API Security in 2025: Akamai Warns of 67% Surge

Listen to this article · 8 min listen

A new report from Akamai Technologies is sobering: web app and API attacks shot up 67% in the first half of 2025. That’s not just a statistic, it’s a direct warning that without a solid API rate limiting strategy, your digital infrastructure is dangerously exposed. If you don’t have effective controls, your backends are just waiting for an overload, a data breach, or a complete service disruption. Building and managing these defenses is the main challenge in this increasingly hostile environment.

Key Takeaways

  • Your rules can’t be static. They need to be dynamic and adapt to weird traffic patterns on the fly to keep the service from degrading.
  • Give priority to API requests based on who the user is, like their role or subscription level, so your critical or paying users get the resources they need.
  • Don’t rely on one tool. Use a layered defense that combines API gateways, Web Application Firewalls (WAFs), and DDoS protection services to cover your bases.
  • Constantly audit and attack your own rate limiting setup with known vectors. It’s the only way to find and patch the holes before someone else does.
  • Teach your dev teams about secure coding and why designing a clear API contract from the start helps reduce the number of endpoints that can be exploited.

The 72% Surge in Credential Stuffing Attempts: A Direct Threat to Backend Integrity

That 72% global jump in credential stuffing attacks, reported in Imperva’s 2025 Bad Bot Report, is a direct threat to your backend’s stability. The damage goes way beyond a few stolen accounts. Every single one of those login attempts is a request hammering your auth APIs, looking for a match. We’re talking about thousands, sometimes millions, of requests an hour. Backends just aren’t built to take that kind of sustained, malicious load. Without controls, these attacks will chew through your database connections, pin your CPUs, and clog your network, which means your real users see slowdowns or can’t get in at all. Standard authentication checks are completely outmatched by this sheer volume. You have to put a protective layer of infrastructure in front of those auth services to even stand a chance.

Feature Dynamic Rate Limiting Layered Defense Approach Regular Auditing & Testing
Adapts to Real-time Traffic ✓ Yes ✗ No ✗ No
Prevents Service Degradation ✓ Yes ✓ Yes ✓ Yes
Combines WAFs & DDoS Protection ✗ No ✓ Yes ✗ No
Identifies Vulnerabilities ✗ No ✗ No ✓ Yes
Reduces Recovery Time ✓ Yes ✓ Yes Partial (indirect)
Addresses Credential Stuffing ✓ Yes ✓ Yes Partial (identifies weaknesses)
Impacts User Patience Threshold ✓ Yes Partial (performance) Partial (performance)

The 40% Performance Degradation Under Load: The Cost of Inadequate Protection

A Gartner Research study from early 2026 found that systems without proper API rate limiting get hit with a 40% performance drop during attacks or even just unexpected traffic spikes. This has real consequences: lost revenue, angry users, and a tarnished brand. When an API endpoint built for a few hundred requests a second suddenly gets slammed with thousands, the entire system starts to buckle. You’ll see database queries timing out, app servers going unresponsive, and cache layers getting blown out. The performance hit isn’t a straight line, it cascades as failing parts put more stress on everything else. For an e-commerce site, a 40% slowdown during a flash sale directly translates to millions in lost sales. For a fintech app, it means critical transactions just fail. The cost of skipping this protective layer is always higher than the cost of building it.

The 10-Minute Recovery Time for Unprotected APIs: An Unacceptable Downtime Metric

Google Cloud’s own API management guide points out that an API without good rate limiting can take over 10 minutes to recover from even a moderate denial-of-service attack. In internet time, ten minutes is an eternity. Think of the chaos in those ten minutes: transactions are failing, support tickets are piling up, and PagerDuty is screaming at your on-call engineer. The recovery itself is usually a frantic manual scramble of restarting services and clearing jammed queues, stretching the real downtime even longer. This 10-minute metric points to a deep-seated flaw in so many API designs which are reactive instead of proactive. A well-placed rate limiter blocks the junk traffic at the edge before it ever gets to your core services, bringing recovery times for most common attacks down to basically zero. This kind of proactive defense is non-negotiable if you care about uptime.

