Akamai DataStream

Use cases: what can you do with DataStream?

For a step-by-step walkthrough of many of the capabilities mentioned below, watch the DataStream use case demo video here.

Use case #1: Before-and-after monitoring

Validate the success—both before and after—of your new code or CDN configuration enhancements in real time. You can do this type of monitoring in production and at scale without risking downtime or jeopardizing end-user experience, while also measuring the impact of a recent code change or new deployments on CDN health, efficiency, and usage.

Use case #2: Continuous monitoring

By continuously ingesting DataStream logs into your preferred log analytics platform, you gain the ability to set up real-time dashboards or alerts for constant, proactive mitigation of issues including:

  • Connectivity problems
  • Service disruptions
  • Configuration-tuning complications

Ultimately, these proactive mitigation efforts allow you to minimize your mean time to recovery (MTTR). In addition, because DataStream gives you the ability to select specific log fields in order to focus on your most important metrics, you can do this continuous monitoring without having to pay for (or be inundated with) data you don’t need.

Use Case #3: CDN behavior benchmarking

With DataStream, you can access recent data (up to 24 hours*) on CDN health, efficiency, and usage for analysis and benchmarking, without the overhead of managing an endpoint infrastructure.

*You can choose to store data beyond the default 24-hour window by setting up your own storage with a secured repository (e.g., Amazon S3 or similar) for offline and historical analysis.

Use case #4: Receiving pre-aggregated metrics in addition to raw logs

With DataStream, you can choose to receive pre-aggregated metrics over a chosen window of time. Unlike raw logs, pre-aggregated metrics are a collection of meaningful, actionable data that don’t require significant downstream processing at your end.

For example, you could choose to have your aggregated stream always on for a continuous, high-level view of your CDN health. If any anomalies occur in the aggregated streams (such as high error count or high average origin response times), you can then turn on raw logs for root cause analysis and diagnostics.

Below is additional detail on raw logs and aggregated metrics.

1. Datasets delivered as raw logs

In the DataStream screenshot below, you can see the range of raw data sets that you can choose to monitor via simple checkbox selection.

datasets1datasets2

NOTE: You can find a full description of the data model for the DataStream API, along with a sample raw log report, here.

2. Datasets delivered as aggregated metrics

Below is a list of aggregated metrics that DataStream can deliver over your chosen aggregation time frame (i.e., 5 minutes, 15 minutes, 30 minutes, or 1 hour):

Traffic volumes
  • Requests per second to edge

  • Bytes per second from edge

CDN offload
  • Count of requests that were a cache hit

  • Count of requests that were a cache miss

  • Offload rate: (cache hits) / ( total requests) over the period

HTTP status codes
  • Count of requests that resulted in 2xx

  • Count of requests that resulted in 3xx

  • Count of requests that resulted in 4xx

  • Count of requests that resulted in 5xx

Edge response time
  • a cache-hit at Akamai

  • a cache-miss at Akamai

  • a cache-hit

  • at edge (child) level

  • at a parent level

  • a cache-miss

  • at edge (child) level

  • at a parent level

  • Average latency observed for non-cacheable requests

Origin response time
  • Typical latency observed between when Akamai requests an object from origin, and when it is returned to Akamai

NOTE: You can find a full description of the data model for the DataStream API, along with a sample aggregate log report, here.