{
  "id": 9112445,
  "title": "SQS Fair Queues: one loud customer no longer ruins your day",
  "url": "https://urgent.news/2026/09/22/sqs-fair-queues-one-loud-customer-no-longer-ruins-your-day",
  "topic": "tech",
  "section": "Tech",
  "published": "2026-09-22T08:40:04.000Z",
  "source": {
    "name": "Dev.to",
    "slug": "dev-to",
    "url": "https://dev.to/ohadtutay/sqs-fair-queues-one-loud-customer-no-longer-ruins-your-day-332i"
  },
  "original_language": "en",
  "account": "SaaS providers often use a single SQS queue to manage messages from all their customers. This works fine until one customer sends a large number of messages at once, causing a delay for other customers. This is known as the noisy neighbor problem. AWS introduced fair queues in July 2025 to address this issue.\n\nUnder a standard queue, messages are processed in the order they arrive, regardless of who sent them. If one tenant floods the queue, other tenants have to wait behind the queue, increasing the dwell time for their messages. SQS fair queues aim to solve this problem by distributing messages from different tenants more evenly.\n\nTo set up a fair queue, you need to include a MessageGroupId in your SendMessageCommand. This group ID should be a meaningful identifier, such as a customer ID or request type. After setting the group ID, SQS will distribute messages from different tenants more fairly. No new queue types are required, and no changes are necessary on the consumer side.\n\nThe fair queue system monitors the distribution of messages across tenants. If a tenant holds more than 10% of in-flight messages and has at least 30 in-flight messages, it is considered noisy. The system also considers the recent share of consumer processing time. Once a tenant's share of in-flight messages drops below 10%, or if there is no in-flight message for five continuous minutes, it is no longer considered noisy.\n\nTo monitor the fair queue feature, AWS provides five metrics: ApproximateNumberOfNoisyGroups, ApproximateNumberOfMessagesVisibleInQuietGroups, and ApproximateAgeOfOldestMessageInQuietGroups. By graphing the approximate number of messages visible against the approximate number of messages visible in quiet groups, you can see when the feature is working effectively.",
  "summary": "If you run a SaaS on AWS, chances are all your customers share one SQS queue. It's far less to manage than a queue per tenant. Until one customer dumps 200k messages at 9am and everyone else's jobs sit there for twenty minutes. That's the noisy neighbor problem. AWS shipped fair queues in July 2025 to deal with it. What goes wrong today A standard queue doesn't care who sent a message. Ordering…",
  "key_points": [],
  "editors_take": null,
  "illustration": null,
  "coverage": {
    "outlets": 1,
    "also_reported_by": []
  },
  "ai_generated": true,
  "disclaimer": "Summaries, key points and the editor’s take are written by software from other outlets’ reporting and may contain errors — always check the linked original."
}