The PerformanceGuard Frontend server handles the communication with PerformanceGuard agents, stores data from agents, and communicates with the PerformanceGuard Backend server to synchronize agent IDs and aggregate performance data. You can use multiple Frontend servers in your PerformanceGuard solution if required.

When you install a Frontend server, it will automatically register itself with the Backend server and receive a unique ID.

PerformanceGuard agents will automatically receive a Frontend server list as a part of their configuration. You can create a list of Frontend servers for a specific location. This can be useful if you want all the PerformanceGuard agents installed on the computers at a location to use one or more specific Frontend servers.

List of Frontend Servers

The Frontend servers list (ADMINISTRATION > Setup > Frontends) shows all registered Frontend servers with the following information:

• Status
• ID
• Frontend Server: Name of the Frontend server. By default the hostname of the Frontend server is used.

• Agent Delivery Address: The IP address or hostname and TCP port number that agents should use when communicating with the Frontend server.

  This is useful if you want agents to use a DNS alias, or if the server is situated behind a firewall (in which case you need to specify the address of the firewall).

• Version: The Frontend server's version of PerformanceGuard.

• Enabled: Whether the Frontend server is enabled (it is possible to disable a Frontend server when you edit its properties).

  You typically only disable a Frontend server if you need to perform maintenance on it. However, if you use multiple, prioritized Frontend servers (see the following), disabling a Frontend server can be a way of actively forcing agents to connect to a different Frontend server.

• Priority: The priority of the Frontend server. Agents use the priority when they decide which Frontend server to connect to (see the following).
• Agents (Online): The number of agents that use the Frontend server, and how many of those agents that are currently online.

Agent-Frontend Server Connection Priorities

An agent will only use one Frontend server at a time.

Agents will automatically receive a Frontend server list as a part of their configuration. By default agents will get a list of all enabled Frontend servers. If some agents belong to a location for which a specific Frontend server list has been defined (see the following), they will get a separate list of Frontend servers.

If your organization uses multiple Frontend servers, agents will connect to the Frontend servers in the following order:

1. Any primary Frontend server in the list
2. Any secondary Frontend server in the list
3. Any failover Frontend servers in the list
4. The Frontend server that was defined for an agent when the agent was installed

Once an agent has successfully connected to a Frontend server, it will stay with that Frontend server for the lifetime of Agent. However, if the agent has connected to a Frontend server that has a lower priority than another Frontend server on the list, the agent will try to connect to the Frontend server that has a higher priority at regular intervals to see if the higher-priority Frontend server has become available.

Agents will receive an updated Frontend server list if there are any changes to the list, that is if Frontend servers are added or removed, enabled or disabled, or if their priorities have changed.

Prioritization Isn't Load Balancing

Bear in mind that prioritization of Frontend servers does not in itself offer any load balancing. PerformanceGuard agents will simply follow the Frontend server list and Frontend server priorities they get as part of their configuration. The agents have no way of determining the load on specific Frontend servers, so they are not able to switch Frontend servers in order to distribute the load across servers.

Any such load balancing must be set up separately outside of PerformanceGuard.

You can achieve Semi Load Balancing by ensuring that:

• All Frontend servers, you need to balance between shall have the same priority.
• All Frontend servers shall be up and running at the time of connection.
• No Agents may ever have been connected to Frontend servers. If they have, then you must remove the RandomSeed.

DNS Alias Tip

Because of the way that agents connect to Frontend servers in an multiple Frontend server environment, the following can be a good way of installing PerformanceGuard agents and Frontend servers:

1. Create a DNS alias like performance.yourcompany.com and make sure that it points to one of your Frontend servers at all times.
2. Install the agents using this DNS alias as the configured Frontend server.
3. The agents will connect to the Frontend server behind the DNS alias. From that Frontend server they will receive the correct Frontend server list.
   
   This way agents will never be lost, because agents will always try to connect to the Frontend server that was defined during installation as their last resort. Because you control which server that is through your DNS administration, you can make sure that agents are always able to connect to a running Frontend server.

Create Specific Location-Frontend Server Mapping

If required, you can create a specific list of Frontend servers and priorities that agents at a specific location should use. This can be sometimes be useful for network infrastructure reasons.

If you want to set up a location-frontend server mapping, select ADMINISTRATION > Setup > Frontends and then select the Locations tab.

See also Set Up Locations.

Changes Take 30 Minutes or More

If you change your Frontend server configuration, for example by adding a Frontend server, by changing location-frontend server mappings or by changing priorities, you should be aware that it takes 10 agent delivery intervals before you can be sure that your changes have been communicated to all online agents. With a default agent delivery of three minutes, that's at least half an hour.

The reason for this is that communication of Frontend server changes takes up considerable agent resources. That's why PerformanceGuard does not let Frontend servers communicate this type of configuration changes to agents more often than every 10th agent delivery interval.