|
| 1 | +--- |
| 2 | +title: AI Rate Limiting |
| 3 | +keywords: |
| 4 | + - Apache APISIX |
| 5 | + - API Gateway |
| 6 | + - Plugin |
| 7 | + - ai-rate-limiting |
| 8 | +description: The ai-rate-limiting plugin enforces token-based rate limiting for LLM service requests, preventing overuse, optimizing API consumption, and ensuring efficient resource allocation. |
| 9 | +--- |
| 10 | + |
| 11 | +<!-- |
| 12 | +# |
| 13 | +# Licensed to the Apache Software Foundation (ASF) under one or more |
| 14 | +# contributor license agreements. See the NOTICE file distributed with |
| 15 | +# this work for additional information regarding copyright ownership. |
| 16 | +# The ASF licenses this file to You under the Apache License, Version 2.0 |
| 17 | +# (the "License"); you may not use this file except in compliance with |
| 18 | +# the License. You may obtain a copy of the License at |
| 19 | +# |
| 20 | +# http://www.apache.org/licenses/LICENSE-2.0 |
| 21 | +# |
| 22 | +# Unless required by applicable law or agreed to in writing, software |
| 23 | +# distributed under the License is distributed on an "AS IS" BASIS, |
| 24 | +# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied. |
| 25 | +# See the License for the specific language governing permissions and |
| 26 | +# limitations under the License. |
| 27 | +# |
| 28 | +--> |
| 29 | + |
| 30 | +## Description |
| 31 | + |
| 32 | +The `ai-rate-limiting` plugin enforces token-based rate limiting for requests sent to LLM services. It helps manage API usage by controlling the number of tokens consumed within a specified time frame, ensuring fair resource allocation and preventing excessive load on the service. It is often used with `ai-proxy` or `ai-proxy-multi` plugin. |
| 33 | + |
| 34 | +## Plugin Attributes |
| 35 | + |
| 36 | +| Name | Type | Required | Description | |
| 37 | +| ------------------------- | ------------- | -------- | --------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------------- | |
| 38 | +| `limit` | integer | false | The maximum number of tokens allowed to consume within a given time interval. At least one of `limit` and `instances.limit` should be configured. | |
| 39 | +| `time_window` | integer | false | The time interval corresponding to the rate limiting `limit` in seconds. At least one of `time_window` and `instances.time_window` should be configured. | |
| 40 | +| `show_limit_quota_header` | boolean | false | If true, include `X-AI-RateLimit-Limit-*` to show the total quota, `X-AI-RateLimit-Remaining-*` to show the remaining quota in the response header, and `X-AI-RateLimit-Reset-*` to show the number of seconds left for the counter to reset, where `*` is the instance name. Default: `true` | |
| 41 | +| `limit_strategy` | string | false | Type of token to apply rate limiting. `total_tokens`, `prompt_tokens`, and `completion_tokens` values are returned in each model response, where `total_tokens` is the sum of `prompt_tokens` and `completion_tokens`. Default: `total_tokens` | |
| 42 | +| `instances` | array[object] | false | LLM instance rate limiting configurations. | |
| 43 | +| `instances.name` | string | true | Name of the LLM service instance. | |
| 44 | +| `instances.limit` | integer | true | The maximum number of tokens allowed to consume within a given time interval. | |
| 45 | +| `instances.time_window` | integer | true | The time interval corresponding to the rate limiting `limit` in seconds. | |
| 46 | +| `rejected_code` | integer | false | The HTTP status code returned when a request exceeding the quota is rejected. Default: `503` | |
| 47 | +| `rejected_msg` | string | false | The response body returned when a request exceeding the quota is rejected. | |
| 48 | + |
| 49 | +## Example |
| 50 | + |
| 51 | +Create a route as such and update with your LLM providers, models, API keys, and endpoints: |
| 52 | + |
| 53 | +```shell |
| 54 | +curl "http://127.0.0.1:9180/apisix/admin/routes" -X PUT \ |
| 55 | + -H "X-API-KEY: ${ADMIN_API_KEY}" \ |
| 56 | + -d '{ |
| 57 | + "id": "ai-rate-limiting-route", |
| 58 | + "uri": "/anything", |
| 59 | + "methods": ["POST"], |
| 60 | + "plugins": { |
| 61 | + "ai-proxy": { |
| 62 | + "provider": "openai", |
| 63 | + "auth": { |
| 64 | + "header": { |
| 65 | + "Authorization": "Bearer '"$API_KEY"'" |
| 66 | + } |
| 67 | + }, |
| 68 | + "options": { |
| 69 | + "model": "gpt-35-turbo-instruct", |
| 70 | + "max_tokens": 512, |
| 71 | + "temperature": 1.0 |
| 72 | + } |
| 73 | + }, |
| 74 | + "ai-rate-limiting": { |
| 75 | + "limit": 300, |
| 76 | + "time_window": 30, |
| 77 | + "limit_strategy": "prompt_tokens" |
| 78 | + } |
| 79 | + } |
| 80 | + }' |
| 81 | +``` |
| 82 | + |
| 83 | +Send a POST request to the route with a system prompt and a sample user question in the request body: |
| 84 | + |
| 85 | +```shell |
| 86 | +curl "http://127.0.0.1:9080/anything" -X POST \ |
| 87 | + -H "Content-Type: application/json" \ |
| 88 | + -d '{ |
| 89 | + "messages": [ |
| 90 | + { "role": "system", "content": "You are a mathematician" }, |
| 91 | + { "role": "user", "content": "What is 1+1?" } |
| 92 | + ] |
| 93 | + }' |
| 94 | +``` |
| 95 | + |
| 96 | +You should receive a response similar to the following: |
| 97 | + |
| 98 | +```json |
| 99 | +{ |
| 100 | + ... |
| 101 | + "model": "deepseek-chat", |
| 102 | + "choices": [ |
| 103 | + { |
| 104 | + "index": 0, |
| 105 | + "message": { |
| 106 | + "role": "assistant", |
| 107 | + "content": "1 + 1 equals 2. This is a fundamental arithmetic operation where adding one unit to another results in a total of two units." |
| 108 | + }, |
| 109 | + "logprobs": null, |
| 110 | + "finish_reason": "stop" |
| 111 | + } |
| 112 | + ], |
| 113 | + ... |
| 114 | +} |
| 115 | +``` |
| 116 | + |
| 117 | +If rate limiting quota of 300 tokens has been consumed in a 30-second window, the additional requests will all be rejected. |
0 commit comments