Understanding the app
The Adobe Commerce Traffic Insights app visualizes raw Fastly Content Delivery Network (CDN) access logs into a picture of a store’s edge traffic. Charts are grouped into the following tabs:
- Bandwidth — How traffic bandwidth is distributed across domains, content and resource types, and Cloud projects over time.
- Full Page Cache Performance — How efficiently dynamic storefront HTML for product detail pages (PDP), product listing pages (PLP), and content management system (CMS) pages is cached at the edge.
- Bots Activity and Requests Analysis — Traffic broken down by known bot agents, geolocation, IPs/subnets, URLs, and Fastly Next-Gen Web Application Firewall (WAF) signals.
A fourth in-app Documentation tab carries conceptual notes and the Investigation playbook.
Who is this guide for?
- Site operators and Site Reliability Engineering (SREs) investigating CDN bandwidth overage, traffic spikes, or origin load.
- Developers tuning Full Page Cache (FPC) coverage and hit ratio, or implementing Fastly Varnish Configuration Language (VCL) rules.
- Administrators and security engineers identifying and mitigating unwanted bots, scrapers, and malicious automated traffic.
Familiarity with Adobe Commerce on Cloud Infrastructure, Fastly CDN concepts, and basic New Relic navigation is assumed.
How it works
Select an account and time range from the platform controls at the top of the page. An optional Project ID can further narrow charts to specific Cloud projects. In a master-account or partnership setup, being able to view an account in the dropdown does not mean that you can query it. If a chart reports a permission error, switch to an account you have New Relic Query Language (NRQL) access to.
You continue to apply filters to turn a broad overview into a focused investigation. Click a value in a facet column, such as a bot, IP, subnet, country, or content type to add a global filter. Active filters appear at the top of the grid and apply across every widget in every tab simultaneously. To widen the scope, remove a filter.
Walkthrough - Consider a scenario where the Total Bandwidth is trending above the contract allowance and you want to know who is driving it:
- Open the Bots Activity and Requests Analysis tab and read Bandwidth Structure to see how much of the traffic is automated versus organic.
- If bots appear to have more traffic, open Known Bots By Bandwidth and click the heaviest named bot, for example, a scraper. This adds a new filter, which means now every widget is scoped to that bot.
- Read Known Bots Impact Details for its request rate, status mix, and FPC hit rate.
- To see where the bot originates, check Bandwidth By Country. To see what the bot is fetching, see URLs By Bandwidth.
- If the traffic concentrates in one network, click through Stats By IP Subnets to confirm an actor rotating across addresses in a single block.
- You now have the who, what, and where needed to write a targeted mitigation. Continue to the Investigation playbook to learn about how to proceed.
The same filtering method works from any starting facet: a suspicious country, a single IP, a content type, or a URL path segment.
How the data is measured
Understanding a few measurement choices makes the numbers easier to trust and interpret.
-
Bandwidth (BW) is the total bytes the CDN served for the matching requests, counting both response headers and body. It is the headline cost metric that counts against the contract allowance.
-
Requests (Req.) is the number of distinct requests, however, with Fastly shielding enabled a single request is logged twice, once on each of the following:
- Internal shield Point of Presence (POP)
- Edge POP
This happens unless the response comes straight from the local POP cache or the shield is itself acting as the POP for the sender’s location. To avoid double-counting theseHIT,MISSandMISS,MISScases, the app’s queries aggregate withuniqueCountover therequest_idfield. This returns a close approximation with an expected margin of error of ~5%, not an exact count.
-
CDN network segments are compressed differently. The response delivered to the client is compressed, but shield-to-POP traffic is not compressed to preserve Edge Side Includes (ESI) support. A low cache hit ratio therefore inflates the internal segment more than the client-facing one, since uncached content must be pulled across the shield repeatedly at full, uncompressed size. This compression is why the CDN Network Segment Bandwidth widget and the FPC hit ratio are two views of the same underlying cost.
Data limitations & performance
- 30-day retention - Fastly CDN logs are retained in New Relic for 30 days per the subscription plan. Any window you pick must fall within the past 30 days. For longer-term total bandwidth, use the direct Fastly integration in the Adobe Commerce admin panel, Dashboard > Fastly > Bandwidth > Total, but consider that it reports per-service-ID, so data must be collected per environment and aggregated to compare against the contract allowance.
- 60-second query limit - Each chart’s NRQL has a 60-second execution limit. For very high-traffic accounts, a widget can time out while scanning too many log records. If this happens, reduce the time range and reload the charts. You can expand it again for lighter tabs.