Understanding LVQ Behavior Under High Message Rates

General Discussions
1

Hi Team,

Today, we had a great discussion during Solace Office Hours.

I would like to better understand the behavior of a Last Value Queue (LVQ) under high message rates. I have observed that multiple messages get queued in the LVQ when the message rate is high. I am attaching screenshots of the queue details for reference.

Additionally, I am interested in understanding how the LVQ behaves when a queue browser or consumer is present. Specifically:

  • Is there a possibility that an older message could be delivered in such a scenario?
  • How does the LVQ handle message replacement and retention under high throughput conditions?

Test Setup:

  • 1 Standard Queue and 1 LVQ subscribing to the same topic
  • Using sdkperf to publish messages
  • Running on the Evaluation Edition of the Solace Software Broker

Would love to explore and gain deeper insights into this behavior. Looking forward to your thoughts!

image
image1600×778 97.2 KB

image
image1156×373 11.9 KB

image
image902×902 30.8 KB

2

Hi there, I’m Solly AI, I generate responses to your questions using the power of the Solace Agent Mesh:

Please remember that AI technology can make mistakes so ensure that you read our Disclaimer.

AI Generated responses are a work in progress. If this response doesn’t make sense for your topic, please react to it with a :shaking_face:.

Let me analyze the behavior of Last Value Queues (LVQs) based on the provided documentation:

The key points about LVQ behavior are:

  1. Message Retention:
  • An LVQ is configured by setting the queue’s maximum spool size to 0
  • It only stores the last message received, regardless of message priority
  • For partitioned queues, each partition holds the last message spooled to that partition
  1. Message Replacement:
  • When a new message arrives, it replaces the older message in the queue
  • The replacement happens regardless of message priority values
  • For partitioned queues, replacement happens within each partition independently
  1. Message Delivery:
  • Like other queues, messages are removed from the LVQ when acknowledged by consumers
  • Queue browsers can view messages without consuming them to maintain the last message
  • Message priority is not respected for queue browsers - messages are delivered in publication order
  • Message priority is not respected for LVQs in general - the last message is kept regardless of priority
  1. High Message Rate Considerations:
  • The documentation does not explicitly address behavior under high message rates
  • Since LVQs only maintain the last message, high rates would result in frequent message replacement
  • Order of delivery to consumers would follow normal queue delivery rules, except priority is not respected

Relevant sources: