* Use a piece of hardware, like Cisco's Local Director * Use round-robin DNS * Use a proxy who passes connects through to the application servers
Depending on the underlying application, it's possible that the user has to be tied to his webserver once a session is established (BroadVision for example needs that - look here for comments).
We're coping with that by Configuring our servers like the following:
* Load Balancer holds address www and transparently redirects to webservers www-1, www-2, www-3,... * The www-n Hosts have VirtualHost sections like that:
This results clients that initially hit on www to be redirect and then hardwired to e.g. www-4, but it has the drawback that the client will also bookmark www-4 and not www.
Handling dead webservers with the Cisco Local Director is no problem because it will recognize the dead link and take the machine out of service. Using a proxy or the round robin DNS solution will most likely require a bit of programming on a monitor machine, but shouldn't be that much of a hassle.
I'd suggest to preconfigure all www-n IP-Addresses on all machines, and as soon as one machine drops, another machine takes the appropriate interface up and gets a load-bonus on the balancing mechanism.
My standard advice for this question is "Read the first two chapters of "Programming Perl". Chapter one gives you a one-man-show version of the bells and whistles. Chapter 2 (The Gory Details) describes how the language works. After that you should have a basic grip on the language and things will be much easier.
These two chapters are the parts that don't come with the online documentation. *EVERYTHING* else in the book is also included in the documentation that comes with perl. I recommend a 'perldoc perl' for an overview of what's there. You'll most propably will need perlfunc, perlvars and perlfaq.
This applies of course only to experienced programmers who only migrate from another programming language. I'd recommend using something more wheenie-like than perl to actually *learn* programming, but that's a completely different story.
Slashdot Mirror
You have a variety of options here:
...
...
...
* Use a piece of hardware, like Cisco's Local Director
* Use round-robin DNS
* Use a proxy who passes connects through to the application servers
Depending on the underlying application, it's possible that the user has to be tied to his webserver once a session is established (BroadVision for example needs that - look here for comments).
We're coping with that by Configuring our servers like the following:
* Load Balancer holds address www and transparently redirects to webservers www-1, www-2, www-3,
* The www-n Hosts have VirtualHost sections like that:
NameVirtualHost IP-Address-Of-Host
<VirtualHost IP-Address-Of-Host>
ServerName www.our-domain.net
Redirect / www-n.our-domain.net
</VirtualHost>
<VirtualHost IP-Address-Of-Host>
ServerName www-n.our-domain.net
DocumentRoot
</VirtualHost>
This results clients that initially hit on www to be redirect and then hardwired to e.g. www-4, but it has the drawback that the client will also bookmark www-4 and not www.
Handling dead webservers with the Cisco Local Director is no problem because it will recognize the dead link and take the machine out of service. Using a proxy or the round robin DNS solution will most likely require a bit of programming on a monitor machine, but shouldn't be that much of a hassle.
I'd suggest to preconfigure all www-n IP-Addresses on all machines, and as soon as one machine drops, another machine takes the appropriate interface up and gets a load-bonus on the balancing mechanism.
My standard advice for this question is "Read the first two chapters of "Programming Perl". Chapter one gives you a one-man-show version of the bells and whistles. Chapter 2 (The Gory Details) describes how the language works. After that you should have a basic grip on the language and things will be much easier.
These two chapters are the parts that don't come with the online documentation. *EVERYTHING* else in the book is also included in the documentation that comes with perl. I recommend a 'perldoc perl' for an overview of what's there. You'll most propably will need perlfunc, perlvars and perlfaq.
This applies of course only to experienced programmers who only migrate from another programming language. I'd recommend using something more wheenie-like than perl to actually *learn* programming, but that's a completely different story.