Rust AI & Data Meetup: Manvika Tuteja, Valkey GLIDE Overview
Recording: Rust AI & Data Meetup: Manvika Tuteja, Valkey GLIDE Overview
Inhale, and exhale. Um, thank you everyone for being here, and thank you for the introduction. Raise of hands, how many of you have heard of Valky? Okay, that's pretty good. Keep your hands raised if you are using that in production in your stack. Not >> Integration with Valky. >> You do? We should chat after this. Okay, for those who don't know, Valky is an open source BSC license high performance key value in memory data store. I know that was a mouthful, let me make it simple
It is a way for you to access your data extremely quickly and store it in your memory, in your RAM. Oops. Um, AI is new, but fast data is not. An application may need to remember a session, fetch contacts, manage a queue, enforce a rate limit, or return a result quickly. That's where a cache comes in comes in. The model is important, but it is only a part of the bigger product. Uh, that is the connection to Valky. Many applications need fast access to data and state around the model
In plain English, um, like I mentioned before, it is completely open source. It is an in memory key value data store, which means the data you need quickly can be kept in RAM. The easiest way to think about it is your main database may hold the durable source of truth, while Valky can help with the fast moving data that your application needs right away. A quick story for Valkey. It is Let me just emphasize how remarkable it is for a project to come about and have a governance structure within 8 days. Since the fork that happened in 2024, Valkey turned around to be a completely independent open source project under the Linux Foundation within 8 days of the license change from the previous key value data store. I'm not going to worry about that. >> [laughter] >> Um 18 days to a generally available release means we forked, set up the infrastructure, validated stability, and shipped production-ready code in less than 3 weeks
Um this speed was possible because the community was ready, the Linux Foundation moved fast, and we had a clear goal, preserve what developers loved about Redis through >> [laughter] >> open source governance, and that's how it came about. We have a lot of industry leaders that are currently using Valkey in production, and we love chatting with them. So, if you're one of those, I would love to chat with you after. Now, where does Rust fit into the Valkey picture? Glide is the official family of Valkey client libraries. Its currently supported clients include Python, Java, Node.js, and Go, with C, PHP have preview releases, and C++ and Ruby are under active development. Across the family, the goal is a natural interface for each application language with a shared Glide core written with Rust. That Rust core centralizes difficult client behavior, such as connection management, command handling, cluster topology, routing, recovery behavior. You all already know this
Instead of every language implementation solving the same distributed system separately more of that work can live in one shared core and that's where Rust comes in. Uh now bringing it back to AI a production AI experience needs more than inference. It may need conversation or session state, prompt and response caching, tool job queues, rate limits, counters, fast fast access to shared application data. Valky can support that real-time data layer around the model. Glide gives application teams a way to reach Valky from languages they already reuse with um shared Rust engineering underneath. Another reason why Valky is here today is because open source grows when people learn how a project works, run it in real systems, share what breaks, improve the documentation, contribute code, teach other people, all of those good things. So we're always welcoming new people to join the community, be a part of like content, code, just be a part of meetups. Not too long ago we launched a meetup program
We're now in eight cities which is super exciting. I recognize some of your faces. I know you were there at the first uh Valky meetup that happened here sometime ago too. So thank you again for being here. Um if you want to explore Valky start at valky.io. You can run the official Valky container and you can run through Glide and Rust module repositories. Or you can find an issue on the GitHub repo. I have linked them there
You can look them up. Now lastly this is the fun part of it. We have a flagship Valky conference coming soon. It's in October. It's in Prague. The CFP is open till October 2nd. So do apply and if you're interested in attending and listening to great talks You can also register. We're also considering some travel funding
So, if that is of interest, please do check it out. That was everything for Valky. Thanks all for being here again. Um the question was if Valky is under an organization. The answer is no, it is not. It is governed by the Linux Foundation and it is under that umbrella, but there are a lot of organizations such as the ones I showed on screen that are deeply invested in the project and its growth. And um yeah, it's a vendor neutral project. >> One more question here
>> Yes. >> Hi. Um so, I just had a quick question. This is the first time I've heard of Valky, um but you mentioned that and you mentioned it's like a drop-in replacement for Redis, um but not maybe with some of the licensing issues. Um what are some ways it can also differentiate itself from Redis? Like, what does it offer outside of what Redis is doing that it currently >> Um I'm not 100% familiar with what Redis is offering right now, but I can talk to what Valky is offering. Is there something specific you are looking for? Like, it has multi-IO threading, it it has atomic sub-migration, it's extremely fast. We don't do public benchmarking as such if that's something that you were looking for. I don't have that answer for you
>> We can talk offline or whatever. >> Yeah, we can talk about it right after this. Any other questions?