Introduction to Redis Labs - Episode 106 - Sponsored

Machine-generated from the episode audio and not hand-corrected, so names and technical terms may be imperfect. The audio is authoritative.

Hey, you're watching another CTO Dose sponsored by Redis Labs. We're going to go right into it. What's the difference between Redis and Cassandra? Why Redis Labs? I mean, I can go and consume Redis from AWS, et cetera. We ask the tough questions of Redis Labs. Stay tuned. Sponsoring yet another CTO Dose from beautiful Las Vegas, as you can tell with the rain pouring outside. Doing AWS re-invent Redis Labs. And from Redis Labs, I have their Chief Product Officer, Alvin Richards.

Alvin, welcome to the program. Great. Thanks for the invite, for coming to talk. Not a problem. So, first off, I know Redis. But when I think of Redis, I hearken back to, I'm going to say a bad word, because, you know, we're now in the era of Kubernetes. And I'm going to say the bad word of OpenStack. So, I don't associate Redis with OpenStack, but I associate it with open source. And I'm an enterprise guy. And when I think of open source, I think about finding talent during the OpenStack days.

And when that talent, I developed that talent around taking a project from open source, adding some of my bits to it, adding my IP to it. And then a year later, my whole team disappeared. I absolutely need a ecosystem around a open source project. I've seen plenty of Redis kind of just bare metal taking going out and pulling down a project and installing it. How does Redis Labs solve this problem of product versus project for me? Great question. So, let's start off with Redis the project.

So, it's been going about 11 years now. It's wildly popular by any metric. You know, it's three years running the most loved database on Stack Exchange, the most deployed database on AWS, billions of Docker pools. And so, by any reasonable metric, it's super loved. It's super used. But why do people, why do developers first pick it up? Mostly because it's sort of, it does this sort of Vulkan mind meld with developers because they use it in a way that's very familiar with them.

They're dealing with data structures rather than object relational mapping layers or object document mapping layers. They're dealing with lists, sets, sorted sets. It's very, very natural. Then to sort of segue into what does Redis Labs do? Well, we're the major sponsor of the project. So, we commit code to that. But we also produce an enterprise version. And so, most people start off thinking of Redis as a cache. They may use it, for example, as a session store.

And that's great for when you're developing. But what you want to do when you're a big global company, you want to distribute that data around the globe. Well, this is where Redis Enterprise comes in. We have active-active replication. So, you can have that same session state in multiple data centers. You lose a data center, that person who goes back to their shopping cart hasn't lost their shopping cart. They have everything there because the data is replicated between all of those data centers in near real time.

And so, whether that's a shopping cart or whether that is the leaderboard in your favorite game that you're playing on your favorite online game. Yeah, I'm playing multiplayer Super Mario Bros. Exactly. Exactly, right? So, you want that session state. You want that leaderboard. Or you want that risk analysis for that credit card transaction to have the same data, irregardless of where you are. You want to be able to access that data super, super quickly. So, we're in an open source world.

We're kind of introducing open source to my enterprise audience. Redis, they might have heard of Redis, again, from a cash layer or developers using it as something to have a basically free database on their development environment. They may move to something else in production for these very reasons. But some of the use cases you gave to me is kind of new to me from a Redis perspective. Redis is not something that I readily work with day-to-day as an operator. So, when I think of, like, these multi-regional global databases, the first thing that I think of is, like, a Cassandra DB.

What is the difference? Can you explain to the audience the difference between the nature between a Redis and a Cassandra? Absolutely. So, you know, I've been in databases a long time. I started with relational databases in the 80s. And, you know, we got to a sort of stasis point with relational databases, which is you had a single server, the database running on it, and you could load up however much data. The problem we got in the sort of, you know, early 2000s is there was more data than a single server could hold.

So, you had to put data across multiple servers. You had to have some form of distribution of that data. And the relational model has some challenges when you start distributing the data. Trying to do a join across servers is very problematic and it's very slow. And so, other data models formed to allow that data to be distributed more easily. So, Redis is a key value store. Cassandra is a wide column store. MongoDB is a document store. These are just different ways of representing the data that have different characteristics on how you can distribute that data.

So, depending on the problem you solve, you choose one database over another one or whatever your way of approaching the problem. So, you can say solve, potentially solve the same problem with the same database, but advantages, disadvantages. We'll leave that for the database experts. Yes. So, if you remember, there was advertising in the 90s by Oracle that said put all your data into Oracle. Right. Everybody tried that and everybody kind of didn't like it. Right. And so, I think the challenge for any IT organization is there are lots and lots of databases.

There's like 250 on dbEngines. But there's these sort of main different categories of databases. And for an IT professional, you have to pick the right tool for the job. And then you end up with this sort of polyglot environment where you still may have some relational. You may have some key value. You may have some document stores. And so, it's really understanding the problem you're trying to solve. What are the scalability requirements? What are the availability requirements?

And can you pick the right tool to solve the problem? For Redis, we think the problems it solves is where you've got very large sets of data or you've got data that you want to access very, very quickly. So, a couple of really simple examples of that would be bidding on a Google AdWord. Google gives you 75 milliseconds. All right. So, you've got to give your bid in 75 milliseconds. So, the more data you can look at in those 75 milliseconds, the better chance you've got of bidding the right offer for your client.

