At the June MongoDB World event in New York City, the company announced a number of significant innovations that attempt to address these concerns.
This aggregation framework now also supports the definition of read-only views. Views can restrict the attributes or documents available to nominated users, providing the equivalent of row-level security that is commonplace in relational databases.
In addition, there are enhancements to MongoDB sharding—the standard means of scaling MongoDB beyond a single writable server. The improvements provide for the establishment of zones, which allows data to be routed to the nodes of the cluster located in local data centers, or allows data to be tiered so that hot/recent data can be placed on systems with an expensive solid-state disk, while colder, older data can be placed on systems with a cheaper disk.
MongoDB has provided cloud-based monitoring, backup, and provisioning services for several years. At MongoDB World, the immediate availability of a database as a service (DBaaS) offering was announced. This service—named Atlas—provides a cloud-based, on-demand MongoDB database cluster with configurable sharding, replication, and server sizing. Atlas runs initially only on Amazon AWS but will also be available eventually on Azure and Google Cloud. A small, three-node replica set cluster (the effective minimum high availability solution) costs around $300/month.
As with most DBaaS offerings, the service offers rapid deployment and a reduction in administrative overhead. However, there are potential performance issues for applications that are not themselves resident in the same cloud, as well as some concerns about security, though the Atlas security features seem pretty robust. Finally, a cloud-based database may cost more in the long run—at least with respect to simple hardware costs—when compared to on-premise hardware.
As with many open core software companies, MongoDB relies on sales of commercial tools to generate revenue. For the long-term success of open source databases such as MongoDB, these business models must succeed. MongoDB has been an amazing success story from an adoption standpoint, and I hope the company can establish an enduring revenue stream—perhaps based on initiatives such as Atlas and Compass—to support the continuing evolution of the technology.