Why developers love Postgres | BRK2064
All. Right hi. Everyone I am, Craig christine's hopefully, you're in the right session otherwise it's too late to leave now because they're closing the doors. So, I'm gonna talk a little bit about Postgres today quick. Show of hands to start off with how many people here already used and run Postgres. Those. Are my people thank you. The, rest of you hopefully I can convince you a little bit by the end so. Uh first, off Postgres is really well-loved believe, it or not it's ranked, as you know the, most loved database for, actually I. Think a couple years in a row now Stack, Overflow just, came out with 20/20 and. DB, engines is a popular, database engines ranking that, names kind of an official kind of the year of the database what was the big one that year and. It won two years in a row so, we'll, drill in a little bit so you know how, it had this resurgence and why it's loved by and particularly developers, and we'll cover a little bit by the by enterprises as well but. Know that it's kind of won over developers, Minds over the past ten years or so. Quick, show of hands how many people in the room would consider themselves a developer. Wow. A lot and the rest of you kind of, admins. Or DBAs or, decision, makers I would assume. So. There, was a been a big shift in the last five. To ten years I would say that developers, now are making kind of the decisions on the stacks to use the frameworks, in. The databases as well if you look at ten. Years ago it was, the DBA s that controlled the database they would say hey you give me the the thing you want to do I'm gonna create the scheme um and create the indexes, I'm gonna manage it all and over the last ten years we've had this big shift to, application. Developers choosing, the database and. Postgres is one that's really really well loved, by developers. Fred monk is a small industry analyst that talks about this a lot there's, actually a really great free book called. Developers or the new king makers it's. Like a hundred pages it's not super long I encourage you to read because it talks a lot about this shift to like developers, kind of making, more and more influential decisions, and driving products. All. Right so a little bit of history about Postgres because I think it's the interesting it's. An open source database but, I think it's unique and open sourced there's a lot of things that are open source like you write something you open source it but it can still be owned by a company open.
Source Just means it's out there and people can contribute, to it Postgres. Has a very unique license, and structure and that no one can actually ever own it it can't go and be bought by a company there, are individual. Committers that, contribute. To it from multiple, different companies but, no one can ever actually own the database so it will forever be open-source and kind of this community, shepherded, project, it, has the same roots as a lot of other databases so, the name Postgres. Comes, from ingress which is one of the earliest databases so, if you look at things like you know db2, and the tizi and other things they have the same, database. Relational. Roots as Postgres. And. Then it releases, really, really regularly on a yearly cadence so, you'd have to worry when the next version is coming it's fall. A full. Winter time of every year sometime. Between September, and November you're gonna get a new release those. Things get committed six months to a year in advance so we already know what next year is going to look like so while it's an open source project and doesn't, quite have a roadmap you can kind of tell what's coming with it so really interesting to be able to follow along and know what you're getting in the future. So. But why do people love it and in particular why, do developers, love it since they're making all these decisions about technology and, what goes into place. Here's. The really high-level summary and I want to drill into a lot. Of these in a little more detail but. It's it's really, really feature-rich, so. It's got a lot. Of data types really, powerful and advanced indexing, full-text search, I. Gengo. As a popular, Python framework and it has this concept of batteries, included, I really. Really like this concept because. It, says you know it has whatever I need to do things like it ships, like on, Christmas morning if the toy comes for my kids and it, doesn't have the batteries and now I've got to run to the store like cool, you gave me a toy but it did nothing for the kids this, idea that everything's in the box that I need is a really powerful one and with Postgres yeah it's a relational database but. As soon as you need unstructured, things it's there as soon as you need tech search it's there geospatial support, so you don't have to go and deploy a new database to get all these features so, let, me drill in just a little bit more so. Data types this is something that's really, really simple and you don't think about as a an, exciting thing when it comes to a database but I really, really like the post-crisis approach they're, really liberal when it comes to data types so, you've got things like text ones and numeric ones you've got geometry, types for geospatial stuff.
