Item: Azure Cosmos DB
Rating: 10
Author: Lars Kemmann

Use Cases and Deployment Scope

We use Azure Cosmos DB as our preferred NoSQL store for custom application development for our clients. It solves many of the onboarding hurdles for graph databases in particular, and the automatic indexing features make it ideal for a variety of scenarios across document and table storage as well. We rely on Cosmos DB when we need high availability, globally-distributed access with very low latency, and always-ready compute over data. In the past two years, I've found our team moving away from relational databases more and more as we're discovering ways to apply NoSQL much more cost-effectively than what could be done with an RDBMS.

Pros and Cons

Turn-key geo-redundancy with multi-master writes is unprecedented and unparalleled in the industry!
Guaranteed low latency makes Cosmos DB an excellent fit for most of our performance-intensive situations.
The tunable consistency model simplifies so many challenges in distributed systems engineering that otherwise require advanced knowledge of computer science topics. I continue to be impressed at how Cosmos DB has abstracted away so much complexity.

Cosmos DB can be very expensive if you're using it for scenarios that are better completed in regular old Azure Table Storage or Blob Storage, specifically if you put some thought into your partitioning schemes. No product is a good substitute for thoughtful system design.
It would be helpful if I had some more insight into how many resources (DTUs) an individual query uses.
The auto-indexing is great, but a little mysterious -- not usually an issue but it does require some intentional thought.

Return on Investment

Tremendously reduced our time-to-market with solutions built on graph databases
Helped our team to think more broadly about data by making NoSQL options much more accessible
Saved us significant effort on geo-replication and global availability

Alternatives Considered

MongoDB, CouchDB, Cassandra and Neo4j

Cosmos DB is unique in the industry as a true multi-model, cloud-native database engine that comes with solutions for geo-redundancy, multi-master writes, (globally!) low latency, and cost-effective hosting built in. I've yet to see anything else that even comes close to the power that Cosmos DB packs into its solution. The simplicity and tooling support are nice bonus features as well.

Other Software Used

Azure SQL Database, Azure Stream Analytics, Azure Active Directory, Microsoft Visual Studio Team System, Microsoft Visual Studio Code, Visual Studio IDE

Likelihood to Renew

Cosmos DB has such a head start and unique position in the market, backed as it is by Microsoft's global backbone network, that it's hard to imagine any other solution coming close to it in terms of overall capabilities. While it's not the right fit for every situation, whenever Cosmos DB is a candidate it should probably be your first choice.

Likelihood to Recommend

Cosmos DB is hands-down the most flexible and performant way to store and access data related to the functionality of your applications. If you don't really need a 3rd-normal-form relational schema, and if you're not working on a (hot or cold) big data analytics scenario, then you should almost definitely be using Cosmos DB! The only major exception is if your use case can be addressed with some thoughtful planning and use of plain old Azure Storage, which might not get a lot of attention but is still a rock-solid platform.

A one-of-a-kind solution to more problems than you probably realize!

Overall Satisfaction with Azure Cosmos DB