Load Balancing

Top  Previous  Next

 

Load balancing and Fault-tolerance are methodologies to distribute workload across multiple services to achieve optimal resource utilization, avoid overload and allow the system to operate properly in the event of failure of some of its components.

 

On this help section you will learn how to set a network configuration using a combination of Round-Robin DNS and the Load Balancing feature included in ThinRDP Server.

 

Round-Robin DNS

Round-Robin DNS is a simple method of load balancing, where a list of IP addresses are associated with a single domain name. The list is continuously permuted, so the returned IP address varies for each DNS response.

 

ThinRDP’s Load balancing feature

 

ThinRDP Server can be configured in two basic ways: normal mode and Load Balancing mode.

 

Normal mode:

This is the default mode in which one single ThinRDP Server centralizes all the web requests and the same server creates and processes all the RDP connections.

 

In some occasions, this configuration may cause an overload of the ThinRDP Server machine resources. Some examples are too many concurrent users establishing connections at a time, or also when ThinRDP integrates applications that handle a lot of graphics, sound and other elements that require a great availability of resources. In all those cases, there is a moment in which one machine is not capable of managing all the required resources. This is when you should start considering using the Load Balancing mode.

 

Load balancing mode:

In this mode, ThinRDP must be installed in two or more servers that will participate in the load balancing /fault-tolerance scenario. Two possible roles can be configured:

 

Connection Broker: Under this role, ThinRDP responds to all web-pages requests and, when an RDP connection to a remote desktop is solicited, it selects the appropriate Connection Server to forward that request. The final RDP connection is done through the chosen Connection Server.

In case any established connection fails, or a Connection Server falls down, the Broker will be able to reconnect to the Server with the highest availability at a that moment.

All the system settings and profiles are centralized and stored on this server.

 

Connection Server: Under this role, ThinRDP processes forwarded RDP connections only. This server is responsible for establishing and processing the RDP connections assigned by the Connection Broker. All Connection Servers must have their IPs public to the client’s Web Browser, so that once the RDP connection has been assigned by the server, the browser can redirect its request to this new server.

 

Some of the benefits of using the Load Balancing architectures are:

 

Avoid the overload by distributing the connections among different servers
Minimize response time
More reliability (redundancy)
Fail over control

 

Before starting to configure a distributed environment to work, there are some previous steps you should go over :

 

1. Choose the architecture out of the three possible ThinRDP Load Balancing architectures, that will best fit your need.

2. Plan the machines that will work as Connection Brokers, Connection Servers/ThinRDP Servers and DNS Servers, depending on the chosen architecture.

3. Make sure all their IP's addresses are public to the Web Browsers that will access ThinRDP.