What, Really really useful data type is a race so. You're probably using a race somewhere in your application. Why map this to a join, within a database when you could just put it an array into a column so. If I got things like you know categories, or tags this is great for an array field and then, I can have the same kind of constraints, range. Types are really really really useful so, it's a from an a2 so in the database usually when you have a you know started. At into that now, I can actually put that into a single column and have various constraints like the ended at can't be after these started at and. I can actually enforce this within the database so it's all correct. We. Can limit the number of people's that can attend a you know a talk session occurring, at 11:45, so. We have too many people signed up in the room so, we can actually do this directly within the database really natively. A, big. Exciting one that people get all really excited about is JSON, be so. Postgres got support for JSON, it's. Going on eight years ago down but we kind of cheated so, when we said we had JSON support we. Accepted. JSON we validated it and we threw it into a text field that was with Postgres 9.2, so we said we had JSON support we were one of the first databases, to do it but, we, kind of cheated, JSON. B is a binary, representation. Of the JSON on disk so much like a Mongo, database it's. Compressing, it it, can index into it it's. Much, much smaller a, great. Great option B, stands for binary I have a colleague as jokes, that B stands for better I like, that approach if. You're using JSON, in your database or you have unstructured, data it's. A great approach. And. It works, pretty natively so if you look here like if I've got a products, table and I have you know the name of the product but I want to have all these attributes right ecommerce. This is really, common if you you know think of an e-commerce store like Walmart and, you've got all these products, and different characters. And categories, and sizes like, you just want this kind of key value structured, nested, JSON document, for it that you can display all these properties about the product. And. Then you can query it on anything right so you can query you on things like give, me the author or filter. For this category of fiction books really. Really simply natively, without, having to go build this to a relational, model but also without having to move to a document, database so, you've got this flexibility, of hey I've got my relational, structure data and I've, got unstructured data. So. Well I would encourage you know when you're doing something in your application, and it's a unique data type take a look at Postgres and see if it has that exact same data type there it'll, allow you a bunch of optimizations. A bunch of constraints you can ensure that the data is correct, really. Really really powerful here. So. Another big one is indexing. Postgres. Adds an index type about every year most databases have one or two index types post-crisis. Has this history of a new database kind of showing a new index showing up every year or so b3, is the really really common one that you expect. Gin. And just a, really, short simplification. Of them gin, is really useful when you have multiple, values in the same color so. If you think of like an array right where you've gotten multiple values in that array, you probably want gin if you think about indexing, JSON. Really. Really useful there as well, a. Quick. Look it's really really simple to actually you know I mentioned JSON right and we looked at how it works to, add an index and index every, key and value.
Within That JSON all I have, to do is create this one index now. Is like query that. Data it's gonna use this gin index filter. Be really really efficient, so, now I've actually just added a bunch of unstructured document and documents, to my database how did one index and then I can you know query this in a really really fast way joining, against my relational tables. And. You've got other ones too so SP just Brin really, really, useful for really large data sets and I say really large. Billions. Tens of billions hundreds of billions of rows that naturally, cluster together so if you think of things like, phone. Numbers that often have a you know a prefix, in kind of a location or zip codes that, kind of data naturally, clusters together, and. An SP just didn't bring to be really really useful. And. So, that's all what's in the box already with Postgres now. What I also, love about Postgres it's it's really solid stable database. The. Number one thing I care about my database is that it's good it works and it doesn't lose data if you have a database it loses data it's it's not so hot so. It focuses on being really, slow and stable when it adds a new data type and make sure like it's a standard, spec I've. Actually had debates with the the core community about like an email data type but, there's actually not a, spec. For what is an email address you, can put all sorts of crazy weird things in there and so that's the reason actually Postgres doesn't support an email data type because you can't validate it in the way you'd expect, but. You know we want we want these new shiny features, post-crisis. Is really unique in that it has this. Low-level extension. Framework. So, these are really really low-level hooks, this is just like an add-on on top this is deep within the core of the database that, you can come in and change how Postgres works so. You can add things like, h2, error was before, json B it's a key value store directly in your database post. GIS is a really, really rich, advanced. Geospatial. Database it is well regarded as the most advanced geospatial. Open, source geospatial database. And. Turns post-crisis into it's like a complete geospatial, tool if. You want partitioning, in your database you've got pdepartment and so, all these tools hooked deeply, into Postgres and