Top 10 Criteria for Selecting a MQTT Broker

Top 10 Criteria for Selecting a MQTT Broker

author Ian Skerrett

Written by Ian Skerrett

Category: HiveMQ

Published: August 19, 2019


MQTT has become the IoT standard for connecting devices to the cloud. It is clear that MQTT’s publish/subscribe protocol is the best way for connecting devices over unreliable networks. Using HTTP oftens leads to unreliable data transfer, lost messages and unresponsive devices waiting to resynchronise with the cloud.

The good news is there are a number of MQTT solutions available in the market. However, as with any new technology space not all MQTT brokers are equal. For this reason, we have put together a top 10 criteria for selecting an MQTT broker provider.

  1. Does the MQTT solution support the MQTT 3.1, 3.1.1 and MQTT 5 specification? Some providers have been known to say they support MQTT but only really support a subset (references: 1, 2).

  2. Does the MQTT broker support high availability? How is the broker deployed in a cluster environment? Does the broker support automatic elastic and linear scalability at runtime? How does it recover from network splits?

  3. How does the MQTT solution support disaster recovery? Does the broker persist to disk and not keep data in memory? If one node goes down, are all the sessions still available? If all nodes go down, are there still all session and undelivered messages? Is there a backup and restore feature?

  4. Does the broker perform under heavy load of publish and subscribe? Make sure you benchmark the broker to match your production scenario. Does the MQTT solution have a limit on the number of topics that are supported? Ask for benchmark data from the vendor.

  5. What type of metrics does the broker produce for monitoring? Can you get live stats based on the number of connections, CPU/memory usage? Is it easy to integrate the metrics into dashboard solutions like Grafana, Prometheus, AppDynamics?

  6. What type of management tools are included? Is it possible to query and manage individual client connections? Are there debug tools to create trace recordings to isolate misbehaving clients? Do these features scale-up to work with millions of client connections?

  7. How robust is the security solution? Is TLS/SSL supported natively? Is it possible to integrate device authentication and authorization systems into the MQTT broker? What type of logging and auditing is provided? Is there a role-based permission system built into the solution?

  8. Can you write extensions for the MQTT broker? How robust is the API? Is there documentation for the SDK? Does the API support hot reload? Is there a marketplace or catalog of extensions?

  9. Who else is using the MQTT broker? Does the supplier have business critical use cases that will act as references?

  10. What type of support is available? Is there a company willing to provide 24 / 7 technical support for your deployment? Is the documentation complete and robust?


Not all MQTT brokers are the same. At HiveMQ, we have been collaborating with large and small companies to build an enterprise ready MQTT platform. We know that to successfully use MQTT in an IoT application, you need to have an MQTT platform you can trust. Feel free to download an evaluation copy of HiveMQ Enterprise and start evaluating HiveMQ based on the top 10 criteria.


author Ian Skerrett

About Ian Skerrett

Ian was Vice President of Marketing for HiveMQ until September 2022.

mail icon Contact HiveMQ
newer posts MQTT Topics, Wildcards, & Best Practices – MQTT Essentials: Part 5
HiveMQ ESE 1.2 released older posts