HiveMQ Cloud aims to provide a fully-managed platform for IoT applications using MQTT to stream data out of devices. HiveMQ Cloud features automatic scalabiilty and a dedicated infrastructure for each customer at a fixed hourly rate.
According to HiveMQ, one of the big advantages of using HiveMQ Cloud is its fully compliant MQTT implementation, including MQTT 5. This contrasts with other providers of IoT platforms such as Amazon, Google, and Microsoft, which only provide partial implementation of the protocol. Not being fully standards-compliant has the downside of subtly locking IoT devices into the concrete implementation provided by the vendor, says HiveMQ.
For IoT applications, the risk of vendor lock-in becomes greater since IoT devices are typically deployed outside of the enterprise, often have long product life cycles and often the sheer number of deployed devices is significant.
Besides the capability of scaling up and down according to telemetric load, HiveMQ Cloud ensures each MQTT cluster gets a dedicated infrastructure in three AWS data centers. This prevents different companies from sharing the same infrastructure, and guarantees predictable performance.
InfoQ has spoken with HiveMQ CTO Dominik Obermaier to learn more.
InfoQ: Could you please share more details about how HiveMQ Cloud implements scalability and reliability?
Dominik Obermeier: We have built HiveMQ Cloud for customers who want to go into production with a business critical IoT application that requires high availability and scalability. For this reason, each customer is set up with an infrastructure of dedicated EC 2 instances. We create three HiveMQ broker cluster nodes in an AWS region that run in three different availability zones. The infrastructure also includes two dedicated load balancers running on EC 2 instances to ensure high availability. We also plan to offer a similar architecture on MS Azure and GCP.
The architecture of HiveMQ is based on a masterless architecture that allows us to create autonomous clusters that can share the MQTT client connections with no impact on latency for the end user. For instance, if a single broker node fails in the cluster, the MQTT client connections are automatically re-assigned to another broker node. The MQTT clients don't need to negotiate a reconnect with another broker instance. We have more details about HiveMQ broker clustering on our website.
InfoQ: About the pricing, while it is true that it is simple and predictable, it may appear too rigid for some kinds of customers, or under-funded projects. Do you have you any plans to introduce more flexible options?
Obermeier: The first release of HiveMQ Cloud is focused on production deployments of companies that require a high degree of reliability and scalability. This is based on our experience working with some of the largest HiveMQ customers. In the near future, we do plan to offer HiveMQ Cloud for companies that don't require dedicated infrastructure for high availability. We will also offer a free trial so developers can create proof of concepts before moving to production.
InfoQ: Could you explain what facilities HiveMQ Cloud provides for integration with other services, which would be useful for consuming the data streamed from the devices?
Obermeier: We believe the IoT industry needs to make it a lot easier to integrate IoT data with different enterprise systems. There needs to be a free flow of bi-directional data movement between IoT devices and services such as Kafka, MongoDB, InfluxDB, Machine Learning services, and so on. To enable this free flow of data, we need to make it easy to integrate different cloud services with HiveMQ Cloud. Our plan is to first focus on some of the best of breed services, such as Confluent Cloud, MongoDB Atlas and InfluxDB Cloud. We will also integrate with some of the more popular services on AWS, Azure and GCP.
We also plan to allow customers to create their own integrations using the HiveMQ Extension SDK. We really see an ecosystem of integrations that customers can leverage to move their IoT data between different cloud services.
If you are interested in taking a tour of HiveMQ Cloud, do not miss this introductory video.