Google Cloud recently announced AlloyDB for PostgreSQL, a managed PostgreSQL-compatible service targeting enterprise deployments. AlloyDB is a full-featured cloud database supporting atomicity, consistency, isolation and durability (ACID)-compliant transactions.
AlloyDB is compatible with PostgreSQL 14, providing portability for existing workloads and transitioning off of legacy databases. Google claims that it is "four times faster for transactional workloads, and up to 100 times faster for analytical queries" than a standard PostgreSQL deployment. Andi Gutmans, GM and VP of engineering for databases at Google Cloud, writes:
Like many managed database services, AlloyDB automatically handles database patching, backups, scaling and replication for you. But it goes several steps further by using adaptive algorithms and machine learning for PostgreSQL vacuum management, storage and memory management, data tiering, and analytics acceleration. It learns about your workload and intelligently organizes your data across memory, an ultra-fast secondary cache, and durable storage.
In a separate "AlloyDB for PostgreSQL under the hood" article, Ravi Murthy, engineering director at Google, and Gurmeet Goindi, director of product management at Google, explain how the storage layer works:
AlloyDB begins by separating the database layer from storage, introducing a new intelligent storage service, optimized for PostgreSQL (...) The fully disaggregated architecture even at the storage layer allows it to work as an elastic, distributed cluster that can dynamically adapt to changing workloads, adds failure tolerance, increases availability, and enables cost-efficient read pools that scale read throughput horizontally. Multiple layers of caching throughout the stack (...) give developers increased performance while retaining the scale, economics, and availability of cloud-native storage.
Source: https://cloud.google.com/blog/products/databases/alloydb-for-postgresql-intelligent-scalable-storage
Many experts suggest that AlloyDB is the response to Amazon Aurora, with Google Cloud claiming that AlloyDB can be two times faster for transactional workloads than the AWS service. Unlike Amazon Aurora, AlloyDB does not charge for I/O but does not currently offer a serverless version. Gutmans adds:
A few folks have reached out to me asking how "disaggregated" compute and storage used in AlloyDB (and Google systems) is different from compute storage separation that other vendors talk about (...) It is quite different, as in our model, compute and storage are both "extremely" flexible at the most granular level and very low latency. Others tend to have high latency between compute and storage and/or co-location of the two.
Mark Callaghan, previously distinguished engineer at Mongo and MTS at Facebook, comments:
Cloud native PostgreSQL is here (Yugabyte, AlloyDB, Amazon Aurora) but upstream PostgreSQL is the only way to take advantage of fast (low-latency + direct attached) storage devices. Will be interesting to see if that changes.
In a popular Reddit thread, some users highlight the lack of a serverless pricing model and the limited features versus Cloud SQL for PostgreSQL. Google Cloud last year introduced a PostgreSQL interface for Spanner, as previously reported on InfoQ.
Compute resources for AlloyDB are priced per vCPU and GB of memory and storage prices depend on the region. For networking, ingress is free and egress pricing varies according to the destination.
