The Aperture Agents are installed in your cloud environment, co-located with your workloads or on a separate cluster. Aperture SDKs, middleware, and proxy integrations connect with the Aperture Agents for real-time decisions and for exporting telemetry data. Aperture Agents are designed to be scaled horizontally as your workloads grow. Aperture Agents constantly share state with the Aperture Cloud and other Aperture Agents in the same cluster.

Aperture Agents can be installed as a DaemonSet in Kubernetes clusters (i.e. one Aperture Agent per node), as a sidecar within your service instances (pods or VMs) or on a dedicated cluster. Though a single Aperture Agent can handle a large number of requests, we recommend installing Aperture Agents as close to workloads as possible to minimize latency and maximize performance. Co-locating Aperture Agents with workloads also makes infrastructure saturation telemetry collection easier.

Aperture Agents relay telemetry data to the Aperture Cloud. This telemetry data includes request metadata, infrastructure saturation, and Aperture Agent health metrics. This telemetry data is used to drive control decisions and to provide comprehensive observability. You can configure what request labels and metrics are exported to the Aperture Cloud. Aperture Agents also receive control decisions and policy configuration from the Aperture Cloud. These control decisions are based on the telemetry data that is collected from the Aperture Agents.

Aperture Agents are designed to be resilient to network failures or Aperture Cloud outages. Even without connectivity, agents continue executing configured policies. Although they will discard any outdated control decisions related to adaptive policies after a configurable timeout.

Integrated SDKs, middleware, and proxies are designed to be resilient to Aperture Agent failures. They fail open by default and continue to serve requests without any policy enforcement. The fail open behavior can be configured for each control point to reject requests, which is useful when running load ramp policies.

Aperture Cloud is a SaaS offering that provides a fully hosted control plane and advanced observability. This includes support for caching requests, rate limiting, and request prioritization. These features are available in both the Serverless and Infrastructure plans.

An agent-hour represents each hour that an Aperture Agent is active within your infrastructure. For monthly billing, we accumulate and aggregate all the agent-hours from throughout your infrastructure. We round up to the nearest full hour for any partial durations. Our pricing is pay-as-you-go model and there are no overage fees.

In Aperture Cloud, a "project" represents a unique entity, complete with its dedicated cloud controller and agents. This is offered at a consistent monthly rate of $50, which includes up to 1 million good requests. Beyond this quota, additional requests are priced at $0.05 per 10,000 good requests. Projects facilitate precise control over permissions, allowing for the assignment of specific roles and user groups to members of an organization to manage policies and Aperture.

Yes, credit card details are required for sign-up with Aperture Cloud. You won't be charged until the free 30-day trial ends, and only if you decide to continue. We'll notify you as your trial nears its end to prevent unexpected charges.

Absolutely! Should you ever decide that Aperture Cloud is no longer relevant for your business needs, you can request cancellation. Simply send an email tosupport@fluxninja.com expressing your intent to do so.

If you decide to stop using Aperture Cloud, we will remove your organization from our system after 30 days. Once your organization is deleted, you will no longer have access to your data. We will send you a notification 7 days prior to the deletion of your organization.

Requests that have been accepted by the Aperture Agent. In other words, requests that meet the criteria and policies set for evaluation and are approved by the Aperture Agent are considered 'good requests'.