change the behavior. We'll. Talk a little bit more about this but or 2pg, is a really. Interesting one to help with, migrating. From Oracle to Postgres so, if you've got this legacy Oracle, application, that you're looking to migrate you, want to see how compatible it is or how much work it is it. Can really help quite a bit. And. Then. You've got a big broad ecosystem so it's been around for twenty years people have been developing, against it. Time. Scale is actually a VC. Funded company that, builds an extension that turns Postgres into a time series database, so. If you're doing things that are time series metric space log analytics. Oftentimes. IOT it can be really really useful to compress this IOT data scan. It down aggregate, it that sort of thing. Sharra really. Interesting, if you're using graph QL and you don't want to have to build something from scratch it'll. Give you basically a out-of-the-box. API, with graph QL that you can work with. Then as you build. In evolve your application, and you want to drop down you can drop directly down, to the database and. Then. Just earlier this year at Build we announced support for Postgres for Azure data studio. Most, people with Postgres are used to working in a CLI I'm personally. A CLI guy it's got a really really powerful good CLI but, I know there's this big desire and kind of need for I want some graphical tools to work with and, so, we have two port riser data studio so you can work with both your sequel databases and your Postgres database is all, in the same way in the same place and collaborate on.
So. If you really think about it right you've. Got Postgres at the core we've got a bunch of extensions, we've got json b which is you know a. Really. Really powerful data type that stands all on its own really. Rich indexing, all. These features built in that i don't have to go to another database. When. I need this feature i can just say hey i can post chris work for me i don't have to go deploy something new and. Then when you come to as you're right you've got built-in eh-eh-eh on top of it the ease of scaling, up and outs. You've. Got all these extensions that are actually already supported, for you so you don't have to go build them and install them for your database it's, simply click a button and you've got this extension, now enabled, to do something that you couldn't before within your database. And. A. Big. Part of can I think the trend of application, developers, doing more and more with the database, is also. Like you're not gonna go and get a PhD in databases like, you're not gonna go learn about the roots of database theory necessarily, if you want to let's hang out and you know chat, and grab a beer after like I think it's super fascinating, but not everyone's, gonna do it so, what we've done actually is take a lot of the common, insights, that you need to do as a DBA things, like what queries are slow and how do i optimize them, and bubble. Those up directly for you within the portal so you've got things like intelligent, performance, that will recommend, indexes, automatically, a. Big. Pain I hear for Postgres so, post-arrest under the covers is just a giant append-only log when, you write data it writes a new record when you update a record it actually marks the old one is dirtied and then, writes. A new one Postgres. Under the cover has something called vacuum that comes up and cleans up all the mess. Now. Out of the box it should work pretty well but really depending on application, you could have to tune this tweak this and, I hear it's a pain for a lot of developers that like hey I've got a lab unch of bloat I've got a bunch of old data that's there consuming, waste how do I clean this up, we. Actually just announced, a, private. Or public preview of the, ability to auto tune your Postgres database so. That you, don't have to come in here and tune this and do this maintenance will, automatically, do it for you and tune it over time. And then of course you want it to be safe and secure so the ability to detect threats early so. We can detect alert. And alert you on anomalies I'll, make it easy for you to investigate is there unauthorized, access is, there something you should look at is there something not configured, correctly that, sort of thing so that you don't have to be an expert DBA to get these kind of insights. So. That's a little bit on I think, why Postgres, from the application, developers perspective like I want to leave this is the TLDR, like if you're. Not sure about something in here. Search, for it you'll find a lot of resources, this. Is very much the 30,000. Foot kind of flyby view, but. I'll give you you know what do you need to know what, do you to look at hey I don't know what that thing is searched, for it, to. Cover a few more of these foreign. Data rappers are super. Interesting, they allow you to connect from within Postgres, to another data source so. You can connect from things like in Postgres. To Redis, from, in Postgres to Mongo, there's, the, ability to query Twitter. From within Postgres, I don't. Know why you would actually want it but. That's that's a different story the, fact that you can is always a fun, opportunity right and as you think about these things like okay what new ones could I use and, could exist there so. If you're curious of this crazy idea if you want to you know do it you can probably dig in and it may be already done for you so. Shifting, a little bit. Why do enterprises, like Postgres, I, hear. More and more of a push that like it's, a well-respected database, it's used. Yeah. We're gonna go ahead and deploy some of it but. I think there's a lot of strategic, decisions, and, as I talk with analysts, all the time to like like Postgres, is the one that's kind of winning a lot of these new enterprise workloads net, new applications, Greenfield applications, a lot, of what we talked about there on the application, and history side right it's a really solid.
