Matthew Brender on Navigating Unstructured Data | GOTO 2015
Hi, it's Mike with UGtastic . I'm here at GOTO Conference 2015. Hi, it's Mike with UGtastic . I'm here at GOTO Conference 2015. Dive into Matthew Brender's talk on navigating unstructured data at GOTO Conference 2015! Learn the differences between key-value databases like React and Redis, and the benefits of distributed systems for handling large amounts of data.
The Interviewer
Mike Hall
Interviewer, UGtastic
The Guest
Matthew Brender
navigating unstructured data
Mike Hall
Interviewer, UGtastic
▶ Speaking
Hi, it's Mike with UGtastic . I'm here at GOTO Conference 2015. I 'm standing here with Matthew Brender, who gave a talk on how to navigate unstructured data. Thank you very much for taking the time to speak with me. So, navigating unstructured data, where did this talk come from and what was it about? Sure.
Matthew Brender
navigating unstructured data
▶ Speaking
So, I'm here on behalf of Basho, my employer, and we 're the creators of React, a distributed NoSQL database, which fundamentally breaks some basic concepts when you've come from a relational world like we all do. So, I've been doing some research on, okay, first off, what type of data is unstructured versus something that fits more naturally into a relational database? And secondly, what are some of the actual system architect ures that people are taking into effect and trying to achieve with their infrastructure, as opposed to having these segmented groupings and data pushing here and there? Just pulling there and refactoring here and there, trying to have more of a raw event stream that results in a repeat able sequence of events that you can analyze after the fact. I'm fascinated by that. There are a few different architect ures that I reviewed.
Mike Hall
Interviewer, UGtastic
▶ Speaking
Okay.
Matthew Brender
navigating unstructured data
▶ Speaking
So, with the unstructured data, I mean, is React a document store database or is it, I mean, NoSQL is such a broad term.
Mike Hall
Interviewer, UGtastic
▶ Speaking
Yeah, and that was part of the thing we have to tackle, right?
Matthew Brender
navigating unstructured data
▶ Speaking
Like, you hear NoSQL and you don't know what to assume. You just know that it's not rows and columns. So, key value is the simplest form. That is, in fact, what React is. But I also went over columnar and graph databases and how they fit different goals. But what's great about a key value database and particularly an AP key value database that is sacrificing some consistency state for in order to have a growing level of availability in an architecture, you find some nuances about that that are a little different. That are really effective for companies that are growing and the amount of data that they're consuming.
Mike Hall
Interviewer, UGtastic
▶ Speaking
Okay.
Matthew Brender
navigating unstructured data
▶ Speaking
I mean, and compared to like another really big player in that key value source, you know, Redis, I mean, how does Redis and React, I mean, do they serve different purposes or? Yeah, exactly. They're very complementary. So, I'd say a decent to a significant portion of our customers also use Redis side by side because Redis is key value, but what it does is keeps everything. I mean, you can use Redis to do both the data and the mapping of that data in memory, and you can perform set functionality on the data and actually aggregate it on the fly as you need it from your application. So, it makes an ideal read cache, but what it doesn't offer is any sort of guarantee on availability in the sense of durability in the case of a failure of a node. So, Redis right now, to the best of my knowledge, it's actually a master-slave architecture, so you have multiple systems and they own portions of the structure. So, Redis is the best way to do that, but it's also the best way for us to keep track of what's going on in the process and keep track of the data that we're sharing. So, Redis is the best way to do that, but it's also the best way for us to keep track of the data that we're sharing. What makes React and Cassandra and others like us really interesting is that data is evenly distributed over a consistent hashing table, meaning each node has a consistent amount of information and it's durable. It's written to disk before you get that acknowledgement, but that's still on a very lightweight millisecond transaction standpoint as opposed to something that could be much larger.
Mike Hall
Interviewer, UGtastic
▶ Speaking
Okay. And so, you know, did you come to React through the open source? I mean, did you come to React through the open source?
Matthew Brender
navigating unstructured data
▶ Speaking
Did you come to Bashow through the React open source, or have you been working with React more because of being a Bashow employee? No, it's very much the latter. I was not at all familiar with databases. I'm going to drift in a little bit. Just go. We have to stand uncomfortably close.
Mike Hall
Interviewer, UGtastic
▶ Speaking
You good?
Matthew Brender
navigating unstructured data
▶ Speaking
No, I'm comfortable. I'm comfortable.
Mike Hall
Interviewer, UGtastic
▶ Speaking
Okay.
Matthew Brender
navigating unstructured data
▶ Speaking
As long as you're comfortable, we're good. Fine. All right, great.
Mike Hall
Interviewer, UGtastic
▶ Speaking
Okay.
Matthew Brender
navigating unstructured data
▶ Speaking
So, I came to React through my work. I joined Bashow because I have a lot of respect and admiration for the leadership and the product. Every single person I spoke to said the product is rock SOLID, is by far really respected in the industry, and I came on to focus on more of our open source contribution strategy, and I've been learning distributed systems from scratch, actually, since I've joined. So, I didn't know databases. I didn't know distributed systems. I've got computer science background and system architecture background and just kind of keep building the skill set like we all do. It's one of those things that it's the employer wants somebody who's smart and can learn things. So, you went through that route. Well, you got the smart part down, so let's see what you can learn. So, with React, I mean, I know that Bashow has been really involved with supporting local community and being part of the go-to events over the years . When you go to an event, just curious about it as an evangelist, what is it like going to speak about React at so many different conferences and touching so many different communities? Yeah, no, it's a great question, and I want to separate it into two points. One is the title. I actually full-heartedly reject the title of an evangelist because an evangelist to me presupposes that I am the expert and you should be listening to me at all times. I am a developer advocate at Bashow, and it's minor, but it 's important to me because it more focuses on how do I cur ate other people's information and their skills and grow with them. As opposed to be the one that's already at some destination that other people are trying to get to. That model doesn't work very well for me, and what that allows me to do is I get to come in and be very honest and upfront that I have some naiveness to my ability to build these systems. I read voraciously about how people are, and then I just try to connect the dots and make sure other people can solve it with me. So it's really honest. I get to be very honest. I get to interact with people who are at all sorts of phases, either. Experts have been running with NoSQL since it was invented to people that still don't know what it means relative to big data and Internet of Things. So they see these trends and want to kind of unpack it. So you've been in a place where you're like, no, React, you just use MySQL. Yeah, yeah. And that's a pleasure is to really say, like, listen, you have what's called a little data problem. Just put it in your favorite relational database and be done with it. You don't need to scale like this. Yeah, yeah. And that honesty, that's core to the culture at Basho and something I really respect and admire about the community around us. We solve a specific set of problems, and if you need that problem solved well, we feel we have the best by far solution to that.
Mike Hall
Interviewer, UGtastic
▶ Speaking
Great. Well, thank you very much for taking the time to speak with me. I appreciate it. Yeah, my pleasure, Mike. It was great to do so and hang out with Shutter. Fluttershy. Fluttershy. You guys will see Fluttershy later. Thanks. Bye.
Critical Insights
durable
"React is a distributed NoSQL database that breaks traditional relational database concepts."
durable
"Redis is a key-value store that serves as an ideal read cache but lacks durability guarantees."
durable
"Distributed systems like React ensure data is evenly distributed and written to disk before acknowledgment."