借助Solace解决生产者发速超消费者的消息拥堵问题（仅需最新消息）

Absolutely! Solace has built-in features tailor-made for your use case where the consumer only cares about the most recent message and can safely discard older, unprocessed ones. Here's how to set it up to eliminate congestion entirely:

Key Feature to Leverage: Last Value Queues (LVQs) #

The star solution here is Solace's Last Value Queue (LVQ). This queue type is explicitly designed for scenarios where only the latest update matters—it automatically discards older messages as soon as a newer one arrives for the same "key".

• How it works: Configure the queue as an LVQ and define a Last Value Property (a message header or custom user property). Every time a new message with the same property value hits the queue, Solace replaces the existing message with this new one. This means the queue will never hold more than one relevant message at a time, so no backlog can build up.
• Perfect match for your case: Even though your producer sends messages twice as fast as your consumer can process them, the LVQ ensures the consumer only ever sees the latest message. Stale, unprocessed messages get dropped automatically before the consumer even has a chance to see them.

Additional Configuration Tips to Optimize #

To lock in congestion-free behavior:

• Use Non-Persistent Messages: Since you don't need to retain messages long-term (and can safely discard unprocessed ones), have your producer send non-persistent messages. This cuts down on broker overhead and speeds up message delivery.
• Tweak Consumer Acknowledgments: Set your consumer to use auto-acknowledge mode (or acknowledge messages immediately upon receipt if using manual ack). This tells the broker the consumer is ready for the next latest message right away, no need to wait for processing to finish.
• Exclusive Subscription: If you only have one consumer for this topic, use an exclusive subscription to the LVQ. This removes unnecessary routing overhead and ensures messages go directly to your consumer.

Example Workflow for Your Exact Numbers #

Let's walk through how this plays out with your rates:

1. Producer sends a message every 5 microseconds to the target topic.
2. The LVQ receives each new message, checks the Last Value Property, and replaces the older message in the queue with the new one.
3. Your consumer picks up the latest message and starts processing it (taking 10 microseconds).
4. While processing is underway, 2 more messages arrive (10 microseconds ÷ 5 microseconds per message = 2). The LVQ replaces the queue's message twice, so when the consumer finishes processing, the newest message is already waiting.
5. The consumer immediately grabs this latest message, and the cycle repeats—no backlog ever forms.

So yes, Solace absolutely can prevent message congestion here, thanks to Last Value Queues and supporting configuration options.

内容的提问来源于stack exchange，提问作者Sergey S