What does IT logging have to do with the internet of things? As companies such as Splunk, Datadog, and Solarwinds recognized the opportunity in the internet of things, they started shifting their marketing copy from emphasizing their ability to log server data to logging machine data with the hope of bringing their logging experience into a new market. Circonus is trying to make that same transition, and so far it seems to have had some success.
The 10-year-old company was created by Theo Schlossnagle, who I know from my past life as a cloud reporter. He created Circonus to manage, monitor, and analyze website and server data for “web-scale” applications. This year, the company brought in Bob Moul as CEO. Moul is the former CEO of Boomi, a cloud integration startup that was bought by Dell.
At Circonus, Moul is going to help sell to enterprise customers, ideally those focused on machine data that isn’t just coming off of servers. One of the tropes of the IoT is that the data machines give off comes quickly, is diverse, and can be incredibly dense (this is usually referred to as data volume, velocity, and variety). Moul says Circonus can handle this data because it has built a dedicated stream processing language called CAQL.
He says the software can ingest a trillion data points a second, which is pretty impressive. Customers include ad networks, which take in millions of data points a second to create true real-time auctions, and Major League Baseball, which is using the API to pull stats on players and show those stats on stadium screens as they walk up to the plate to bat or pitch. These use cases don’t actually count as IoT in my book, but the use cases for rapidly ingesting and analyzing of a lot of data are there.
Stream processing is the analysis of data in close to real time, which is valuable for use cases such as predictive maintenance, factory automation, or traffic analytics. The Circonus software consists of a local agent that runs on a data-gathering device, and software that can run on the cloud or on-premise in a server or gateway.
When it comes to the edge, the Circonus agent is fairly lightweight. It can’t run on a microcontroller, but it could run on a Raspberry Pi with a single core and 100 Mb of memory, which means it could end up on a gateway or device that aggregates sensor readings. As for the proprietary programming language, Circonus supplies an API but also has a module that can convert CAQL to Python or R.
All of which means we could see customers put the software on physical devices that gather data and then send that data to the cloud or other software. And that’s what Circonus is hoping for.