The 3-5 Second Threshold for User Patience: The Human Element of API Performance

This isn’t just about security. User experience research, like the kind from Nielsen Norman Group, has shown for years that users start getting antsy after a 1-2 second delay, and many will just leave after 3-5 seconds. This simple fact about human patience makes a strong case for vigilant API rate limiting. Even if you’re not under attack, a spike in legitimate traffic from a marketing campaign can introduce enough latency to kill your user experience. If your mobile app’s API calls suddenly take 4 seconds instead of 500ms because you got popular, people will close the app and go to your competitor. Rate limiting is also about guaranteeing fair access and consistent performance for everyone, especially during high-demand periods. It’s about preserving the quality of service that actually keeps people using your product.

My Disagreement with “One Size Fits All” Rate Limiting

I hear this advice all the time: just apply a simple, uniform rate limit across your endpoints. Something like “100 requests/minute for reads, 10 for writes.” Honestly, this approach is far too simplistic and often does more harm than good. Sure, it gives you a paper-thin layer of protection, but it completely ignores how your app is actually used and how much work each API call creates on the back end. Think about it: a user fetching their profile data once is completely different from a trading bot polling for stock prices every second. An API call that returns a cached value is cheap, but one that fires off a complex database join and hits another service is incredibly expensive. A blanket limit will inevitably throttle legitimate, high-value traffic or, even worse, leave your most resource-heavy endpoints wide open for attack. We have to get smarter with intelligent, adaptive, and context-aware rate limiting. This means looking at things like user roles, subscription tiers, past usage patterns, and the real computational cost of an API call. A dynamic system that adjusts limits based on what the backend is actually experiencing is more work to set up, yes, but the payoff in security and performance is massive. For any startup trying to scale, pairing this kind of nuanced rate limiting with intelligent AI backend scaling is going to be non-negotiable.

Putting effective API rate limiting in place isn’t some optional extra anymore. It’s a fundamental part of building any reliable digital service. By actively managing your API traffic, you can prevent overloads, keep performance steady, and give your users a consistent experience they can depend on. This forward-thinking posture is a huge part of overall app resilience when the tech field gets chaotic. And with everyone paying more attention to app privacy lessons, locking down these critical interfaces is just common sense.

What’s rate limiting actually for?

Its main job is to control how many times a user or client can hit your API in a given period. This stops abuse, keeps things fair for everyone, and most importantly, protects your backend servers from getting overwhelmed by traffic, especially from denial-of-service attacks.

How is this different from a WAF?

Rate limiting is all about managing the *volume* of requests. A Web Application Firewall (WAF) is different. It looks at the *content* of each request to find and block attacks like SQL injection or cross-site scripting. They are two different layers of security.

Can this mess things up for my real users?

Yes, if you set it up poorly, you can absolutely end up blocking or slowing down legitimate users. This is a big problem during traffic spikes. That’s why smart, dynamic rate limiting that adjusts to real-time conditions is so important, it’s designed to minimize that negative impact.

What are the common ways to set this up?

The common technical approaches are fixed window counters, sliding window logs, and token bucket algorithms. In practice, you’d implement this using an API gateway, a reverse proxy like Nginx, or a dedicated service that enforces rules based on things like IP address, API key, or user ID.

Is it okay to just do rate limiting on the client-side?

No, not for security. While client-side limiting can be a nice-to-have for user experience, it’s useless as a security control because any attacker can easily bypass it. You always need to enforce the limits on the server-side for real protection.

Editorial Team

The Editorial Team at Apps Scale Lab is a group of writers and editors who cover technology, products, and how teams ship. They research each story and publish under a shared desk byline — experts in their fields, working as one team.