Database, It's flexible, I don't have to always go into ploy something new but, I think there's a few more reasons that are worth drilling into a little bit. So. The first right it's. The same roots as all, other relational, databases like the name itself. Postgres. Is posting grass so, any of these other databases that you probably had deployed had, those same roots it's not like this fancy, new reinvented. Kind of thing. Tran, a transactional, database is safe you trust it there's. A lot of eventually, consistent databases, out there I don't, want my bank, typically, running on one right I want, the data to be there to be safe and secure and to not lose it and. This. Is a one, that not, everyone likes equal, how many people here like writing sequel, oh. That's. Actually way more than I expected how many people like reading someone else's sequel, I. Saw. One hand we didn't, talk after I think yeah. But. Sequel is everywhere like it is the lingua franca of data it's how you access it so as soon as you have a new no sequel system what, are you gonna do you're gonna go layer a sequel, API on it to perform your analytics, like. Sequel. Works it is the most powerful way, to access. And work with your data, Postgres. Has you know very rich, anti sequel support it follows the standard pretty well. It. Works businesses, like this you don't have to learn a net new thing to go of how do I analyze this data. From. Most organizations, I talked to you. Already have a relational, database this isn't something new a DBA knows how to administer. Oracle. Or sequel server you probably know how to administer, a Postgres, database it's. Not something completely, net new yes. There's it's not the exact same but. It's much much closer than significant. Changes there. This. Is important, as you bring on new tools and systems there's a overhead, to ramping, up to learning these really. Important that you can bring in a new system and just, operate it this. Is a big kind of win, here. Every. Framework you're probably already working, with already. Works with Postgres you, don't have to go and write, well sequel if, you're using Ruby on Rails if, you're using dotnet, if you're using, Python. All these, tools and frameworks already, work well with Postgres it's, a thing that's heavily tested against heavily used so, again. It's just you know your usual tool section toolbox. Oh. And. This is a big one it's. Much easier to hire for Postgres developers. Want to work with Postgres like. It's a stodgy, old database that's been around a long time but, somehow it became a cool database I don't know if you can say that about databases can they be cool. But. People. Want to work with it and I. Don't. Know about you all like I like. Hiring is not easy like we're. In a great industry it's wonderful, to be in demand but, so like go and hire and recruit someone is a month long process, I'd much, rather give, them the tools they want to work with it's. Much much easier to recruit for and. Postgres is one of those ones that checks the box if you looked at how loved it is by developers, they, want to work with it they'd rather work with it than other tools often. And. The really big one great for. Businesses out here is reduced, total, cost of ownership. To, to kick out an Oracle, database and bring in Postgres to, get the same experience. It's, much cheaper, works. The same way you don't have to learn something that new but, you can actually reduce your total cost of ownership and so, I actually want to invite a couple of people on stage so. I want to invite roll and Shum, on stage to, talk a little bit about what. Chevron, did with Open Text in migrating, from their. Oracle set up over to Postgres, thank. You guys. So. Move away a little bit to the middle. Yeah. Thanks for being, here or. Fighting. Me yeah so my. Name is Google Alyssa I'm a program, manager for the Microsoft solution from Open Text so. If you don't, know Open Text we, are the largest. We. Are the largest. Independent. ECM provider in the world and if you look in at, the gardener, come to service quadrant we're right, next to Microsoft, up there and, we.
