text stringlengths 0 2.67k |
|---|
**Justin Garrison:** Yeah, we're like "This happens at a certain minute. Let's see what's going on." So we add skew to it in Telegraph. We're like "Oh yeah, add a one minute skew to these metrics, and we're fine. We don't actually need it on the minute." And the problem went away. |
Those were my first two "Oh, I get the power of this thing as time-series data, what it's supposed to do. We can get rid of some of those older style tools, and kind of reduce the overhead and the maintenance and all that stuff just storing four terabytes of data." |
**Autumn Nash:** \[19:45\] I think once you get to a certain size for SQL though, you have to figure out something else or you're going to have to get somebody who's really good at architecting your data. But - I mean, you're kind of sort of stuck at a certain point... But I think a lot of what influences databases too... |
**Lili Cosic:** Get started, yeah. |
**Autumn Nash:** Yes. And they have such good documentation, and it is just -- it is very much made in a way that it's like, you can just throw whatever you want in here, not have a DBA and get stuff out, you know? |
**Break**: \[20:24\] |
**Justin Garrison:** Do you remember when Mongo was exposed to the internet because the defaults was no password? |
**Lili Cosic:** Oh, really? |
**Justin Garrison:** People started getting ransomware in their Mongo database... People would steal all the data and leave one object in it, that's like "Pay me this Bitcoin." |
**Autumn Nash:** I don't know, I think that there are lots of great -- I love open source and the fact that you can do all the things with the databases, and I love that you can choose between managed... But sometimes when you don't know what you're doing and you just want -- just get a managed database. |
**Lili Cosic:** As someone working on a managed database, I couldn't agree more. |
**Autumn Nash:** You know? Sometimes it's worth paying stuff so you don't have to worry about it, and so that you can call and get someone to help you. Because honestly, just the fact that you can call and ask people, like get customer service with someone who's an expert to do your data schemas, and really talk about ... |
**Lili Cosic:** Support, yeah. |
**Autumn Nash:** Yeah. |
**Lili Cosic:** I did notice that there's a lot of -- in the recent, I would say, years, there has been a shift in databases where they're really specialized, they're really focused. |
**Autumn Nash:** They really are. And then it's funny, because you're going for an interview and they're like "We want you to have 20 years of Postgres and Kafka experience", and I'm like "There are certain things that two people are not going to have 20 years of... But how old do you think I am?" And did I have two jo... |
**Lili Cosic:** No, I don't think so... Especially in this job market. |
**Autumn Nash:** I know, they're asking for stuff that -- I swear, they'll be like "We want you to have 20 years of experience in something that was made two years ago." And you're like "Oh, okay..." It's so wild. What did you do with Prometheus? |
**Lili Cosic:** Yeah, I worked on -- so I was at Red Hat and I was part of the OpenShift organization... And we essentially -- OpenShift is the Red Hat Kubernetes offering that uses the components from Kubernetes, but ships itself, or unpacks it and then packs it neatly into -- |
**Autumn Nash:** Corporate and open source. |
**Lili Cosic:** Yeah. |
**Autumn Nash:** They have the weirdest babies. |
**Lili Cosic:** But it works really well. It was definitely built by experts who work on upstream Kubernetes still. And I was part of the cluster monitoring team, which essentially was in charge of the monitoring component, as every OpenShift came with monitoring. So out of the box you had a Prometheus stack, you had m... |
**Justin Garrison:** Were you there before the CoreOS acquisition? Because all the operator stuff was CoreOS, right? My experience with OpenShift running it was back in the gears and cartridges days, before Kubernetes... And I'm like "This isn't fun." And then they went into the Kubernetes phase and were like "Ah, this... |
**Lili Cosic:** 4. |
**Justin Garrison:** 4, okay. |
**Lili Cosic:** Yeah, 3 was the -- I think three point something was still... At some point they did a switch to operators. If I remember correctly, it was three or it was four. I don't know. I worked on four. I joined the week before Red Hat got sold to IBM. And I also worked at HashiCorp, and... Yeah, same thing happ... |
**Autumn Nash:** Acquisitions are following you. |
**Lili Cosic:** \[28:02\] So that was fun. But yeah, no, I worked closely actually with the CoreOS people before, as I worked at a small company called Kinvolk. They were a Linux consultancy. They were bought by Microsoft. |
**Justin Garrison:** Yeah. |
**Lili Cosic:** Yeah... Again. And I worked there on some of the early operators. I was working with Chef on the Habitat operator, which was back in 2016, or something... So it was definitely in the early days. So it was nice to just watch what the CoreOS folks were doing, and... Essentially, they invented the term ope... |
**Justin Garrison:** And that's something that I think is really -- like, CRDs were novel at the time, right? |
**Lili Cosic:** Very. |
**Justin Garrison:** Because this was coming off of third-party resources. We shifted from third-party resources to custom resource definitions... But people were still only building -- |
**Lili Cosic:** Nikita, right? That was the one, I think -- working on that with Dr. Stefan, if I remember correctly, back in the day. |
**Justin Garrison:** Yeah. But there was only controllers on the core Kubernetes objects. You were just doing pods and deployments, and that sort of stuff... And then they throw out this data, like "What if everything's a CRD, and you could do operators on all of them, or controllers on all of them?" We call that an op... |
**Lili Cosic:** To the extreme. |
**Justin Garrison:** ...and they went hog wild on it. |
**Lili Cosic:** I know, I know. Not everything has to be an operator, people. |
**Autumn Nash:** Am I the only one who wonders why they named CockroachDB CockroachDB? Because I have so many questions... |
**Lili Cosic:** If I remember correctly, it's because it's the last thing that dies, or something. It's like indestructible. |
**Autumn Nash:** Oh, okay. That's actually pretty good marketing. |
**Justin Garrison:** You want your database to always be available, and always be up, and that was... Yeah, it's always alive. |
**Lili Cosic:** And if I remember correctly, TigerBeetle, another database with similar naming team, where it's also indestructible, or it's like... |
**Justin Garrison:** TigerBeetle? Never heard that one. |
**Lili Cosic:** Is it TigerBeetle? |
**Justin Garrison:** I'm waiting for DungBeetle, or something. |
**Lili Cosic:** TigerBeetle, definitely. That's a good one. It's a database, again, specialized in finance. So they do just a very specialized database in finance, and they do it really well for a very specific use case, and it outperforms everything in transactions, and to just a regular SQL. |
**Justin Garrison:** How did you get into this space? You've been at a lot of these companies, and it seems like IBM is going to buy Influx at some point now... But you were doing Red Hat, you were doing CoreOS, you were doing... |
**Lili Cosic:** Kinvolk. |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.