Good API rate limits protect backend capacity while giving legitimate clients a predictable share of it. Choose a policy by deciding what resource you need to protect, which callers share an allowance, how much burst traffic to permit, and what should happen when the limit is reached. The algorithm matters—but it does not, by itself, determine whether requests are rejected, delayed, or counted consistently across gateway replicas.
What an API rate limit controls
A rate limit is a policy that evaluates requests over time for a defined identity or scope. It can protect an upstream service from overload, prevent one consumer from taking an unfair share, or enforce an aggregate capacity boundary. A policy therefore needs more than a number: it needs a counting interval or refill rate, a key that identifies whose requests count together, and an enforcement action.
As an Amazon Associate I earn from qualifying purchases.
Do not treat every traffic control as the same thing. A request-rate limit controls how many requests arrive over time; a longer-period quota caps use over a broader period; a concurrency limit controls how many operations are in flight at once. An expensive operation that holds database connections for a long time may need a concurrency cap in addition to a request-rate limit.
Which rate-limiting algorithm fits the traffic?
The algorithm determines how usage is counted or smoothed. It does not guarantee identical behavior across API gateways: blocked-request counting, window resets, distributed state, and the action taken on excess traffic depend on the product and configuration.
#1 Best Overall
- Dual band router upgrades to 1200 Mbps high speed internet (300mbps for 2.4GHz plus 900Mbps for 5GHz), reducing buffering and ideal for 4K stream
- Full Gigabit Ports - Gigabit Router with 4 Gigabit LAN ports, ideal for any internet plan and allow you to directly connect your wired devices
- Boosted Coverage - Four external antennas equipped with Beamforming technology extend and concentrate the Wi-Fi signals
- MU-MIMO technology - (5GHz band) allows high speeds for multiple devices simultaneously
- Access Point Mode - Supports AP Mode to transform your wired connection into wireless network, an ideal wireless router for home
| Approach | Short bursts | What happens to excess traffic | Boundary behavior | State and coordination considerations |
|---|---|---|---|---|
| Token bucket | Allows bursts up to the bucket capacity while limiting sustained traffic through a configured refill rate. | The algorithm alone does not decide whether excess requests are rejected, delayed, or queued. | No fixed-window reset boundary; burst allowance is governed by available tokens. | Requires maintaining token state for each key. Cross-replica consistency depends on how that state is stored and coordinated. AWS API Gateway documents token-bucket throttling for HTTP APIs, with configured rates and bursts treated as best-effort targets rather than guaranteed ceilings (AWS documentation). |
| Leaky bucket / traffic shaping | Can absorb or smooth bursts, depending on how the implementation handles incoming work. | Some implementations delay or queue excess work; others reject it. Check the gateway’s behavior rather than inferring it from the name. | Typically aims for a steadier output rate rather than relying on a fixed reset boundary. | Queueing adds waiting work and can increase latency; distributed enforcement still depends on the implementation. Apache APISIX describes its limit-req plugin as leaky-bucket based (APISIX algorithm overview). |
| Fixed window | Can allow a concentrated burst near a window boundary. | Whether excess is rejected or handled another way is implementation-specific. | A caller can use much of an allowance just before a window resets and again just after it; Kong illustrates this boundary effect (Kong window types). | Discrete counters are conceptually simple, but storage, reset timing, and replica coordination vary by product. |
| Sliding window | Reduces the sharp burst opportunity at a fixed-window reset. | Does not prescribe whether over-limit requests are rejected, delayed, or counted. | Evaluates usage across a moving interval, avoiding the same reset-boundary effect as a fixed window. | Accuracy, storage cost, and treatment of rejected requests vary among implementations. Kong documents sliding-window support and contrasts it with fixed windows (Kong window types). |
| Concurrency limit | Does not grant a request-per-time burst allowance; it caps simultaneous in-flight work. | Excess work may be rejected or otherwise controlled according to the product. | Not a window-based request counter. | Useful when work duration and resource occupancy vary substantially. APISIX documents concurrency control separately from its rate-limiting plugins (APISIX algorithm overview). |
Use a token bucket when the service can handle brief spikes but needs a sustained ceiling. Consider shaping when a downstream system benefits from a more regular flow and can tolerate queued work. Fixed windows are straightforward when occasional boundary bursts are acceptable; sliding windows are a better fit when those boundary bursts would undermine the policy. Add concurrency control when simultaneous work—not just request arrival rate—is the capacity problem.
