michelangelus - Fotolia
The much vaunted internet of things could be too much of too many good things -- at least for databases. That is because IoT data from remote sensors can arrive too rapidly for traditional databases to handle, especially if the objective is to quickly act upon that data.
To handle the firehose of incoming sensor readings, Roomonitor, a company that monitors and controls air conditioners in hotels in 17 cities across Europe, recently decided to go the NewSQL route. For that purpose, Roomonitor selected CrateDB from Crate.io.
Along with the ability to scale database clusters to handle large amounts of incoming data, CrateDB's support of SQL was a factor in its selection, according to Ignacio Suarez, CEO of Barcelona-based Roomonitor. That is because SQL support simplifies the work of developers who build queries that scan IoT data.
"We needed something that could quickly store and access large amounts of data," Suarez said, who described CrateDB as "purpose-built for IoT."
Suarez said the Roomonitor AC climate-control system grew out of earlier work that the company did to help communities deal with noise issues encountered in densely populated European cities, where air conditioners, among other things, can disturb the peace.
He said Roomonitor takes readings from sensors that monitor activity, and it detects patterns in that activity. Now, somewhat like the Wi-Fi-enabled Nest thermostats that have gained some attention in the U.S., the Roomonitor system employs the data it gathers to adjust air conditioners, reducing both noise and overuse, and in turn reduces the costs of cooling rooms.
Suarez indicated the Roomonitor system stores data in CrateDB in JSON format -- a favored data type for both NoSQL and NewSQL implementations.
He said Roomonitor's technical team reviewed other alternatives for working with JSON, including MongoDB. But ease of scaling and simpler SQL development gave an edge to CrateDB for the monitoring and control system, he said.
CrateDB, which first became generally available last year, enters a varied -- and very crowded -- field. That field includes NoSQL players ranging from startups, like Couchbase and MongoDB, to established names like IBM and Oracle, as well as NewSQL players, such as MemSQL, NuoDB and VoltDB.
This year, CrateDB has updated its database, launching CrateDB 2.0 and CrateDB Enterprise Edition. The database adds a new index structuring architecture to speed queries on IP address, range and geospatial queries, and SQL improvements that allow users to execute more complex queries, according to Andy Ellicott, chief marketing officer for Crate.io, based in San Francisco.
Ellicott said IoT applications like Roomonitor's are a prime use for CrateDB. He described the IoT data type as "data created by machines, not people."
He suggested that IoT growth is rooted in strong ground, because so much modern machinery already includes sensors. The trick, he said, is to make that data more accessible, even as it becomes more widely prevalent.
Today's firehose of data could be just beginning. Analyst group Gartner has estimated that 20.4 billion IoT units will be connected by 2020. Going forward, NewSQL offerings like CrateDB could play a key role in handling that gush of new data.
Find out about data ingestion combining MemSQL and Kafka
How AI will lead to more use of DevOps in data management