Five Patterns, Five Different Communication Problems
12 entriesA LinkedIn post by Nikki Siapno (31 October 2026) explains that request-response, message queues, publish-subscribe, streaming and batch processing all move data between systems, but each solves a different communication problem. Request-response is the simplest model: a caller sends a request and waits for a reply, which suits APIs that need an immediate answer but couples the request to the latency and availability of the responding service. With a message queue, a producer sends work to a queue for a consumer to process asynchronously, useful for background jobs, absorbing traffic bursts and decoupling services. In publish-subscribe, a publisher emits an event to a topic and multiple subscribers independently receive and react to it, useful when the same event must fan out across a system. In streaming, data flows continuously and consumers process it as it arrives; streaming platforms often retain events so consumers can process or replay them independently. Batch processing handles bounded datasets in groups, often on a schedule, suited to ETL, backfills, reports and large recomputations. The post's one-line summary is: request-response answers now, a message queue processes later, pub-sub notifies many, streaming processes continuously and batch processes in groups. It adds that real distributed systems rarely use just one pattern and combine them as each part needs. This page organizes standards, tools and design references around those five patterns. Descriptions are paraphrased from each source and are not rankings; the post's handbook and newsletter promotion is not included.
1. Request-Response: Answer Now
6 entriesA caller sends a request and waits for a reply, which couples the request to the responding service's latency and availability.
2. Message Queues: Process Later
6 entriesA producer sends work to a queue for a consumer to process asynchronously, for background jobs, traffic bursts and decoupling.
3. Publish-Subscribe: Notify Many
7 entriesA publisher emits an event to a topic, and multiple subscribers independently receive and react to it.
4. Streaming: Process Continuously
7 entriesData flows continuously, and platforms often retain events so consumers can process or replay them independently.
5. Batch Processing: Process in Groups
6 entriesJobs process bounded datasets in groups, often on a schedule, for ETL, backfills, reports and large recomputations.
6. Combining Patterns and Design References
8 entriesReal systems rarely use just one pattern; these references cover choosing and combining them.
Frequently Asked Questions
How do distributed systems communicate?
Mainly through five patterns the LinkedIn post this page follows describes: request-response, message queues, publish-subscribe, streaming and batch processing. Each solves a different communication problem.
What is request-response?
A caller sends a request and waits for a reply. It suits APIs that need an immediate answer, but it couples the request to the latency and availability of the service responding.
When should I use a message queue?
For background jobs, absorbing traffic bursts and decoupling services: a producer sends work to a queue and a consumer processes it asynchronously.
What is the difference between a queue and pub-sub?
A queue hands each piece of work to a consumer to process, while pub-sub lets a publisher emit an event to a topic that multiple subscribers receive independently, so one event fans out to many parts of a system.
How is streaming different from batch processing?
Streaming processes data continuously as it arrives and platforms often retain events for replay. Batch processes bounded datasets in groups, often on a schedule, for ETL, backfills, reports and large recomputations.
What is the quick way to remember them?
Request-response is answer now, a message queue is process later, pub-sub is notify many, streaming is process continuously, and batch is process in groups.
Do systems use only one pattern?
Rarely. The post says distributed systems combine these patterns depending on how each part needs to communicate.
How current is this page?
Links and release facts were checked on 1 November 2026. LinkedIn may ask visitors to sign in. Descriptions follow each source's own pages and are not rankings or guarantees.