F5 Load balancing

I am putting tomcat web servers behind the F5 LB.  How does it balancing the load.  Per connection or per load on the server?  How much the load is too much per server?  How does it know when too much so to point to the next server?
Can anyone explain please?
LVL 17
Tiras25Asked:
Who is Participating?

Improve company productivity with a Business Account.Sign Up

x
 
giltjrConnect With a Mentor Commented:
No problem.  The F5 can have two levels of monitoring.

First level is to ping the IP address of the host.
Second level would be a simple HTTP get of a static page that is served up by the tomcat server.

The either one of these fails, the F5 marks that node/service as being down and stops sending requests to it.
0
 
giltjrConnect With a Mentor Commented:
The F5 has multiple load balancing options:  You have to choose what is best for you.

Are all of the servers it will be front ending at the same physical location?
Are they all equal in performance?
Do you need to have session persistence?

Ref: http://www.fir3net.com/Big-IP-F5-LTM/f5-balacing-methods.html

You can choose:

Round Robin - Evenly distributes requests to all available pool members.

Ratio - Ratio allows each server to be assigned a ratio value. This is useful for pool members that have greater or lower computing resources then others. Example : Ratio 3:2:1:1. Based upon 8 requests, 3 requests would go to 1, 2 to 2 then 1 to 1.

Least Connections - Traffic is balanced to servers with the least total of current connections.

Fastest - Connections are distributed to pool members based upon server response time.

Observed - This method is the same as ratio but the ratio is assigned by BigIP.  Each ratio is calculated based upon the total number of connections currently active on each pool member. A pool member with a lower then average connection count is assigned a ratio of 3. A pool member with a higher than average count is given ratio of 2.

Predictive - Predictive is similar to observed but ratio`s are assigned using much more aggressive ratio values. A pool member with a lower then average connection count is assigned a ratio of 4. A pool member with a higher then average count is given ratio of 1.
0
 
Tiras25Author Commented:
Thank you very much for extensive answer.  Let me read through again and digest this.  

Yes, all the servers are windows virtual machines on ESX and will be in the same physical location.  They all will be an equal performance.  
Yes, we need a session persistence.
Thanks!
0
 
giltjrConnect With a Mentor Commented:
With all servers being equal Round Robin is the suggested/preferred method of load balancing.

The F5 provides multiple methods of session persistence.  You can read the the doc to see what may be best for you.  Off hand I would say Cookie.
0
 
Tiras25Author Commented:
What would happen if one of the servers tomcats slows down.  Then I'd want the traffic to go to another web server.
0
Question has a verified solution.

Are you are experiencing a similar issue? Get a personalized answer when you ask a related question.

Have a better answer? Share it in a comment.

All Courses

From novice to tech pro — start learning today.