Amazon Managed Apache Cassandra Service stokes competition

Vendors in the open source Cassandra NoSQL database market voice their concerns about AWS' entry into the field, as the cloud giant looks to expand its database portfolio.

Amazon Web Services' recent unveiling of its Amazon Managed Apache Cassandra Service did not go unnoticed by other vendors active in the open source Apache Cassandra market.

"The AWS announcement is further validation of the reach and growing importance of Cassandra," said Jonathan Ellis, co-founder and CTO of DataStax, one of the largest vendors in the Cassandra market. He added that enterprises in every industry have seen big returns from their investments in Cassandra, an open source database uniquely designed to meet the needs of hybrid and multi-cloud deployments.

As an open source project, there are multiple vendors that have commercially supported services for Cassandra. With the addition of Cassandra, AWS now has the most complete set of database management systems (DBMSes) in the cloud, according to Donald Feinberg, a vice president and distinguished analyst at Gartner. In his view, Cassandra was the only major missing DBMS in the AWS portfolio prior to the announcement at the re:Invent 2019 conference this month.

"Not only did they add Cassandra, but it is the only serverless Cassandra available in the cloud. This makes managing a Cassandra cluster easy to manage," Feinberg said. "The most difficult part of managing Cassandra is the multiple cluster environment, and serverless removes that obstacle."

Debate over the architecture

Instaclustr provides hosted and managed Cassandra as a service. Ben Slater, chief product officer at the company, called AWS' entry a validation for Apache Cassandra, and said he expects to see continued growth and interest in his company's offerings as a result. However, in his view the Amazon Managed Apache Cassandra Service isn't actually managed Cassandra, he added.

AWS CEO Andy Jassy announces Amazon Managed Apache Cassandra Service at re:Invent 2019.
AWS CEO Andy Jassy announces Amazon Managed Apache Cassandra Service at re:Invent 2019.

"It is an AWS proprietary version of a portion of the Apache Cassandra codebase running as a translation layer over DynamoDB. The service is only partially compatible with Apache Cassandra," Slater said.

The most difficult part of managing Cassandra is the multiple cluster environment, and serverless removes that obstacle.
Donald FeinbergVice president, Gartner

AWS disagrees -- according to Herain Oberoi, AWS marketing general manager for database, analytics and blockchain, Amazon Managed Cassandra Service uses the Apache Cassandra code.

"We modified Apache Cassandra to enable us to run it as a highly scalable, serverless and managed offering," Oberoi said.

Oberoi explained that AWS built Amazon DynamoDB on its own to provide customers with a fast, scalable NoSQL database service. He said when AWS built Amazon Managed Cassandra Service, customers were asking for similar qualities in a Cassandra-compatible database service.

"We learned a lot from our experience operating DynamoDB, and we were able to apply that knowledge to build a highly scalable and serverless, purpose-built Cassandra-compatible database service, but the two are different," Oberoi said.

Serverless approach could be a big boost

ScyllaDB, another player in the Cassandra service marketplace, recently updated its namesake database platform to improve performance. ScyllaDB builds its own open source database, which is compatible with Apache Cassandra.

Dor Laor, CEO of ScyllaDB, singled out the serverless capability as the most impressive part of Amazon Managed Apache Cassandra Service.

"The user doesn't need to provision any servers, and it's possible to create a table immediately," Laor said. "A pure service approach and friendliness are the key strengths of AWS, and I am sure that users who struggle with the complexity of Cassandra will appreciate it."

Though the new AWS database will be a competitive offering to what ScyllaDB provides, Laor said he sees that as a good thing. He said he expects ScyllaDB's price and performance capabilities to draw new users as they look at what options are available in the market; he foresees the AWS service elevating the problems Cassandra users face, bringing more attention and appreciation to ScyllaDB. 

Laor said he also anticipates plenty of technical challenges for the new AWS database service, including compatibility with existing Cassandra deployments.

"Obviously, the goal of the service is to convert Cassandra users to AWS managed Cassandra, yet the service lacks several key features -- from multi-region to materialized views, counters, limitations on object size, etc.," Laor said. "Of course, all these can be solved over time."

AWS acknowledges that it is does not currently have every Cassandra feature. AWS' Oberoi cautioned that Amazon Managed Cassandra service is in preview and will add features over time. That said, he noted that most customers can already use and test services during the preview, with little or no modifications to their existing Cassandra application code.

"Over the course of the preview and at GA, we will continue adding features to Amazon Managed Cassandra Service that help customers run Cassandra workloads in production, such as cross-region replication," Oberoi said. "We built Amazon Managed Cassandra Service to provide customers with single-digit, millisecond performance at any scale, and as a result we made the decision not to add support for experimental Cassandra features or specific Cassandra features that customers have told us do not scale, do not perform well or have stability issues, such as aggregator functions, counters and materialized views."

Next Steps

Apache Cassandra 4.0 advances open source database

Dig Deeper on Database management

Business Analytics
Content Management