Focus On. Connecting. Unstructured, content to, business, applications, right so think about Dynamics. CRM, as. AP sales force that's, what we focus, on so. You, know because our systems are always right. Next to the business systems, you, know they're really important, for an organization right, so you. Know you have test environments, you have acceptance. A fireman's and you fill over backup so, that means that you, know just like your business applications, it needs, to be fully skilled. In. In our case that means that sometimes you have four or five different. Running. You. Know besides your production environment and if. We then you know talk about a failure right if all, of those applications, use expensive, database, licenses, then, you know your total, cost of ownership go sky-high right, and you. Know as a vendor you know the only thing we focus, on with organizations. Is the return, on investment, of your investment. Right on the Open Text investment, so, that means that you know if you have a really high TCO, then the return on investment will, be you know longer, right, and for. Us as a vendor that's bad but, also for the customer right and you, know the less money you can spend on new investments. Or new technologies, means you, know less profit for the organization. So. As I open decks right so so I personally I take, care of the go-to-market for Microsoft, solutions with open for, Open Text but I also manage the Microsoft, partnership, and as part of that you. Know we also need to think about how we embrace, these, technologies, and you, know although we are clouds. Agnostic. Right for, protects it doesn't matter where you run your software, are. You know 25, plus history, partnership. With Microsoft, showed. That you know adoption, of those technologies, is is you know really key and easy, to do from, our experience, so we. Supported. Postgres. A while, ago for a lot of our solutions including, adoption, but also the, mainstream. Application. From open text to, you know lower the cost. Of ownership but also being able you know to, run copies. In VMware else you know the. Applications. Are. Numerous. So. As. Such we also now, are, supporting. As a Postgres. In production, environments which we couldn't, do in the past because of you know performance, maturity, things like that but, right now you know we're happy to be able to support, that also in the production environment, so. You, know that is a value for us as open text and like I said we. Most. Of our solutions, run on Asscher we. Also try to adopt all of the technologies, like blob storage as. A sequel, and all those there's technologies, and this, is you know one key, one. Key solution. Of that so. Shamans, gonna tell, a little bit about the experience, of using that Thank You Craig thank you also I'm Sean Patton Matthew I'm the, manager. Within, Chevron. Who supports, the content, management product. Suite, not, all of content management but, content management mainly focused, to, support our facilities, engineering, so, all the mission-critical data. Document. Um is a mission critical application. For us and. What. I wanted to do is I asked break how much time I have he told I have like 10 minutes ourselves so this is not going to be a Bollywood, story there's going to be a Hollywood story I'll get straight to the point.
What, A. Year. Before. Where. Bieber were we. Had, open. Text document I'm up and running for almost, 15. Years or so every. Time we do an upgrade every, time we read platform, every time we need to put an Oracle patch it, takes, humongous. Effort, from our team there. We were pretty. Much running an IT shop. Within. Chevron, so. What. We did was a year before we partnered, with Open Text and Microsoft. We. Had a week-long. Hackathon. Where, we thought what we can do best. And again to, Craig's point why developers. Love Postgres. Right so, we asked to open text what tools, do you guys use to support your thing what do your developers, use and they told we, use Tomcat. We use Linux. We use phosphorus, and. If. The developers, are using that why can't we take that to production and just use that right so if it's tested by developers, it should be pretty easy for us students so a year ago you did a hackathon and you dropped it in anywhere in production, in a week right yep. So. What we did was we took our. Platform. Which was which, consisted, of like almost hundred and sixty servers millions. With. Several, instances, of Oracle, database and everything and, we. Started, moving into Azure kubernetes. And using. Azure kubernetes, weari platform, the entire document. Em suite working, with open text and Microsoft and we. Implemented. Helm charts on it and then, also. We. Partnered with Microsoft to, get everything in Postgres, as. A past service, and, right. Now we are live. As of like last, two months and we are in the progress of migrating. The on-prem applications, from Oracle, to, Postgres. We were talking a little earlier that like, you, didn't just say okay let me just drop in this like you you kind of modernized, everything like, you, before it, was very much that analogy. Of like you. Had your server and you named it and you knew what it was and it was this special VM that you didn't touch and as, you modernized, you kind of went all-in with kubernetes, and containers. Postgres. Is a big part of that. And. Thanks for bringing that up right so I. Love. To, I'm. Very proud to say this before. If, we were even - or. Even if you are trying to bring up a development environment from, ground up it, will take my team at least two or three people. Probably. Two weeks to get the whole stack up and running but, the way we modernized, over the last year, it's, one are from. Two weeks it's it's it's. Not the number I'm cooking, it up it's really one hour end-to-end. With Azure DevOps, clicking. The helm install, and the, whole stack of dark, when I'm up and running with Postgres. Behind, the scenes and so I'm curious a little bit right so you said you've got some of them in production I think we were talking that I apologize, I think this slides out-of-date like it was 13. Deployments. When I created, this slides and I think you said you were up to 15 now because. In production like how, is that journey right because like moving a database is not, like. I talk, to people and every time it's not a relaxing, thing I've helped people move a bunch but, it's a little nerve-racking everything, single time cuz it's, it's a sensitive thing right, you, lose day though or you know there's performance issues or how do you know what's going to work what was that journey like um.
It Was a scary. Journey. It, was very scary but I. Think we did, several. Hackathons. With Microsoft, and open text and. Once. Open tech started looking at how this migration process, works and and. They, were comfortable, with it and Microsoft, was comfortable, but it we validated, all the data and everything then, it dis like a clockwork, it, just started working how is the data, validation process, can you talk a little bit like how did you move things over how to do testings. Pretty. Much with the we. Are doing evaluation. With the DMS solution, where actually, it takes from. Oracle. Directly into, post Chris and and. It looks at all the tables and all the data and what field and what not so we have the complete history of what's happening at it okay and then how are things looking now like you mentioned about that time to deploy what's, the overall stack look like now I'm kind of curious if you would from it on-premise. Kind of database. Where you were you know watching, carrying after probably, getting paged for all sorts of things, patching. Frequently, like what is the experience like now on the ones that are in production so there are still some on primum two implementations. In. Chevron, which. Uses Oracle, and last, night I was in a firefighting called troubleshooting, and adding indexes, and so on and so it wasn't your Postgres database is you it wasn't my post press okay but with, post cursor I mean with being in an agile platform and everything I have everything in my cell phone I can watch what my. CPU, is with any nodes and everything, without your log analytics, all my logs are being pumped who, is accessing, my database, to the nth degree I will be able to tell, from my phone and. That. Sounds much more. Peace of mind so I like I think the idea to not have to manage your database is it sounds kind of refreshing yeah I don't. The. Database gods that's, what they unfortunately. I have to do that so that's a horrible, business don't don't do it. Cool. Anything. Else they'd like from that journey that you think you know it's worthwhile to share. I think. The key was our. Management support, where, chevron, has, always. Been, supportive of innovation, so go, figure it out and if, there is value as long as we are not, increasing, the cost which is the major driver we, are actually reducing a lot of cost here by moving to Postgres and. With. The management buy-in, where. They are saying go in a wait go figure it out we just started very, simple, how, we can take our Stone, Age not. Documentum but Stone Age infrastructure. Into. The, great modernized, infrastructure, where I can control everything that. Was the key value and one more point I wanted to stress is Chevron is an oil and gas company right so. We. Are not here to do IT we, will leave the IT to the hands. Of the experts, where, we want to focus on is set the foundation, right where, we don't need to worry about patching, we don't need to worry about upgrades, we don't need to worry about do.
