Microsoft has unveiled a raft of updates to its Event Grid platform, which it says will allow users to build a far broader and higher powered, range of event-driven and serverless applications.

Event Grid, launched in 2017, is an event routing service that allows a resource or application to publish an event, which is picked up by Event Grid, and then routed to an appropriate application or service such as an Azure function. Microsoft says removing the need for services to continually poll the network looking for event triggers, greatly reduces the demand for compute and network resources.

In a blog post outlining the changes, Bahram Banisadr Program Manager for Event Grid, said the service’s integration with the Azure IoT Hub has been expanded to include device telemetry events. The integration is in public review, though not in the East and West US or West Europe regions. Microsoft said this will ensure easier integration between data from IoT devices and applications, including Azure Functions of Azure Logic Apps, or non Azure services using webhooks.

Microsoft also announced general availability of Event Domains, which it describes as a way to “organize your entire eventing infrastructure under a single construct, set fine grain auth rules on each topic for who can subscribe, and manage all event publishing with a single endpoint.” Put simply, they should handle much of the management of complex event-driven architectures.

There are standard limits though, with each event domain allowed a maximum of 100,000 topics per Event Domain, 100 Event Domains per Azure sub, and a max of 5,000 events per second per Event Domain.

Microsoft said “We also acknowledge that advanced event-driven architectures don’t always fit in the confines of 64 KB.” So, it has launched a public preview of events of up to 1MB.

Should you be nervous about the possibility of a service interruption taking out your event architecture, Event Grid “now has built-in automatic geo disaster recovery (GeoDR) of metadata, applicable to all existing Domains, Topics and Event Subscriptions”.

Should an outage take out an entire region, customers’ eventing infrastructure metadata is already synced to a paired region, and “new events will begin to flow again with no intervention from your side required, avoiding service interruption automatically.”