Google Cloud is looking to take on rivals such as Oracle Corp. and Amazon Web Services Inc. in the database arena with AlloyDB, a new, fully managed and PostgreSQL-compatible relational database service for top-tier workloads.
Announced today at the Google I/O virtual conference, AlloyDB is built atop the open-source PostgreSQL database management system and powered by Google Cloud’s compute and storage services. It offers dozens of additional capabilities too, including artificial intelligence-powered management and full integration with Google Cloud’s array of AI and analytics services.
In a blog post, Andi Gutmans, Google Cloud’s general manager and vice president of databases, said AlloyDB is powered by an intelligent, database-optimized storage service that allows it to scale up seamlessly with predictable performance. In addition, he said, it’s backed by embedded AI and machine learning, analytical acceleration capabilities and automatic tiering of data, making it ideal for any high-value workload, with minimal management overheads.
Gutmans explained in an interview with Tech To See that enterprises are becoming increasingly desperate to move to the cloud with open-source database services that eliminate their dependency on legacy tools. However, he said, there’s a need to standardize on open systems such as PostgreSQL to eliminate vendor lock-in and the sometimes “unfriendly” and expensive licensing terms that come with legacy systems.
“Migration to the cloud is also migration to open source and getting the flexibility to run workloads anywhere,” Gutmans said.
The problem enterprises face is that running and replatforming business-critical workloads for open-source systems can be daunting tasks, with teams often struggling with issues such as performance tuning and managing application availability.
That’s where AlloyDB aims to make a difference, by bringing the best of Google to the PostgreSQL ecosystem, the company said. Notably, AlloyDB maintains full compatibility with PostgreSQL 14, the latest version of the open-source project. That means customers can migrate existing PostgreSQL applications without making any code changes.
“Postgres from a compatibility and feature set perspective is best for lifting and shifting to the cloud,” Gutmans said. “AlloyDB is really focused on bridging that gap to lift-and-shift… to run your high-end workloads.”
Gutmans explained that AlloyDB has been under development for a long time to meet customer’s needs for additional security, performance, scalability and reliability. However, he promised that customers will be happy with a product that’s even better than what the company had planned, with Google making some big claims over AlloyDB’s overall performance.
For instance, Google said, AlloyDB is four times faster than standard PostgreSQL when it comes to commercial-grade transactional workloads, and two times faster than the AWS Aurora database service it will also compete against. That’s possible thanks to multiple layers of caching that are automatically tiered based on workload patterns, Gutmans explained.
AlloyDB also offers a high-availability service-level agreement of 99.999% inclusive of maintenance. It has the ability to detect and recover from most database failures automatically and within seconds, Gutmans said, independent of database size and load. The architecture also supports features that include nondisruptive instance resizing and database maintenance, ensuring customers will have a reliable and continuously available database to support their most critical workloads.
Database patching, backups, scaling and replication are all handled automatically by AlloyDB’s autopilot system. It also uses adaptive algorithms and machine learning to perform all manner of tasks including vacuum management, storage, memory management, data tiering and analytics acceleration. In other words, AlloyDB learns about each workload and then intelligently organizes the data across its memory to improve access.
Perhaps the most impressive claim is that AlloyDB can deliver analytical queries at up to 100 times faster than standard PostgreSQL. That’s thanks to a vectorized columnar accelerator that works by storing data in-memory in an optimized, columnar format so it can be scanned and aggregated much faster, Gutmans explained. As such, AlloyDB is ideal for workloads such as business intelligence, reporting and also hybrid transactional and analytical workloads.
“We built AlloyDB to address the most demanding workloads,” Gutmans said.
There’s more to like too, with Google promising predictable and transparent pricing with no input/output charges that can cause a spike in costs.
“Opaque I/O charges are hard for customers to control and can account for up to 60% of the bill,” Gutmans said. “So Google has no I/O charges. Customers pay for compute and storage only. We take the risk on ourselves [because] we want to be the friendliest cloud.”
Holger Mueller of Constellation Research Inc. told Tech To See that AlloyDB looks to be another great Google Cloud product offering and that it will be interesting to watch how much market traction it can get.
“AlloyDB is a solid offering that’s highly differentiated from the competition, supports a popular standard and comes with innovative pricing plus superior performance,” Mueller said. “It’s going to be interesting to see how many AlloyDB workloads can wrestle away from competing offerings.”
International Data Corp. analyst Carl Olofson was similarly optimistic about AlloyDB’s prospects, explaining that databases are increasingly heading into the cloud as more businesses undergo digital transformation.
“With AlloyDB, Google Cloud offers large enterprises a big leap forward, helping companies to have all the advantages of PostgreSQL, with the promise of improved speed and functionality, and predictable and transparent pricing,” he said.
Google said AlloyDB is available to everyone in public preview now, with general availability slated for the second half of the year.
With reporting from Robert Hof