Operations | Monitoring | ITSM | DevOps | Cloud

Valkey JSON module now available on Aiven for Valkey

The Valkey JSON module implements native JSON data type support within Valkey, allowing users to efficiently store, query, and modify complex, nested JSON data structures directly. This overcomes previous architectural complexities, such as needing to serialize entire documents as strings or flatten data into hashes, by providing native handling for nested data models.

Benchmarking Diskless Topics: Part 1

We benchmarked Diskless Kafka (KIP-1150) with 1 GiB/s in, 3 GiB/s out workload across three AZs. The cluster ran on just six m8g.4xlarge machines, sitting at <30% CPU, delivering ~1.6 seconds P99 end-to-end latency - all while cutting infra spend from ≈$3.32 M a year to under $288k a year. That’s a >94% cloud cost reduction. Extending Apache Kafka does come with an explicit tax.

Different ways to Search Text in PostgreSQL

When it comes to text search, PostgreSQL offers a surprisingly rich set of tools. Initially, text search capabilities were quite basic, often relying on the LIKE operator. This is inefficient for large amounts of text and lacks the nuance that comes with language. A major breakthrough came with the introduction of the built-in tsquery and tsvector data types, along with the associated functions, as part of PostgreSQL's core distribution. But tsquery is not the only option.

Replacing cron jobs and dbt pipelines with ClickHouse Refreshable Materialized Views

For ClickHouse to achieve the speed it's known for, it has historically relied on a trigger-based and incremental architecture for materialized views. This architecture is known to be very efficient but lacks flexibility, such as when working with data that needs to be backfilled rather than append-only.