AlloyDB is a proprietary fork of PostgreSQL created by Google Cloud.[03]
- Developer
- Country of Origin
- US
- Start Year
- 2022
- Project Type
- Commercial
- Derived From
- PostgreSQL
- Compatible With
- PostgreSQL
- Operating System
- Hosted
- License
- Proprietary
AlloyDB is a proprietary fork of PostgreSQL created by Google Cloud.[03]
Checkpoints[03]
In AlloyDB, WAL records are written to the storage layer synchronously. Then, the log processing service in the storage layer is responsible for replaying the logs continuously. AlloyDB supports checkpoints but there is no need for flushing the changes in the database layer to the storage layer.
Hardware Acceleration[08]
The system runs on Google Cloud instances but there is no specific hardware required, although two VM instances are required for an AlloyDB primary instance.
Indexes[09]
The concurrency control AlloyDB uses is the same as in PostgreSQL and PostgreSQL supports B-tree, Hash, GIST, SP-GIST, GIN and BRIN index types.
Isolation Levels[10]
Isolation level in AlloyDB can be set through database flag "default_transaction_isolation", where it can be set to either “read uncommitted”, “read committed”, “repeatable read”, or “serializable”.
Joins[11]
AlloyDB builds on PostgreSQL where PostgreSQL supports the following three join strategies: nested loop join and it is possible to utilize the values from the row of the left relation as keys for scanning the right relation's index given that the right relation can be processed with index scan, merge join and hash join.
Parallel Execution[13][14]
AlloyDB employs PostgreSQL, which enables both inter-operation parallelism, allowing for parallel execution of parts of the plan, and intraoperator parallelism, enabling partitioning and sharing of specific operators such as scans, joins, aggregation, and append across multiple processes.
Query Compilation
AlloyDB does not support code generation or JIT optimization at the moment.
Query Execution[15]
The columnar engine of AlloyDB uses a vectorized model where SIMD instructions are used when deemed suitable. Bloom filters are also used for optimization purposes, depending on the selectivity of the workload.
Query Interface[16]
AlloyDB supports SQL and Stored Procedures. Additional REST API and CLI tools are available for managing the database system resources.
Storage Architecture[03]
For AlloyDB, all the WAL records are written synchronously to the storage layer, where the records get processed by the log processing service.
Storage Model[15]
AlloyDB stores the data in the storage layer following the same format as PostgreSQL. However, a columnar engine is used where the data that gets loaded into memory can take on either the row-based format or the columnar format based on the workload, which is decided using machine learning models. As a result, query optimizations relating to columnar formats can be applied to the relevant data.
Storage Organization[17][03]
PostgreSQL uses heap files to store the database data. For AlloyDB, after the log processing service processes the WAL records, the resulting database blocks get sent back to the primary and replica instances from the storage layer.
Stored Procedures[16][18]
AlloyDB supports stored procedures. SQL, PL/pgSQL, and C are supported by default. In addition, extensions to PL/proxy and PL/v8 can be added to support additional procedural languages.
System Architecture[03]
AlloyDB disaggregates its database layer and its storage layer. As a result, each component of the database system becomes more elastic to scale. AlloyDB refers to the main unit of resource as a cluster, which consists of one primary instance and multiple read pool instances. The primary instance handles read/write requests while the read pool instances only have read access. The storage layer is shared between all cluster resources.
Views[19][20]
AlloyDB follows the same rule as PostgreSQL where both Materialized Views and Virtual Views are supported.
Citations
20 sources- AlloyDB for PostgreSQL | Google Cloud google.com
- https://cloud.google.com/alloydb/docs google.com
- AlloyDB for PostgreSQL intelligent scalable storage | Google Cloud Blog google.com
- 5. Concurrency Control :: Hironobu SUZUKI @ InterDB interdb.jp
- PostgreSQL: Documentation: 7.1: Multi-Version Concurrency Control postgresql.org
- PostgreSQL: Documentation: 18: 1. What Is PostgreSQL? postgresql.org
- PostgreSQL: Documentation: 18: 3.3. Foreign Keys postgresql.org
- View instance details | AlloyDB for PostgreSQL | Google Cloud Documentation google.com
- PostgreSQL: Documentation: 18: 11.2. Index Types postgresql.org
- Supported database flags | AlloyDB for PostgreSQL | Google Cloud Documentation google.com
- PostgreSQL: Documentation: 18: 51.5. Planner/Optimizer postgresql.org
- 9. Write Ahead Logging (WAL) :: Hironobu SUZUKI @ InterDB interdb.jp
- PostgreSQL: Documentation: 18: 15.1. How Parallel Query Works postgresql.org
- PostgreSQL: Documentation: 18: 15.3. Parallel Plans postgresql.org
- AlloyDB for PostgreSQL Columnar Engine | Google Cloud Blog google.com
- PostgreSQL: Documentation: 18: CREATE PROCEDURE postgresql.org
- Introduction to PostgreSQL physical storage rachbelaid.com
- Supported database extensions | AlloyDB for PostgreSQL | Google Cloud Documentation google.com
- PostgreSQL: Documentation: 18: 39.3. Materialized Views postgresql.org
- PostgreSQL: Documentation: 18: CREATE VIEW postgresql.org