Azure Cosmos DB: Paper about the TCO of running a NoSQL database is available
Posted on Saturday, November 19, 2016
Azure Cosmos DB Principal Engineering Manager Kirill Gavrylyuk recently published the paper The Total Cost of (Non) Ownership of a NoSQL Database Cloud Service. The paper compares the total cost of ownership (TCO) of running a NoSQL database in the following scenarios:
- An OSS NoSQL database like Cassandra or MongoDB hosted on-premises
- An OSS NoSQL database hosted on Azure Virtual Machines
- A managed NoSQL database as a service, such as Azure Cosmos DB
The paper uses the same scenarios, parameters, and assumptions used in a paper previously published by Amazon, and it yields similar conclusions. The main finding is that using a managed NoSQL cloud database is five to ten times more cost effective than the alternatives. In addition, a new finding in this paper is that Azure Cosmos DB is about 10 percent cheaper than DynamoDB due to the lower cost of write requests.
A qualitative comparison between Azure Cosmos DB, DynamoDB, and Cassandra also yielded the following findings:
- The TCO of Azure Cosmos DB is comparable to that of OSS Cassandra running on Azure D14v2 VMs for scenarios that involve high, sustained, predominantly write workloads with low storage needs.
- If more storage is needed, or the workload involves a balanced read/write mix, or the workload is bursty, the TCO of Azure Cosmos DB can be up to four times lower than that of OSS Cassandra running on Azure VMs.
- Azure Cosmos DB is up to three times cheaper than DynamoDB for the high-volume workloads examined.
Read the Azure blog post now, or jump right into the paper The Total Cost of (Non) Ownership of a NoSQL Database Cloud Service.