.
In any way url column from first table is not indexed.
I'm not sure UNIQUE KEY on url column from second table is useful. Guess there should be INDEX as well.

Second query is fast because id columns from both tables are INDEXED.

thanks MapMaster, I think I'll re-construct the 2nd table with integer primary key than the string one
Thanks again

1st, what mapmaster told you is VERY valid.
2nd, do you have indexe(s) created on table1.url and table2.url ?
This could help dramatically.

create index t1_url on table1(url);
create index t2_url on table2(url);

Try again with those.

3rd, what is the result of this query, that first compute a list of unique url's from table1 and then join to table2?

SELECT 
  SQL_CALC_FOUND_ROWS 
  t1.url, t1p.Title, t1p.Description, t2.col16, t2.col17, t2.col18, t2.col19
FROM (
  select url, max(id) as id
  from table1
  where topic = 'Arts'
  group by url
) as t1
  inner join table1 t1p  --join again to get Title and Description
    on t1.url=t1p.url
    and t1.id=t1p.id
  inner join table2 t2 
    on t1.url=t2.url
LIMIT 0, 10