How to design a limit that protects the right thing
- Identify the protected objective. Establish what the backend can safely sustain and identify routes whose work is unusually costly. Measure representative service behavior before setting a public requests-per-second value; there is no universal safe quota.
- Choose the counting key and scope. Common keys include account, API key, authenticated consumer, IP address, route, or service, and a policy can combine dimensions. IP-only controls can group unrelated people behind one shared address. Unauthenticated endpoints may still need IP- or network-level safeguards. Kong documents consumer, credential, IP, service, and route scopes; AWS API Gateway documents account, stage/method, and usage-plan client scopes (Kong rate limiting; AWS REST API throttling).
- Layer fairness controls with capacity safeguards. Per-consumer or per-route policies can limit abuse and distribute access, while an aggregate service or regional limit protects shared infrastructure. AWS describes multiple REST API throttling layers, including per-client or per-method usage-plan limits, stage/method limits, account throttles, and regional throttles, with documented precedence (AWS REST API throttling).
- Set sustained rate and burst capacity separately. The rate governs continuing demand; the bucket capacity determines how much work can arrive together. Tune burst size against queue depth, downstream concurrency, and the latency budget. AWS exposes refill rate and bucket capacity as separate token-bucket controls for HTTP APIs (AWS HTTP API throttling).
- Choose how replicas share state. Local counters are fast and avoid a shared-store round trip, but each replica can grant its own allowance, multiplying the effective limit as traffic spreads. Shared counters can make enforcement more consistent across replicas, at the cost of coordination latency and dependence on the state store. Kong documents Redis options for its rate-limiting plugins, but consistency guarantees and outage behavior depend on the chosen product, datastore, and configuration (Kong rate limiting).
- Specify the failure policy. Decide whether the limiter fails open (allowing requests) or closed (rejecting them) when its backing state is unavailable. Set timeouts, any fallback limits, and monitoring deliberately; this behavior is not standardized across products.
- Define the enforcement action. Decide whether traffic above the limit is rejected, delayed, or queued, and verify what the configured gateway actually does. Kong documents delayed-and-retried throttling as an optional capability in its advanced plugin; it is not an automatic consequence of choosing a particular algorithm (Kong rate limiting).
What should a client do with HTTP 429?
429 Too Many Requests signals that the client has exceeded a rate limit. When the response includes Retry-After, use it to determine when to try again instead of immediately resending the request. Slack documents this response for its HTTP-based APIs, with Retry-After expressing seconds until retry; its example value of 30 is illustrative, not a general waiting period or a universal API limit (Slack rate limits).
Rank #2
- 【Five Gigabit Ports】1 Gigabit WAN Port plus 2 Gigabit WAN/LAN Ports plus 2 Gigabit LAN Port. Up to 3 WAN ports optimize bandwidth usage through one device.
- 【One USB WAN Port】Mobile broadband via 4G/3G modem is supported for WAN backup by connecting to the USB port. For complete list of compatible 4G/3G modems, please visit TP-Link website.
- 【Abundant Security Features】Advanced firewall policies, DoS defense, IP/MAC/URL filtering, speed test and more security functions protect your network and data.
- 【Highly Secure VPN】Supports up to 20× LAN-to-LAN IPsec, 16× OpenVPN, 16× L2TP, and 16× PPTP VPN connections.
- Security - SPI Firewall, VPN Pass through, FTP/H.323/PPTP/SIP/IPsec ALG, DoS Defence, Ping of Death and Local Management. Standards and Protocols IEEE 802.3, 802.3u, 802.3ab, IEEE 802.3x, IEEE 802.1q
- Honor the supplied retry time. If many clients may retry together, add jitter so they do not all resume at once.
- Cap retry attempts and use backoff when no meaningful retry time is supplied.
- Check whether replay is safe before retrying. A 429 response alone does not establish that every operation can be repeated without duplicating side effects; use the API’s idempotency protections where available.
For API developers, return 429 for rate-limit rejections and include Retry-After when a meaningful retry time is known. Make the response actionable rather than forcing clients to guess.
How major gateways illustrate the differences
- AWS API Gateway HTTP APIs: AWS documents token-bucket throttling. It warns that configured throttles are best-effort targets, not guaranteed request ceilings; exceeding rate and burst targets can lead to 429 responses (AWS HTTP API throttling).
