New Data Types – immudb version 1.5 released
After many months of hard work we’re thrilled to introduce version 1.5 of immudb, which comes with two major features: ...
After many months of hard work we’re thrilled to introduce version 1.5 of immudb, which comes with two major features: New data types and the Document model additional to SQL and KV.
MVCC: Multi-Version Concurrency Control*, Data Retention, Performance Improvements, Float64 data-type and Document Data Model APIs. In addition to fixing bugs, we enhanced the overall experience and last but not least improved the performance of some workloads by 60 times.
You can find the new release here for all major platforms:
MVCC: Multi-Version Concurrency Control
Detection of conflicts is automatic when multiple transactions are processed concurrently. MVCC in conjunction with snapshots provides a lock-free and optimistic approach to ensuring data consistency.
Data retention refers to the practice of keeping data for a specific period of time before deleting it. This practice is commonly used in various industries and organizations to comply with legal and regulatory requirements, as well as to manage storage space and maintain data integrity.
One of the primary benefits of data retention is its ability to help maintain disk space. By setting a retention period, organizations can automatically delete data that is no longer needed, freeing up disk space for new data. This can be particularly useful for organizations that deal with large amounts of data, such as those in the financial or healthcare industries, where storing vast amounts of data can be costly.
To get more information about data retention please refer to the documentation on docs.immudb.io.
Concurrent bulk indexing makes indexing faster than ever. It simultaneously indexes all entries within a single transaction. Moreover, multiple transactions can be indexed at once.
As of this release, values can be stored in the same transactions logs as headers. When this mode is enabled, transaction commits may be speed up by up to 60%, but data retention will not be supported. The setting can only be set at the time of database creation, but a replica can have a different setting.
To further speed up transaction processing, file preallocation can also be enabled. On Linux, fdata_sync will be used when this setting is enabled. As a result of enabling file preallocation and embedded values mode in our initial evaluations, transaction throughput was increased by up to 100% (in comparison with the previous immudb release v1.4.1).
It is now possible to use double-precision floating-point data types with SQL according to IEEE-754 64-bit floating-point specification.
Document Data Model API
The document-data model provides simple yet powerful storage. It is possible to store and retrieve documents represented as JSON documents based on conditions that relate to indexed fields.
You can find the full change log and the release download here: