In this episode I speak with Patrick McFadin about Datastax and their plans with Apache Cassandra, ponder on the release of MIDI 2.0 after 37 years, plane security, what happened in Iowa, and more
Chris Ward: [00:00:00] . Welcome to the weekly squeak. Your weekly geeky squeak with me chris Chinchilla a slightly short show this week. Basically, it’s going to sound lame. I finally decided to take my older MacBook pro. I mean, not old, I mean it’s one of the newer ones, but one of the first generation of the newer ones. Yes.
[00:00:26] The ones with the dodgy keyboard. I finally decided to go and get the keyboard repaired under Apple’s extended warranty program, and I’m using my backup windows Linux machine. And Oh my, how difficult it is to switch operating systems when you’ve got used to one for decades. So I am very, very unproductive right now, trying to figure out how to set things up and how to use things and all sorts of wonder that you suddenly realized you need to set up when you haven’t set up a computer from scratch in nearly 20 years anyway, and hopefully also the audio quality here comes out.
[00:01:03] Okay. I know my interview later. Which is with Patrick McFadden of DataStax, um, of, um, uh, Apache Cassandra amongst other projects. Um, I suddenly realized audio hijack, which I normally know use for recording interviews, does not exist on windows. Um, so we just took the zoom recording, which. It is not going to be as good quality as it normally might be, but the interview was great to so hopefully stick around for that.
[00:01:29] Enough complaints, enough disclaimers. Let’s get started with my links for the week. First, an article from Andrew Leonard on a one zero this is entitled, the blockchain is a reminder of the internet of failure. Now, this article may not necessarily be what you think effect. It’s quite a pragmatic article.
[00:01:49] He’s not necessarily criticizing blockchain as a concept or as a technology, but more as a way that engineers have a tendency to create solutions without ever asking people what problem they had and how that the internet tried to do this in the first place and then kind of reinvented itself. And that, um, blockchain is kind of a manifestation of this engineer’s a way of thinking.
[00:02:12] It’s not even a particularly new article. I’m not entirely sure why it popped up on my medium digest, I guess, because I’d be looking a lot of blockchain articles maybe. Um, but I enjoyed it. And, uh, if you feel that way, and actually this, this harks back to mini interviews and conversations and articles I’ve covered in the past is.
[00:02:30] Aspect of making sure that you solve problems people actually have as opposed to just creating technologies that solve no one’s problems apart from your own and maybe a select handful of other individuals. So enjoy your comments, please. As always, you can find contact firstname.lastname@example.org slash contact I’d love to hear from you next somewhat fitting.
[00:02:53] This is from Zed dinette, but obviously it was widely reported. This is particularly from a friend of the shows, Steven J, a Vaughn Nicole’s, and this is an article about the Iowa caucus app and what went wrong, and I heard more detail on this, I think on a Tom Merrick DayTech new show, actually, uh, when they went into a bit more detail about the companies behind it, these companies have, and he was like, shadow this.
[00:03:17] I mean, you couldn’t make this up. Um, and. How it was completely untested, how there was no backup, and it was a real, it is a shame actually, how every time there has been a proposal to use digital solutions for voting, it has gone terribly wrong. Meaning that, that anyone’s chance of wanting to do it well is always set back by another few years.
[00:03:41] Um, and maybe one day we’ll actually get this right and there will be online or digital voting. But in the meantime, I think we’re going to be sticking with paper a lot longer. It has its flaws, but I guess it’s easy to recover from. It’s easy to implement. It’s lowest common denominator. Everyone can understand it, et cetera, et cetera.
[00:03:57] But it’s such a shame that this story or this, this, this case, I suppose not only was a negative for digital voting, but also for the Democrats as well, and gave Trump a lot of ammunition against them and how they couldn’t organize a. A real piss up in a brewery. Maybe that’s not exactly what he said. I don’t know.
[00:04:16] Potato. So it’s a shame on many, many counts. Um, and the, and the company behind it. I think I used to work in agency as well, and I like to think I worked for a good agency. But how mediated is just kind of pump out things for as little money as possible and don’t necessarily build the best quality they could either.
[00:04:35] So a whole combination of. Problems there leading to this unfortunate situation. Next, an article on quartz from Dan . There’s a wonderful, wonderful article and I, there’s been a lot of this in the news recently, I guess because of the new version of the release after 37 years of Amity 2.0 and media was also featured.
[00:04:55] In an hour book. I did not enjoy as much as I would hope. Um, you, what was, it was a, you a a, not a pro, I can’t remember the name of the book, but, um, how algorithms are running the world basically. And he complained about MIDI there as well for sanitizing music. And the article goes into this little bit, the current or the old implementation of MIDI.
[00:05:17] Was very restrictive and kind of dictated how music happened, especially when media was very popular in the 80s lesser solely around the seventies and 19th as well. And how many people kind of recreated music to suit the style and the media suited keyboards and not stringed instruments or processed judgments and the subtleties that were lost immediate because of it, but no longer MIDI to actually, um, I was.
[00:05:43] It goes deeper, ed, a lot more subtlety and maybe, I mean, media is obviously had a bad name, well, not a bad name, but a reputation for such a very long time that will this version changed people’s opinions and how quickly will software and operating system and hardware operators incorporate it to mean that it can finally become an expressive protocol?
[00:06:02] Uh, I’m interested to see, actually, I’d be interested to follow. The, the, the progress of it. And when devices start coming out to support it, and then software of course, and, and what kind of creations emerge? I remember the, I remember when Wayne, I guess they still made like MIDI guitars and things, but what was very odd, and it’d be interesting to see how that would happen now because obviously with, with guitars, unlike piano’s traditional pianos anyway, you can’t get it.
[00:06:28] You can get like . Bed notes and quarter note and all sorts of things with strings and wind that keyboards don’t like to do. So I’d be interested to see how many 2.0 represents all those happy birthday, happy, happy anniversary of the 37 years. It’s quite amazing. Next article on dig, actually a website I haven’t ever really mentioned before, but this is.
[00:06:48] From a business financing dot. Co. Dot. UK originally, and I’m not going to go too much detail here a bit. It, I love maps and this is a map of all the oldest businesses and every country on a map. And actually, um, I was expecting to have heard of some of them and I actually haven’t heard of most of them.
[00:07:05] Um, and it’s interesting to see per continent are their kind of companies they are. So in the new world, they tend to be primary resources or banking, and in the old world they tend to be actually bruised is one. Just there is things like that proving that people have always needed to drink, uh, when they went to the bank or when they went to work.
[00:07:23] I suppose if ancient mapping and history kind of interested you, then take a look. And I suppose it’s interesting to see the. The differences between the get the new world, Australia, America, Canada, et cetera, and the old ordinary, the ages, the relative ages of those businesses as well. Anyway, take a look.
[00:07:39] It’s a nice little map and not much more to really mention, but just take a look and, um, yeah, uh, do already have these businesses familiar to you would actually be an interesting question to answer on the Kristen chiller.com/contact cause I haven’t heard of most of them, but maybe you have an article from, uh, Danny van Cooten, just his private blog.
[00:07:59] And this is fairly Swift. I would have actually liked to have maybe seen a little bit more detail, but, um, it’s certainly an interesting opening conversation on a topic that we kind of know about, but we ignore, and this is co two emissions of websites. Um, and. And how he went into to some links to try and reduce the CEO to emissions of his websites as much as he could, I suppose.
[00:08:21] I mean, if you’re running on shared hosting or any kind of cloud hosting and you can’t even really guarantee where your site is going to end up, it’s like, if you say, so here in Germany, we often have to split water and, um, heating bills between the whole house. So if you save, but then someone else uses a lot, then.
[00:08:40] I don’t know how much have you accomplished, and I wonder if it’s similar here. Your website runs perfectly, but everyone else in your sort of hosting providers, VPs is running inefficiently. Then what have you accomplished? Or if they run their business very badly, what have you accomplished? I don’t know.
[00:08:55] But it’s a fascinating thing to start thinking about. And how you can optimize your websites, reducing dependencies. Every tiny kilobyte that you reduce means less CO2, I guess. So if you’re interested in starting this conversation, it might actually be someone I try to do an interview with on the show actually, um, that, uh, take a look in them.
[00:09:14] I’d be interested to, to hear your feedback on how realistic you think this is and what real impact it will have on co two emissions. And finally, I’m rounding off with two airplane stories. I don’t know why I should work. You get these airplanes doors, but the first one is from scientific American and the headline is slightly scary.
[00:09:32] But, um, I mean, I suppose we have decades of proof to make us feel a little more comfortable. And if it’s called, no one that can explain why planes stay in the air or to be more precise. There are two schools of thought about why planes stay in the air and no one has quite decided which one is true. We know how they take off, but we don’t know how they stay.
[00:09:51] And that’s something I, a scary thought, but quite amazing to think that something that is used so widely and we know really very little about how we accomplished what we accomplished. And whilst that’s a little bit scary, it’s also kind of cool and kind of amazing to think about it. So, yeah, if you’re not about to fly soon and don’t want to, to be scared of the concept, then I have a read and I suppose this ties nicely into this year two emissions.
[00:10:18] But anyway, um. Yes. Uh, how, why do you think pine stay in the air? Which opinion do you go with aneurysm? Great. That is a scientific website too. There’s great details about each theory, and I’d love to know which one do you think is true. I’m not quite. Sure. Yet I hadn’t ever given it much thought. I think generally I don’t want to give it much thought.
[00:10:41] I suppose it’s one of those things where sometimes you’re just like, I don’t actually want to think about how this is working, but it is, and I will just be happy with that fact. And finally, another article from Zed D net, but covering it slightly different angle from Katelyn campaigner a are called another slightly scary title for anyone who travels a lot by plane.
[00:11:02] I’m the three of the top 100 international airports pass basic security checks. And before we go any further, let’s, let’s just recap who those three are, where those three are. And uh, Europeans, we’ve very happy because it is in, uh, Amsterdam. Helsinki airport and Dublin airports are three European airports.
[00:11:19] And, uh, I have actually been to two of the, I had been to all three but two regularly. And they are very nice airports actually. So this was more about cybersecurity. The apps. Um, systems and things like that. Um, and whether they use best practices for cybersecurity. And I mean, we, I think we’ve all singing in how shonky some airport applications and technology are.
[00:11:40] And I guess, you know, everything is as strong as its weakest link. So if I’m an airline is focused on security, but then the airport security around it is weak, then what have you accomplished? So, yeah. Um, interesting. And is it included some very basic things like using HTTPS, which you would hope was just standard these days.
[00:12:01] A proper setup of email servers using firewalls and it removing of vulnerabilities from mobile apps and antibiotics with you. Like this is kind of. Maybe there’s this feeds back into the Iowa caucus problem. These feel like such basic things to be checking in. Any developer should be checking in. Is it just money saving?
[00:12:19] Is it incompetence? Is it not updating? What is it that is causing these problems? If they seem such basic things to fix, pretty much every service recommends fixing these sorts of things now. So how do these problems get created in the first place, I think is the more concerning thing to think about. And they’d be, next time you pull, you don’t flow from any of these airports, which is leaves quite a lot left.
[00:12:41] Maybe you should complain, maybe you should point out the palms they have and the vulnerabilities caused and complained. And I know my own experience, I’ve complained to airports and if it gets you very far, but you can give it a go and maybe enough people will make the industry change and become better and follow pretty basic best practices, some of which.
[00:12:58] Well, not even take half a day to fix. So really some with some of them, there are no real excuse. Now, enjoy my interview with Patrick McFadden of DataStax, where we talk about Cassandra DataStax, the company, and some of the other products that DataStax producers. And uh. I maintain, enjoy.
[00:13:18] Patrick: [00:13:18] I’m Patrick McFadden and I work in developer relations.
[00:13:22] I also work in the Cassandra community in the project and user community. Um, I have for 10 years more, um, since early days at Cassandra. So I’ve been around since, uh, I mean I know all the characters in here cause I’ve been around with them. Um, DataStax was a company that was built. Specifically for Apache Cassandra support and services, and just being, you know, as, as open source projects emerge and, and become mature, there’s usually a commercial company that pops out, and I think that’s a pretty standard trend.
[00:13:58] Well, we’re, that commercial company that popped up for Patrick is Sandra. Um, our co founder, Jonathan, was the, the first project chair for Apache. Cassandra. Um, and we actually started out as a company called Rick Tano, and later changed her name to DataStax. But that’s what we’ve been doing for the past, say 10 years.
[00:14:18] But, um, just, uh, trying to grow Cassandra into the biggest, baddest database on the planet. And, um. Figuring out how to make, bring it to enterprises, which is tough because, you know, the open source software fails at the enterprise gateway. Uh, and finding a way to make it easier for them. Mostly it’s around support and services, um, but also building our own distribution that, um, that hopefully solves a lot of the ticks, a lot of boxes.
[00:14:47] But, uh. And we may have multiple things that we do now that are in support of that. Like, um, we’re getting ready to launch. We’re in beta with our cloud product, which makes that easier. Um, we have some other products that are coming soon, like app stacks, which makes the data modeling better. So we’re just continuing to advance the needle, uh, on all the products that are, that are required or needed.
[00:15:12] But. Um, probably what’s most interesting for you right now is we have a new CEO. We have new direction, and it’s pretty exciting.
[00:15:20] Chris Ward: [00:15:20] Uh, I think I’m more interested in digging into some of the technical stuff, but let’s go back a step.
[00:15:25] Patrick: [00:15:25] I will go into the technical stuff because our CEO is very technical.
[00:15:31] Chris Ward: [00:15:31] Let’s just go back a step. Um. Cassandra is reasonably well known, but let’s go back a step. So from there, I can’t quite remember. It was born out of LinkedIn or Facebook. I think
[00:15:42] Patrick: [00:15:42] face Wells, it was two engineers at Facebook. Uh, Avinash and Prakash were engineers there. And, uh, the, the fun part of that story is Avinash was, uh, one of the original officers of the dynamo paper.
[00:15:55] Chris Ward: [00:15:55] Hmm.
[00:15:55] Patrick: [00:15:55] Okay. And he, and he worked at Amazon, went to go, you know, this is how engineers go. Right. And say he was an Amazon, went to go work at Facebook, um, said, Hey, that dynamo thing was really cool that I worked on it at Amazon. But I think that, uh, I think it’d be better if we. Change the data model. So they developed a, around the dynamo model, the dynamo, not dynamo DB, like you can buy it
[00:16:19] Chris Ward: [00:16:19] dynamic dynamite that,
[00:16:20] Patrick: [00:16:20] yeah, the data bookkeeper.
[00:16:22] Chris Ward: [00:16:22] Yeah. People assume they’re the same thing, but I guess,
[00:16:26] Patrick: [00:16:26] no, and it’s funny because I think that that was like when I think they just said, wow, that’s an awesome name. What can we use it with? Um, and because it’s not the same, uh, it’s not the same database. But what’s cool about it is it’s. Um, I think that they tried to take some of the same concepts.
[00:16:43] It’s, uh, dynamo paper was originally a key value store. Yep. It was trying to answer the question is how do we keep a database online 24, seven, um, and deal with enormous amounts of failure or like an entire data center going offline. Um, because if you think about it, Amazon’s in the business of making money and from everything I’ve read lately, they’re doing a good job of that.
[00:17:05] And. When you’re offline, you’re not making money. Yeah,
[00:17:08] Chris Ward: [00:17:08] sure. Let’s, I actually, there’s a lot on the, the, the usage side I’d like to dig in, but I’d just like to get one more bit of the backstory. Cause there’s interesting in that with most Apache projects, the, the company and the Apache projects, uh, generally.
[00:17:27] Fairly well connected, but it’s interesting that DataStax was not the company that originated the project. Um, or not, not the initial company. Anyway, I’m sure you’ve been involved from very early days, but. Is there a particular reason that DataStax kind of stepped in as the main, um, maintainer slash kind of overseer as opposed to Facebook, or is that the company culture or,
[00:17:56] Patrick: [00:17:56] well, Facebook is not in the, there lately have been.
[00:17:59] A little better stewards of open source projects like
[00:18:03] Chris Ward: [00:18:03] I was looking [00:18:03] Patrick: [00:18:03] for, yeah. Yeah. Stewards. Yes. That’s what we, that’s how we call it. We don’t call it owners. There’s no owners and open sellers. It’s stewards. Um, they, Facebook has done a better job lately. Um. You know, I think there’s, there’s also some debate there, but that’s not their business. [00:18:21] You know, that it runs their business and they’re happy to make it an open source project. Um, but as I mentioned, uh, the reason DataStax exists is because the project chair, and this is kind of, this is kind of a fun and interesting part of it, is you think a company like Mongo DB, um, or Mongo who has Mongo DB, they own the whole thing. [00:18:42] It’s not an Apache project. It’s theirs that they open source and they can do whatever they want. They own the trademark, they own everything. So that’s, that’s pure top-down open source company. Apache. Cassandra is an Apache project. It is owned by Apache software foundation. Um. And, uh, as I mentioned how DataStax got involved. [00:19:03] Well, well, dataset became a company from the project chair of Apache, Cassandra, Jonathan Ellis. Um, so there was a, there was a connection there. At that point, Facebook had not put any energy into Cassandra and Abinash and Prakash. Uh, that’s why they donated it to the Apache project. [00:19:23] Chris Ward: [00:19:23] Okay. Interesting. And just out of interest from DataStax perspective, you just mentioned you’re coming into beta with the cloud, uh, offering on top, which is, which is a difficult time to be doing this. [00:19:37] We know we could, we could talk about that later maybe. Um, but, so what does DataStax being doing. Apart from that in terms of a monetization strategy for the core open source project? [00:19:51] Patrick: [00:19:51] Well, support services are, are more or less the, that’s the bedrock of open source, trying to make money around open source, uh, you know, and we, we’d sell a licensed product. [00:20:01] It’s enterprise and it adds a lot of things like, uh, particular security things, integrations, um, those sorts of things. I mean, I think primarily what, um. W when you talk to a large company, uh, you know, a fortune 100 company and they’re looking to implement open source, they get really nervous about just doing it on their own. [00:20:24] Um. And they, they have the resources, they can, they can put engineers to it. And that happens. Um, Apple employs a lot of Cassandra commuters because they have that kind of cash, but they, you know, they’re also, um, in the mind, uh, well, we can also rent that too. Um, and. So when when we work with enterprises, they’re like, we, we believe in what Cassandra is. [00:20:50] I mean, we, how the technology works is awesome. Always on scales like crazy, multiple data centers. Perfect. That’s what we want, but we want a little more and we’re willing to pay for it. That’s a perfect opportunity for DataStax to create a great relationship. Um, the database isn’t going to fail you and we’re going to make sure you do it right. [00:21:10] Chris Ward: [00:21:10] Okay. And I mean, we know this space is, is relatively busy, the sort of no sequel space. There’s still, it’s not as much as it used to be, but still kind of new offerings emerging. But Cassandra does have something of a reputation for. Working pretty UL on large data sets and obviously comes from, from, from creatives who are used to it. [00:21:35] Uh, I’m looking on the Apache page. Uh, and you mentioned Apple. They have, and I think I remember when they switched over to Cassandra 75,000 nodes, which seems. Ridiculous. [00:21:47] Patrick: [00:21:47] I think the latest one is like 150 or 200,000. Yeah. There. So a DataStax accelerate, which is our user conference in may. Um, Dinesh, he’s like the spokesperson for the iCloud engineering team that runs Apple. [00:22:01] Um, I’m waiting for he, and he does this every time. The first slide he puts up is all the numbers. And they’re always ridiculous. I mean, petabytes of storage, bazillion nodes. I mean, they, it’s kind of fun. I mean, it’s one of the highlights of the conference for sure. [00:22:17] Chris Ward: [00:22:17] It’s actually quite interesting looking at these numbers because Apple has 75,000 nodes with 10 petabytes, and Netflix has two and a half thousand node with 420 Tera of terabytes. [00:22:26] Sorry, I missed, I missed a few zeros there. That’s okay. Yeah. So, uh, and then added to that is also Sen, uh, get hub. I mean many others, I’m skipping over a few here, but just ones that, especially a technical audience will, will understand and appreciate the kind of data throughput they’re probably going through. [00:22:49] So [00:22:50] what [00:22:50] Chris Ward: [00:22:50] do you think at the core, what do you think made Kasandra so well designed to handle this kind of capacity in the first place? [00:23:01] Patrick: [00:23:01] That, and that is exactly what Cassandra is, is the way it was designed. Um, and this is the magic sauce is how it started. And I’ll go back to the dynamo paper. Um, it’s really the law. [00:23:13] It’s the only data. It’s the only database the grew up and made it out of the dynamo paper that has usability. I mean, it’s usable. So, um, the, the three things that are really important are the scale. Um, the way it scales, I mean, it can go from zero to petabytes and, uh, it’s designed to do that without having any downtime. [00:23:35] You can do this online. Um, it’s the speed. Uh, it is really a fast database. Um, very rarely do people say it’s not fast enough. Um, and then, and then in the resilience, um, you know, the, the way that it replicates data and the, because it’s an, um. availability in partition tolerant database. That’s, that was the, the two, you know, the out of the cap theorem, uh, consistency, availability and partition. [00:24:03] Cassandra was built around availability and partition, um, with tuneable consistency. That AP part is so a, think of every business needs availability. And partition tolerance because this is the world we live in. Like I said, Amazon had it right, like we’re offline, we’re not making money. And you look at all the list of companies there. [00:24:25] They were like, yeah, we can’t deal with an hour of downtime because our database went sideways, or we lost a data center. Um, you just have to be ready to deal with failures. So Cassandra started with that. And what’s fascinating to me as you see databases now. That, you know, older databases, you know, like relational databases that are trying to glue on these features, um, like replication and things like that. [00:24:54] And it’s just, it’s a mess. It’s hard to work with and it isn’t very resilient because it’s an afterthought. Um, that wasn’t one of, it was primarily designed to do so. When you look at that list of companies, they’re like, yeah, we, this was, this is what we need and we’re going to make it work. Um. And that, and you mentioned that, you know, it’s been a long time. [00:25:17] That’s the thing is we’ve proven it. It’s been, it’s been in the most critical production workloads on the planet and it has held up just fine. It’s proven, and I love it when people throw FID at Cassandra. I’m like that. That’s like saying the moon doesn’t exist. I mean, come on, or the earth is flat. Well, believing in something that doesn’t exist. [00:25:42] Chris Ward: [00:25:42] Yeah. Let’s don’t dig too deeply into that one video. [00:25:45] Patrick: [00:25:45] Yeah. I mean, you can, when you hear this stuff, it’s like, well, Cassandra will fail and it’s not good enough, and I’m like, are you kidding me? Can we get over this now? It’s great [00:25:57] Chris Ward: [00:25:57] on on that point though, one, one of the facts on the. The Apache page that jumps out at me is this. [00:26:04] Every note in the cluster is identical. Yeah. I haven’t counted a few distributed databases that claim this. And it’s always kind of, it’s always not strictly true. Uh, is that completely true with Cassandra? There are no concepts of, um, masters and Nope. Yeah, I [00:26:25] Patrick: [00:26:25] know it’s a more [00:26:26] Chris Ward: [00:26:26] appropriate words these days. [00:26:28] Patrick: [00:26:28] Yeah. The leader follower, [00:26:31] Chris Ward: [00:26:31] no one’s quite decided on though. The consistent replacements to those horrible words yet, but yeah. Anyway. You know what I mean? [00:26:38] Patrick: [00:26:38] Yeah, yeah. There’s no elected leader and, um, though any, any, um, you noticed a very important thing. All of those databases have elected leaders are more, they have consistency first. [00:26:49] So they’re like CP database or CA databases and, um, and which are. Most, or if not all of the other databases in the world. Um, so of course, but no, Cassandra is definitely a shared nothing architecture. And so, um, of course there’s, uh, they’re not soup and not like every node stores, all the data that would be. [00:27:11] Ridiculous. Um, but it’s architected in a way that each node stores a certain percentage of the data plus replicates from other systems. And it’s a very consistent way to do it. We use consistent hashing to do this. Um, it, it, uh, is deterministic. We can, we know where things are. It’s not random. Um, but, um, that the way that basic architecture of that ring architecture works really well. [00:27:36] And yes. So if you lose a node. It doesn’t bring down your database. Um, and it’s really, it’s, I used to do this thing with raspberry pies. Um, I’d have a small raspberry PI cluster set up and I would have it, you know, have an outside client, like pounding away at it, and I would take a hammer and smash the shit out of one of the nodes. [00:27:57] And I mean, of course for a fact, but everyone was like, what happened there? And the application doesn’t even blink. [00:28:04] Chris Ward: [00:28:04] It doesn’t mean a flinch. Yup. Yup. This is a, I have seen a few distributed databases, uh, actually maybe I seen you. Do that. [00:28:13] Patrick: [00:28:13] I have, you might’ve seen me do that. I did it on stage with a fire ax. [00:28:17] Yeah, that’s a blender. [00:28:20] Chris Ward: [00:28:20] Just pull a knocker a pie out of a cluster. But destroying it is more memorable and I have definitely seen someone do that. I did it at [00:28:29] Patrick: [00:28:29] Oz con, uh, one year, and it was, it was almost semi controversial. People were really upset about how violent it was. Like. All right. It’s a raspberry pie. [00:28:41] Chris Ward: [00:28:41] Anyway, [00:28:43] Patrick: [00:28:43] we had a drill and we were drilling through them. I mean, it was fun. [00:28:47] Chris Ward: [00:28:47] I would shut it. I’d actually like to see what the pie is doing whilst you’re doing that as well. [00:28:53] Patrick: [00:28:53] Yeah, there was some, we had to really, we had to test it a couple of times because what I was worried about as a shorting and causing a fire, and I’m like, okay, we can’t have a fire in expo [00:29:00] Chris Ward: [00:29:00] hall. [00:29:02] Wow. I’m actually frankly amazed at something that small would cause a fire anyway. [00:29:06] Patrick: [00:29:06] Well, you know, electricity. Yeah. And this is, you know, that’s what I’m worried about. I’m not worried about the database going down. I’m worried about a little node causing a fire. [00:29:15] Chris Ward: [00:29:15] Other than that, that’s a headline in itself. [00:29:17] Then you caught fire database. Okay. [00:29:21] Patrick: [00:29:21] See, all right, Chris, you and I get along great because I am in the PT Barnum school. There’s no such thing as bad news. [00:29:29] Chris Ward: [00:29:29] One more quick question on Cassandra itself, and I’d just like to ask you a couple of quick ones on DataStax a bit more. And that I think maybe, maybe one of the few negatives, uh, especially for people starting with Cassandra is this custom query language. [00:29:45] And I just, I mean, to be honest with you, I think back in the day this used to be in comparison to sequel. And I don’t know how many big application users these days are really coming from SQL backgrounds anyway, so maybe the comparison doesn’t matter so much, but have you noticed any kind of difficulty in people. [00:30:05] Understanding that or all the sort of SDKs and rappers so mature now that most people are just doing queries through those. [00:30:12] Patrick: [00:30:12] Have you been a fly on my wall? This is a current conversation for sure. Um, but I’ll tell you the CQL language because intercore language was. And I was involved in that design early days because it’s meant to be familiar. [00:30:28] Instead of having, Hey, let’s have yet another query language out there that is just completely unfamiliar. Let’s build it as a subset. Now there’s good and bad. Um, but my, my hypothesis has always been. Everybody comes from relational. Everyone knows relational databases. So you know the syntax, you know how to insert data, insert in, you know, insert into table where, you know, values were, I mean, all these things are there. [00:30:51] Select from where that’s a pattern. You understand. Um, I think what gets people into trouble is then they start thinking, well, this is SQL. Then they start doing things like joins or full table scans. And so I spent a lot of my time educating users on and we start our training. Um, we do these workshops all over the place, these free workshops. [00:31:13] And, um, the first thing I say is, all right, I know you all have SQL background. Awesome. I’m going to explain how to go from there to here in a language you are familiar with and I’m going to show you all the sharp edges and explain why this is the right language. You get the data out. Um, and it, it just goes and really we could skip the rest of the day and just go into the primary key, how it works. [00:31:37] Um, you know, I mentioned the primary key or I mentioned like how the data is distributed. That primary key is the key to your success. And I think what you, there’s a light bulb moment once people get, ah, that’s what, that’s what the primary key does. Um, and this is how it works for me and my data model. [00:31:56] Then we’re off to the races. [00:31:58] Chris Ward: [00:31:58] Okay. And coming back to DataStax, I see that, I mean, whilst the cloud offering is coming and the enterprise offering, you’ve mentioned, I can see you have some other things as well. Uh, you have a graph database. Intriguingly, you also have CAFCA connectors and a few other, and also Kubernete is in containers, tools. [00:32:20] I mean, um. I think he’s basically just from OBS observing how people usually integrate Cassandra into their application and you figuring out the places where you can help with that or are these very separate offerings. [00:32:36] Patrick: [00:32:36] There. Yeah. It’s, if you think of anyone deploying Cassandra, of course, it’s not an Island. [00:32:43] It doesn’t just exist with no other connections to the world. Um, Kafka is a great, great example of getting data from point a to point B. We’re probably going to be point B. um, so let’s make that work. We also have, it also uses our CDC on the. Uh, where you, when you insert data into the database, it is committed. [00:33:02] It emits data. So, um, we also feed into a Kafka queue and so into a topic. So, um, I mean, the Kafka is an important part of our ecosystem. The Cuban, anything is developing and stay tuned. We’re going to be a cube con with a bunch of stuff, but, um, that. I think that’s just, we know that that’s the winner, good or bad. [00:33:25] That’s the winner of deploying large infrastructure, and if we’re not supporting it in a really important way, then we’re not helping people get it done. Um, so there’s more coming there. That’s actually one of those things that’s, we’re really working hard to. Improve and change the story on Cassandra and Kubernetes. [00:33:44] Okay. Yeah. [00:33:46] Chris Ward: [00:33:46] And what’s the graph database thing? [00:33:49] Patrick: [00:33:49] A graph is, uh, is an interesting debate. I was just in graph day in Texas and you know, it’s, it’s, there’s this raging debate right now is graph a, is a graph, a product or a feature. And, um, I think without a doubt. So our graph, what our graph is, is, um, it’s a top down graph, uses gremlin, ticker pop, uh, open source, um, very, this is very much rooted in open source and other project. [00:34:16] We support two projects. Mainly it’s tinker, pop, Apache, ticker pop, and, uh, Patrick is Sandra. But, um, we’re looking at doing things like how can, how can we use graph. In really interesting ways because it is very useful. I mean, um, and this is the quick takeaway on what graph is. And my main opinion is, you know, where Cassandra is, um, is completely de-normalized datasets. [00:34:43] Um, graph is infinitely normalized. Everything’s a relationship. So you’re covering, I mean, you’re, you’re, you’re railing on both ends of the spectrum when it comes to data models. By that. Um, that to me is the most exciting part. [00:34:59] Chris Ward: [00:34:59] I wonder, is there much crossover or do people tend to implement both or one or [00:35:07] Patrick: [00:35:07] it’s, yes. [00:35:08] That’s what we continue to see is it’s not, that’s why it’s like, it’s a debate. Is it a feature or a product? Because, um. Very rarely do we see people just implementing graph only. I mean, it does happen, but it’s like, Oh, I want to, I want to map a relationship to something and then go deeper into the data with something like, [00:35:30] Chris Ward: [00:35:30] okay, and aside from CubeCon, which I now will be at as well. [00:35:35] Excellent. Look coming from there in a few weeks time. Um, what else is on the roadmap for the next six months? [00:35:44] Patrick: [00:35:44] Uh, for DataStax or, or for the project? Actually both, [00:35:52] Chris Ward: [00:35:52] let’s say both. Where they crossover [00:35:55] Patrick: [00:35:55] both of the crossover. Alright. Well, we, we are doing a lot, as much as we can. Cassandra. Ford auto is a probably the biggest frigging deal of the project ever. [00:36:08] And the reason being is because it is, um, I think all of the committers, the PMC on that project, uh, operators, it’s a biggest operators who have agreed that this has to be a, because it’s so much data on the world relies on it. The reliability has to be. The top. And so, um, what’s exciting to me, and I know it’s taken a long time, but it’s, it’s going through some of the most incredible testing you’ve ever seen. [00:36:35] And there’s some open source projects that are gonna pop out of this around correctness testing of databases. Um, here’s, here’s the interesting, and Chris, I haven’t told anyone else this yet. So you’re the first one. Um, the commitment in the project is the largest operators want to put Cassandra in production before they say it’s worth releasing. [00:36:58] They want.zero to be the best database that they can ship. And. And some of the testing goes on. Like here, we’ll move to move a petabyte of data. Now, did a single bike get lost in that movement? [00:37:14] Chris Ward: [00:37:14] Okay. Oh, there was, [00:37:15] Patrick: [00:37:15] let’s find out why. Um, you know, this is, this is why it’s exciting for me. It’s because I don’t think there’s any database in the world. [00:37:25] Is, is going to be this solid and steady. [00:37:29] Chris Ward: [00:37:29] Yeah, I’ve heard, I’ve definitely interviewed companies that are kind of doing this as a bolt on, but, um, [00:37:35] Patrick: [00:37:35] this is the first class. [00:37:36] Chris Ward: [00:37:36] Yeah, that’s, that’s quite interesting. And does that also relate to things like, um, uh, like sort of developer environments? The ability to, I don’t think necessarily aplastic Zandra, but the ability to. [00:37:52] Kind of collaborate on data. Like you can move version controlling code. [00:37:59] Patrick: [00:37:59] It’s not, it doesn’t go up. We’re talking baseline. Like, um, what happens whenever you’re in a new node is going into the cluster, like adding to the cluster and you turn it off. Okay. Yeah. Bad things happen. All [00:38:13] Chris Ward: [00:38:13] kind of chaos engineering element, I guess, [00:38:17] Patrick: [00:38:17] that it’s embracing chaos, like it’s your best friend. [00:38:19] Chris Ward: [00:38:19] Yeah, and I’m guessing there’s a lot of work there with Netflix. They kind of. Kings and Queens of that sort of domain. [00:38:28] Patrick: [00:38:28] Oh, absolutely. Without a doubt. And it’s, it’s funny because, you know, Netflix, I think, popularized it. There’s some, there’s some great companies out there, like gremlin is a company