- AWS API Gateway REST APIs: AWS documents usage-plan client limits alongside stage/method, account, and regional throttling layers. These controls have documented precedence and should not be collapsed into one assumed quota (AWS REST API throttling).
- Kong Gateway: Its rate-limiting plugins apply to services, routes, and consumers. Available algorithms, Redis options, and delayed handling differ by plugin and configuration, so verify the relevant product version and settings (Kong rate limiting).
- Apache APISIX: Its overview maps
limit-reqto leaky bucket,limit-countto fixed or sliding windows, andlimit-connto concurrency control. Those mappings describe APISIX, not a universal naming convention (APISIX algorithm overview).
What to monitor after rollout
A configured limit is only useful if you can see how it behaves under real traffic. Monitor the outcomes and dependencies that reveal whether the policy is protecting the service without blocking normal use:
Rank #3
- Dual-band Wi-Fi with 5 GHz speeds up to 867 Mbps and 2.4 GHz speeds up to 300 Mbps, delivering 1200 Mbps of total bandwidth¹. Dual-band routers do not support 6 GHz. Performance varies by conditions, distance to devices, and obstacles such as walls.
- Covers up to 1,000 sq. ft. with four external antennas for stable wireless connections and optimal coverage.
- Supports IGMP Proxy/Snooping, Bridge and Tag VLAN to optimize IPTV streaming
- Access Point Mode - Supports AP Mode to transform your wired connection into wireless network, an ideal wireless router for home
- Advanced Security with WPA3 - The latest Wi-Fi security protocol, WPA3, brings new capabilities to improve cybersecurity in personal networks
- Allowed and rejected requests, broken down by route and limit key where practical.
- Key cardinality and saturation, to catch unexpectedly broad or rapidly growing scopes.
- Backend latency and resource pressure, including whether delayed work is building a queue.
- Rate-limit state-store health and coordination latency in multi-replica deployments.
- Actual enforcement versus configured targets. In AWS API Gateway, for example, throttling targets are best-effort rather than hard ceilings.
Adjust the rate, burst, scope, or action based on these observations rather than treating the first configured number as a permanent capacity guarantee.
Quick Recap
Best Value
- Next-Gen Gigabit Wi-Fi 6 Speeds: 2402 Mbps on 5 GHz and 574 Mbps on 2.4 GHz bands ensure smoother streaming and faster downloads; support VPN server and VPN client¹
- A More Responsive Experience: Enjoy smooth gaming, video streaming, and live feeds simultaneously. OFDMA makes your Wi-Fi stronger by allowing multiple clients to share one band at the same time, cutting latency and jitter.²
- Expanded Wi-Fi Coverage: 4 high-gain external antennas and Beamforming technology combine to extend strong, reliable, Wi-Fi throughout your home.
- Improved Battery Life: Target Wake Time helps your devices to communicate efficiently while consuming less power.
- Improved Cooling Design: No heat ups, no throttles. A larger heat sink and redefined case design cools the WiFi 6 system and enables your network to stay at top speeds in more versatile environments.
Rank #4
- DUAL-BAND WIFI 6 ROUTER: Wi-Fi 6(802.11ax) technology achieves faster speeds, greater capacity and reduced network congestion compared to the previous gen. All WiFi routers require a separate modem. Dual-Band WiFi routers do not support the 6 GHz band.
- AX1800: Enjoy smoother and more stable streaming, gaming, downloading with 1.8 Gbps total bandwidth (up to 1200 Mbps on 5 GHz and up to 574 Mbps on 2.4 GHz). Performance varies by conditions, distance to devices, and obstacles such as walls.
- CONNECT MORE DEVICES: Wi-Fi 6 technology communicates more data to more devices simultaneously using revolutionary OFDMA technology
- EXTENSIVE COVERAGE: Achieve the strong, reliable WiFi coverage with Archer AX1800 as it focuses signal strength to your devices far away using Beamforming technology, 4 high-gain antennas and an advanced front-end module (FEM) chipset
- OUR CYBERSECURITY COMMITMENT: TP-Link is a signatory of the U.S. Cybersecurity and Infrastructure Security Agency’s (CISA) Secure-by-Design pledge. This device is designed, built, and maintained, with advanced security as a core requirement.
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
