STRATEGIC BUSINESS PARTNERS

Platin Bilişim has cooperations with the strongest and leader brands. Our target is not only make a difference on our rivals via our certificated and powerful engineer staff but also; providing IT security products and services for our customers from end to end.

F5 Networks Load Balancing

Load balancing methods are considered in two categories: Static or Dynamic. Static load balancing methods distribute received connections in a uniform and a predictable manner without considering received factors or current conditions. Dynamic load balancing methods distribute the connections with considering current conditions when it gives a load balancing decision.

Static Load Balancing Methods:

  • Round Robin: This is the default load balancing method that is considered at BIG-IP system.  System is static because of not using the information of dynamic working hours to select next pool member. It is a method that consequence distribution is made without making analysis.
  • Ratio: It is a method that makes distribution according to given user-described ratio to the pool members.

 

Dynamic load balancing methods:

  • Least Connection: It’s a method that steers requests to the server that has the least connection. This method considers the number of existing connections between BIG-IP system and server as valid. It doesn’t account connections of servers with other systems.
  • Fastest: It steers to the server that response faster. Decision depends on the response speed that turns to the applied monitor.
  • Observed: If the response times are equal, the pool member who has the least connections, going to be selected and then next connection is transferred. Observed method is the condition that the fastest and the least connection methods are used together.
  • Predictive: It is same as observed method instead of including rationing with time. This method decides after watching in a certain period of time.
  • Dynamic Ratio: According to SNMP queries, it takes weights as a dynamic and makes the steering according to it.

 

There are two different load balancing methods that are determined according to pool member or node.

  • Member load balancing is made with looking at only member’s certain pool statistics
  • Node load balancing looks at all service pools that are defined for node

 

As an example, if both HTTP and FTP executes over one server and services are taken as http_pool and ftp_pool, then least connection (node) looks at total connection number over two services.

 

Priority Group Activation: Priority Group Activation load balancing allows usage of pool members only if preferred pool members aren’t accessed. BIG-IP system assigns a priority to each pool member and first sends the link to the members of the pool with the highest priority. System assigns the lowest possible number of members and if a member of the pool becomes less available, the system enables the members of the next highest priority pool.

 

Fallback hosts: If all members of a pool are unavailable, a fallback host configured on virtual servers which can be added into an HTTP profile, can send an HTTP redirect message to the client. This allows you to send a connection to the alternative website or server. You can configure this condition with HTTP profile.