But you can't take 76 milliseconds. You need something that's got very guaranteed latency. So, that could be an auction bid. It could be, you know, risk analysis when you swipe your credit card. There are more and more use cases where the more data you have, the richer experience you can give, you know, personalization on the web. Do you like getting those ads for products that are completely irrelevant to you? Or do you want those ads that are really tailored to who you are and what you're doing right now?

And so, this is the sort of where Redis comes in because you want to look at those big data sets. You want to look at them real quick. So, as I'm looking at databases, these 250 different databases that I have to pick from, the way enterprises typically select databases is that we'll say for this type of problem, if you have a key value store problem, as an enterprise, we standardize on this DB. For column databases, we standardize on this DB. That way, I can do business with a single DB vendor and I can get my support from that vendor.

What's the elevator pitch for choosing Redis Labs? I mean, Redis is an open source project. As you said, it's all over the place. Why go to Redis other than the fact that you guys sponsor the open source project? Yes. So, I mean, one route in is you're in a regulated industry and you need to have a relationship with the vendor. And that's one sort of traditional way of you build that relationship between the vendor and the customer. But, you know, when I look at it from a technology perspective, there are criteria that the business needs in order to meet their commitments to their customers or to other business units.

And that's all around SLAs, around availability. It's around throughput and performance. And so, in a very small Redis cluster, you can do 200 million operations a second, right? So, the question then becomes, what can I now do with that? How does that change the way I think about my business process or how I engage with a customer or how I can optimize and schedule traffic? So, for example, you know, you can imagine you're a cell phone operator and you've got trucks moving around, right?

Let's say part of your network goes down. Would you want to route that truck to go and do a home install or to go fix the network problem? Now, if you can reroute dynamically, you can also then inform that customer that's waiting for an install that the person's going to be 20 minutes late. So, is that a feature of native Redis or is that value add that Redis Labs adds on top of Redis? So, what it's allowing our customers to do is look at a greater data set and a number of different feeds so that they can make more sophisticated answers.

And that may be because it's a better routing. It may be a better credit score. It may be a better personalization experience. It's all a question of being able to look at that big graph of data for the relevancy and then be able to actually do more sophisticated processing. So, I get the sophisticated processing. One of the things that I run to consistently is reliability as a challenge when it comes to databases being the underpinning of mission-critical applications. And I hear open source.

So, open source I get at Linux. Linux has gotten to the point I trust Linux. However, when I'm talking about databases, I understand some of the world's biggest businesses run their mission-critical solutions on Redis. But I don't have a bunch of PhDs to make this thing work. How do you guys help me to make sure that it's reliable? Great question. So, there's two ways we do that. So, Redis Labs as a company, we started off as a cloud service for Redis.

And so, the last eight years, we've been running and managing hundreds of thousands of databases for thousands of customers. And we built up a huge amount of operational experience, which we then bake into the product. So, those customers who don't want to use cloud, they can take the same product we run in the cloud, and they can run that on-premise. Or they can run it in their private cloud instance. And so, what we've focused our attention on is all of the technologies and all of the edge cases so that we can actually give five times of availability out of the box without having to have people's PhDs operating in there.

I was talking to a customer yesterday, and they said, you know, we love Redis because we installed it, we set it up, and we haven't had to touch it since in 18 months. And so, this is, you know, the benefits of being born in the cloud has allowed us to instill all of those operational things that we have learned running at scale into the product that benefits people who may just want to run it on-premise themselves. So, for those of us that are new to kind of the AWS world and the Redis world, what's the relationship?

We're at reInvent. Why are you guys at reInvent? I mean, it seems like AWS has their competitive view of what Redis is, and you guys are here at reInvent talking to customers. What's the conversation like? What's the value pitch? Great question. So, you know, we're here at reInvent. AWS is one of the cloud providers that we run on. We also work on GCP and Azure. We announced earlier this year a partnership where Google would actually promote Redis Enterprise as a first-party service along with their own Redis offering.

And so, we work with all of the cloud providers. And so, one of the value pitches is hybrid cloud deployment. Some of it you want to run on-premise. Some of it you want to run on cloud. You may actually want to span both of those. If you've got a transition strategy to get from on-prem to cloud, well, Redis is going to help you doing that because it's going to allow you to cache the data on both sides so you can maintain the application performance as you're moving both the application load and the data load between those deployment characteristics.

But you need to do that in a reliable way. And so, technologies like ActiveActive give you that five nines of availability across those domains. Now, most people will tell you cloud deployments are much more unreliable than traditional on-premise, and that's absolutely the case. And so, this is when technologies like Kubernetes actually help those migrations into the cloud deployment because they simplify the automation of the day-to-day operations. One of the things we announced recently was an extension to our Kubernetes operator that would do automatic cluster recovery.

Again, this alleviates the burden for day-to-day operations because it litches with set and forget it. So, Alan, I really appreciate you taking out the time. You're busy reinventing the schedule. We're all the way over here at the Cosmo. If you're not familiar with the setup here at reInvent, it's spread across like eight different hotels, and it takes 45 minutes to go from one hotel to another. So, I appreciate your time coming over from the Venetian. com, or follow us on Twitter at CTO Advisor.

Until the next CTO Dose, talk to you then.