Frank,
Tyan is a brand name for a motherboard. I like this board because it is really stable. It took me less than 30 minutes to put together a dual server and get it going. You can buy Tyan on Ebay (search by Tyan Tiger 133) or you can get it retail.
RAID is redundant array of independent disks. It is a cool thing, and every high traffic server should have one. Linux supports RAID on software level, meaning that no hardware necessary. This is how RAID works... Let's say you have two hard drives - and a chunk of data. What is done next is that this data distributed evenly between these two hard drives. 50% goes on one hard drive and 50% on another. So access time to all data is much faster compared if you have only one hard drive. Of course if you have 5 hard drives, that are SCSI and rotate at 10,000 rpms that'll help a lot.
(http://www.redhat.com/support/manuals/RHL-7-Manual/ref-guide/ch-raid.html) This link describes RAID levels in detail, and also contains instructions on how to set up software RAID.
On a server hard drives are a real bottleneck, because they are slower than CPU, memory, and motherboard bus, so if you can speed them up a bit, than your server's performance will definitely go up, and RAID is the way to go.
As far as mysql servers - they definitely have to be on a LAN. Only on the LAN you can get speeds of 100 Mbps. T-3, which is 30 T-1 lines combined is 45 Mbps, and it'll cost you 20 grand or more a month. T-1 is a good line for remote access, but at 1200 dollars a month it is high. So bottom line is this - you set up your main server and connect it to 2 more mysql servers locally. Use switch. Switch establishes dedicated connections between communicating entities, and rate of collisions is very low compared to hubs.
So, let's say one server has LAN address 192.168.110.2 and another 192.168.110.3. PHP script will say $conn=mysql_connect("192.168.110.2", "user", "password"), and then it can query. The query will be processed on a mysql server, and not the main server - and that is the main advantage, since databases eat a lot of memory and CPU cycles. Because connection between main server and mysql server is 100 Mbps through switch - the traffic on the network should be flowing nice and smoothly.
Make sure that network settings should be configured to read hosts file first.
As far as knowing how many connections well - I never did before, but below is just an idea
$conn=mysql_connect("locahost", "user", "password);
if($conn)
{
$fp=fopen("myfile.txt", "a+");
$fw=fwrite($fp, "Conn");
}
Basically, if $conn is established, than PHP will append "Conn" to the file. The problem with it is that there is no such thing as auto_increment, however this is how you can count how many connections were made.
$fl="myfile.txt";
$fp=fopen($fl, "r");
$fr=fread($fp, filesize($fl));
$num=explode("\n", $fr);
$how_many=count($num);
Basically, "myfile.txt" is opened and read into variable $fr, than variable $fr is exploded by using "\n" (endline element). Output from explode is an array, so the only thing you have to do is count how many elements are in array, which is what I did in the last line, and that'll tell you how many mysql connections were established. Of course you can use time functions to pinpoint when were the most connections made.
Best way to analyze data is using database :0), but that'll really overload it, and I think that you can do pretty good job using just files.
Hope that helps,
Di
