You can design a deployment that supports mobile and API Access Management, web access management, or auditing and proxying. For each of these environments, you can choose a stand-alone deployment for proof of concept or deploy multiple PingAccess servers in a cluster configuration for high availability, server redundancy, and failover recovery.

You have a choice between using PingAccess as a gateway or using a PingAccess agent plugin on the web server. In a gateway deployment, all client requests first go through PingAccess and are checked for authorization before they are forwarded to the target site. In an agent deployment, client requests go directly to the web server serving up the target site, where they are intercepted by the agent plugin and checked for authorization before they are forwarded to the target resource. The same access control checks are performed by the PingAccess policy server in both cases and only properly authorized client request are allowed to reach the target assets. The difference is that in a gateway deployment client requests are rerouted through PingAccess gateway, while in an agent deployment, they continue to be routed directly to the target site, where PingAccess agent is deployed to intercept them.

PingAccess agent makes a separate access control request to PingAccess Policy Server using the PingAccess Agent Protocol (PAAP). The agent request contains just the relevant parts of the client request so that PingAccess Policy Server can make the access control decision and respond with instructions to the agent regarding any modifications to the original client request that the agent should perform prior to forwarding the request. For example, the agent can add headers and tokens required by the target resource. Under the PingAccess policy server's control, the agent might perform a certain amount of caching of information in order to minimize the overhead of contacting the PingAccess policy server, thus minimizing response time.

In both gateway and agent deployment, the response from the target resource is processed on the way to the original client. In an agent deployment, the amount of processing is more limited than in a gateway deployment. The agent does not make another request to the policy server, so response processing is based on the initial agent response. Consequently, the agent is not able to apply the request processing rules available to the gateway.

When designing a deployment architecture, many requirements and components must be identified for a successful implementation. Proper network configuration of routers/firewalls and DNS ensure that all traffic is routed through PingAccess for the resources it is protecting and that alternative paths, such as backdoors, are not available.

The following sections provide specific use cases and deployment architecture requirements to assist with designing and implementing your PingAccess environment.