We've been leveraging Akamai Prolexic for DDoS mitigation on several client-facing API endpoints for about 18 months, and the performance has been generally excellent. However, following the most recent platform update (I believe the rollout was around late Q3), our monitoring is showing a consistent, measurable increase in latency for traffic routed through the Prolexic scrubbing centers.
Our typical baseline for round-trip time from East US users to our origin, through Prolexic, was 42-48ms. For the past three weeks, we're consistently seeing 68-85ms for the same routes. This is significant for our real-time bidding API, where we have SLAs under 75ms.
We've ruled out our own infrastructure and peered networks. The latency jump appears immediately after the traffic is handed off to Akamai and persists until it's returned to our edge. A traceroute analysis shows 2-3 additional hops within the Akamai network post-update, and increased latency at those specific hops.
**Our current hypothesis:** The new routing logic or additional security inspection layers in the updated scrubbing centers are adding processing delay. We've opened a ticket with Akamai support, but their initial response was that "all metrics are within normal ranges."
I'm curious if others in the community have observed similar behavior and, more importantly, if you've found any configuration levers to pull to mitigate this.
* Are you seeing this in specific regions (e.g., US-East, EU-Central) or globally?
* Has anyone successfully negotiated or configured a more performance-oriented routing profile with Prolexic, perhaps at the expense of some less-critical checks?
* For those using APIs, have you had to adjust timeout thresholds or implement more aggressive regional failover?
Our temporary workaround has been to shift some non-critical traffic to a direct route, but that's not a sustainable solution for our core services. Any shared experiences or diagnostic comparisons would be invaluable.
API first.
IntegrationWizard
We saw the same latency pattern after their Q3 update. Our finance clients were hitting timeouts on settlement calls. Akamai's support took a week to escalate beyond "within acceptable thresholds" - you have to push hard and present the SLA impact data.
It turned out, in our case, the new geo-routing rules were sending our East Coast traffic through a Midwest scrubbing center for "capacity balancing". They re-pinned our prefixes after we showed the breach reports. Might be worth asking if they've changed their traffic engineering logic for your IP ranges.