MongoDB announces new search, cloud and edge capabilities

MongoDB is taking aim at developer silos, says senior product director Mat Keep

MongoDB, the producer of database software and services built around the eponymous document store, has announced the general availability of a number of new features in the database and the MongoDB Cloud platform and the availability of the Realm mobile database.

MongoDB users can now create unions between collections of different structures, so long as they have a common field to act as a key. "Union operates across multiple servers so you can start to explore very large datasets in new ways," said Mat Keep, senior director, product & solution. Anticipated use cases include IoT and trading applications that create ‘buckets' of time-series data but which also require schema changes from time to time.

Shard keys (indexed fields that determine the distribution of a collection's documents across the cluster's shards) can now be altered without having to take the database down. "If your application changed in ways you hadn't anticipated, you couldn't go back and change that shard key, what you'd have to do is dump the data, and then reload it with a new shard key", said Keep. "With 4.4 you can change the shard key and it does that without downtime."

A third major change is called ‘hedged reads'. This is designed to deal with latency caused when a query gets sent to a node that's busy or faulty, by instead distributing the query to multiple nodes simultaneously and then returning the quickest response to the user.

The MongoDB Cloud database platform sees the Atlas Data Lake released for general consumption along with new search capabilities. The data lake will integrate data stored in AWS S3 with other services to allow for data tiering, with ‘colder' data stored where it's cheaper. A new feature called Federated Query allows queries to run across both hot and cold data in their different tiers.

There are other developments in search too.

"Developers work with data in much more than just the database, they also work with it in search. So, they're trying to build Google-like search experiences into their apps," said Keep. Atlas Search, which is now generally available, is designed to meet this need.

In combination, these cloud features are designed to make life easier for application developers by reducing the number of separate data sources and APIs they need to deal with, said Keep.

"From a developer standpoint, cloud hasn't really moved the needle all that much because they still have to use all these different databases, data lakes, data models and APIs. MongoDB Cloud is designed to address that because you have a flexible data model and you can query all of that data with a consistent API. So, it's a way of really compressing down all of these different silos, all of these different APIs, into something that is much more unified."

The Realm mobile database, the fruits of an acquisition last year, is also available together with a synchroniser to automatically keep data up-to-date across devices and with the backend. This represents MongoDB's first real move to the edge, Keep said.