Throttle Limits
---------------

* Admin (by default, this is disabled). We recommend that the Admin throttle is
  enabled and set to 450 API requests/min. The setting is per Unified Node.
 
  * Service Inventory (SI) Report: Relies on the per-user throttle setting to
    ensure adequate throughput.
    
    * For the Active-StandBy deployment model, we highly recommended that the SI
      report is configured to run on a specific non-Primary Unified Node
      (preferably in the Secondary Data Center as those nodes are likely to have a
      lower load). This results in faster performance, but there is not any
      protection against a single node failure in the middle of an SI report run
      (not very likely). The use of a Web Proxy is *not* recommended here as 25% of
      the initial requests are routed to the Primary Unified Node based on the
      recommended web weight settings at either Web Proxy. 
    * For the Active-Active deployment model, you can use a Web Proxy instead.
      The SI report would run slower, but this configuration provides protection
      against a single Unified Node failure during the SI report run.
      If a Web Proxy is used, then:
      
      * Use Web Proxy 2. This prevents routing to the Primary Unified Node based
        on the recommended web weight settings.
      * The Web Proxy knows the health of the UN and can route requests accordingly.

* Per User throttle for API Clients:

  * Default setting is 20 API req/sec per Active Unified Node. 4 Active Unified Nodes x 20 = 80 API
    req/sec (system wide) or 4800 API req/min (system wide). We recommend that
    you keep this setting.
  * This limit applies to all admin users, but in practice serves to
    limit API clients. Human admin users are not likely to create a traffic rate
    of 80 API req/sec.

* Self-Service throttle. The default setting is 300 API req/min (per Unified Node).
  APIs for logins and actions would count against this throttle.