Monitoring overview

Prev Next

What to monitor

System monitoring and logs should be used to answer three broad questions:

Is the infrastructure operating as intended?

The AppGate ZTNA Collective can span many locations and networks and include multiple redundant instances. Because of the core design principle, each instance can operate with a high degree of autonomy. A Site with only three operational Gateways out of five appears healthy until available capacity is exhausted.

Verify that the entire system operates as expected. If you have multiple Controllers, confirm that all are active. If you have a primary and a backup system, confirm that both are available. If you are exporting audit logs, verify that exports are succeeding rather than filling the disk

It may not be possible to answer all infrastructure questions from AppGate ZTNA logs or metrics alone. You may need to gather additional information from other related systems to get a complete picture of the infrastructure.

Are functions performing to required specifications?

Functions refer to the different functions you can enable on an appliance: Controller, Gateway, Connection Broker, Connector, Portal, LogForwarder, LogServer, and Metrics Aggregator.

Display metrics that confirm each function performs as intended. For example, if a function forwards logs, the metrics must show how many logs it forwards. A system designed to forward a maximum of 1,000 log records per minute may perform differently after commissioning—the metrics reveal the actual throughput.

In addition to confirming that a function is working, metrics show how well the job is being done. For example, you might find that the Ax-H3 appliance signs in users in an average of 200 ms, whereas the reserve machine (a VM) takes 600 ms for the same operation. This historical data is useful for capacity planning and hardware investment decisions.

Most systems operate in an HA configuration. Clear graphical metrics that confirm appliances are in balance with each other are therefore an important consideration.

When evaluating a function’s ability to perform, identify conditions that could prevent it from doing so. Denial-of-service (DoS) attacks are one example—certain metrics can indicate that a DoS attack is beginning on a specific appliance.

Many metrics are provided to help you spot sudden changes, differences between appliances, and trends over time. Absolute values are included for completeness rather than as exact measurements. For example, several metrics relate to sign-in times on the Controller, and multiple sign-ins may run concurrently, so summing these values does not produce a meaningful total.

Is operation impinging on any physical limits?

Without a clear understanding of the inner working of the AppGate ZTNA system, it can be difficult to know what to monitor and which audit logs matter. Many customers start by monitoring familiar metrics such as CPU, RAM, and disk. This is a useful starting point, but the results often surface either design characteristics—such as Java allocating large amounts of heap space that it does not release because heap resizing is expensive—or design faults such as a file that is not being cleaned up and is consuming an increasing amount of disk space.

Controllers, for instance, should not normally consume significant RAM. If RAM consumption is elevated, a backlog is likely forming, such as from long-running scripts. A Controller can operate at near 100% CPU, but disk space exhaustion causes serious problems

The Portal supports a maximum number of concurrent user sessions. Monitor active user sessions to verify that the Portal operates within its supported limit. RAM utilization is not a reliable indicator of Portal health. The Portal dynamically allocates memory based on available RAM and always uses a minimum of 25%.

Also consider limits imposed by external systems. The system relies on DNS, and DNS servers or firewalls between the AppGate ZTNA appliances and the DNS server may enforce a per-second request limit. Exceeding that limit can drop requests and deny user access. Monitoring DNS requests per second is therefore relevant in some deployments.


Monitoring

Plan to monitor the AppGate ZTNA system continuously. Where appropriate, configure alerts and take corrective action before user access is affected.

Additional monitoring is not required from the outset. The Collective performs about 50 site, appliance, and functional health checks by default. When the system is unhealthy, the Dashboard displays the status in the Sites widget or the Appliances widget, where you can view appliance health details.

The Appliance Health Details window of the admin UI.

This provides only a short-term snapshot of the Collective’s health. As you scale up the number of users, longer-term monitoring is important for understanding how the Collective is performing.

The AppGate ZTNA system has two means of providing metrics used for monitoring the system:

  • SNMP is available via the AppGate MIB.

  • Prometheus metrics, which are equivalent to those in the AppGate MIB. Prometheus is the recommended method for enterprise usage.

The built-in healthcheck status is also available as a metric and is a useful starting point. The apn_function_status and apn_status metrics reflect the status shown on the dashboard (0=healthy, 1=busy, and so on).

Prometheus Exporter

The Prometheus Exporter runs an HTTP server that listens on a configured port and serves appliance metrics at GET /metrics.

The Metrics Aggregator

The Metrics Aggregator collects and groups Prometheus metrics from the Collective by Site, then exports them in a single endpoint. This removes the need to configure the Prometheus Exporter on each appliance and eliminates the need to open inbound firewall access to each appliance in the Collective.

From a connectivity perspective, the Metrics Aggregator functions the same as the Prometheus Exporter—it is a metrics endpoint that requires a Prometheus server to connect and scrape metrics at a defined interval. The key difference is that the Metrics Aggregator presents metrics from multiple appliances, grouped by Site, in one place. This reduces the required inbound firewall rules to one and simplifies configuration.

You can configure multiple Metrics Aggregators—for example, one per geographic region.

For more detail on what to monitor, see the Monitoring the AppGate System white paper, which includes three example Grafana dashboards published in the AppGate Github repository.

The Telemetry Aggregator

The Telemetry Aggregator receives performance signals—such as logs and system metrics—from multiple appliances in your Collective. Once the data is collected, the Telemetry Aggregator sends it to your preferred monitoring tool or security platform (SIEM), such as Grafana LGTM or OpenObserve. You can configure the Telemetry Aggregator to collect data from all appliances or limit it to appliances from specific Sites, depending on your monitoring needs.

See Observability and Adding a Telemetry Aggregator for more information.


System logs

AppGate ZTNA continuously records all system, user, and admin actions. You can export these logs and, where appropriate, configure alerts and corrective actions when the system detects specific events.

AppGate ZTNA produces two types of log records:

  • Daemon logs. Used to examine the internal workings of the AppGate ZTNA system.

  • Audit logs. Used to record actions that the system performs.

AppGate ZTNA processes these two log types independently. Audit logs are handled by one of two appliance functions: the LogServer or the LogForwarder.

LogServer

The LogServer provides a local log server for use within the Collective. After you enable the LogServer, sign out and sign back in for the Audit Logs page to appear in the admin UI (Usage > Audit Logs). The LogServer is not intended for enterprise use, but is suitable for commissioning a new system or for smaller deployments that fall within its audit log capacity.

LogForwarder

The LogForwarder collects, groups, and securely distributes audit logs within an enterprise environment. Use the LogForwarder to export audit log records to an external system that retains persistent logs.The LogForwarder includes connectors for leading security information and event management (SIEM) systems, including Splunk, Kibana, and Azure Monitor.

Multiple LogForwarders may be needed when logs from different appliances must go to more than one destination. Consider using multiple LogForwarders when log volume is high, such as when collecting logs from multiple busy Gateways. Multiple LogForwarders can also be configured to send logs to the same destination.

Use the Usage menu to access information about system usage, including users, devices, applications, and audit logs


Usage menu items

The Usage menu contains six items by default. The following four items appear only when their corresponding feature is enabled:

  • App Health Reports. Available when the Digital Experience Monitoring license is active. Summarizes unresponsive hosts that users have attempted to connect to in the past 24 hours.

  • Discovered Apps. Available when the Application Discovery license is active. Displays usage of discovered apps defined by entitlements with broad access rights.

  • Audit Logs. Available when the LogServer function is enabled. Use audit logs to monitor, search, and report on user and device activity.

  • Issued Certificates. Available when certificate revocation list (CRL) is enabled. Tracks all certificates issued by the system.