AWS’ open source chief claimed major corporations are in the midst of large scale migrations from Redis to Valkey as the open source in-memory database project hit 8.1 this week.
The release, just over a year after the project launched, delivers significant performance improvements maintainers said, in large part to the involvement of major Chinese contributors including AliBaba and TenCent.
Key additions in 8.1 include native JSON support, and native Bloom filters. The new version also gets a vector similarity search module contributed by Google.
The project claims performance improvements include a “memory efficient dictionary implementation” leading to a 20% memory footprint reduction for common key/value workloads, and 20% better performance for encryption workloads using I/O threading.
AWS principal engineer, and Valkey maintainer, Madelyn Olson said Alibaba, Tencent and Huawei were all big contributors to the project.
“In China, hardware is a little more expensive, so they care more about being more hardware optimized. So that’s why I think there’s a lot more interest in China around Redis compatible APIs.”
The project was spawned after Redis (the company) last year switched from a BSD license to a source available license. That followed a move to an earlier server side public license strategy for some of its technology after it accused cloud giants of exploiting the project without contributing. Major backers of the fork included Ericsson and AWS and Google.
AWS director of open source strategy David Nalley said the memory performance improvements would accelerate adoption by large users.
He said that following “cautious optimism” from “household brand” users last year, there were meaningful migrations in progress. “There's a number of really large household name brands that we're seeing who are talking to us and telling us, ‘we're 20% of the way through migrating all of our clusters.’”
“If you’re in a regulated industry, moving a workload is not just pushing a button,” he said. “Even though in the AWS console you can literaly push a button and move from Redis to Valkey,” Nalley said.
“I talked with one customer recently who had migrated just over 700 clusters," he continued. "But they still have several 1000 more to make before the before they've actually migrated completely,”
Nalley added the vector search module choice followed a “bake off” between AWS and Google, both of whom wanted to see the functionality incorporated into Valkey.
“We brought our implementation, they brought theirs, and ultimately, the community decided to go with Google's implementation for vector search. But we're really happy about that.”
The link has been copied!