I Need this patch or that patch or what not but where we want to worry about is once I have my data I know it's reliable, I know it's scalable how, to can I take that data and and. Make business add business value how, can I do, data analytics, on it how can I. Show. More metrics, around it right so that. Was the key driver it sounds sounds like you know an exciting way and a really good synergy on the you know what developers wanted as well as what management wanted right so, thank, you great to hear and then I think if you're curious hear more you're giving a talk a little bit later in, the theater around 4:20, yes yes, so what, we did was we again partnered, with the Open Text and Microsoft. And what, we are doing is we are doing an evaluation or, actually, finished an evaluation, with the DMS where we are take, Oracle database and have, migrated. Successfully. Into, the Postgres, and I. Would love to share that experience in that. So. Yeah if you're curious I would encourage you to stop by for a little bit more thank, you guys. All. Right, so a few, more things like I think to summarize. It a little bit, so. When the Postgres side right a big one for businesses is lowered TCO, I think Shawn mentioned it several times right it's reducing costs like a business cares about that they love that, because. It's open source no, one can ever own it. You. Won't get locked in there is no expensive licensing it is, sequel Postgres. Is one of the closest, databases, to Oracle, compatibility, out there so, if you're you've got some Oracle and you want to get rid of it Postgres, is probably the closest thing to drop in there and. It's not a new shiny data store right like you, hear that he's managing it from his phone now it's, not a new thing to go figure out how to manage it's easy drop-in, very, similar to what you're already doing. So. You, know all of these build on each other so you've got a really really rich feature-rich. Database. You. Already know how to work with it it's a it's a sequel relational database and then, you have the ability to reduce cost but. I. Was, gonna do one more thing but I think it's about five more minutes so my. Impression, of Steve Jobs my best attempt there. Hyperscale. Siteís so. One of the big, knocks against relational databases is that. They don't scale you, scale up you scale up you scale up until you can't scale any further and then you get this expensive, appliance. That you put on premise or, you. Migrate, to a distributed. No sequel store. Hyper. Scale sightedness we announced public preview back at build we, ga8, it I think you heard in, some of the keynotes earlier this week, it's.
Pure Postgres it's not a fork we haven't gone and taken, post crest and changed it and manipulated, we've taken and used all those low-level extension, hooks and transformed. It into a sharded, horizontally, distributed, database under the covers so. As a developer it still looks like a single node database it, doesn't look like anything, complicated. So looks like a single no database and you, can now just scale, it with dragging a slider so you drag a slider it adds nodes data rebalances, it automatically, routes queries under the covers. So. To give you a little bit of a look at like how it would kind of work if. You're doing something like a you, know update on some sequel. You're, gonna talk to a single coordinator, node it's, a standard Postgres, database, that you connect to. The. Coordinator is gonna know that this data lives down on node one two three it's, gonna rewrite on the fly route, that query down update. That single record send, that back to the coordinator, and. Just say okay cool this is committed so everything's fully transactional. You. Don't have to know that all of those other nodes are there and. You're looking at single millisecond, latency so it's still like a standard database we're not fanning things L we're not kind of waiting, for some consistency, to get there it's, committed it's it's, there same. Thing for inserts, for, selects. If. You run a basic select and you want to just get hey I want purchases, for this given user that only lives on one node it's. Gonna push that down really, quickly give you that really sub, sub second kind of response, now. You. Also have the ability to do more advanced analytics, on it so, what happens if you want to say hey I want to do a count, star well. A count star is now split up across all of these three nodes my data is spread out how does that work so. As soon as you actually query that the, coordinator, is gonna recognize in this case we're actually gonna do it average so something more complicated than the count star we're, gonna get a Selman account from every single node, bring. That back to the coordinator, and and. Then. Give you the result so the coordinators basically get a sum up everything of the, sums of the counts knew that final math fair now. The query we executed, which is a standard, sequel query it wasn't anything complicated, it's a basic average query. Under. The covers it's gonna actually turn this into distributed.
Sequel. That it can execute now. It's gonna run this across every, core in the system so, you can actually see you know performance, benefits of 100x over a single node database because we're paralyzing, this workload so, similar. To some. MPP, systems but, we're doing this in a transactional. Way so. We're really interesting in that you can start on a really small single, server Postgres as you need to we, since failed to hundreds, of terabytes. Double, digits nodes extremely, kind of large-scale databases, we have customers in production was. Up to a petabyte. Of data all, as, a transactional, database. So. Now, on Azure right you've. Got a feature-rich database, in Postgres, if, you have any feedback on a you, know as your own Postgres experience, I would encourage you to come up happy, to chat afterwards swing, by our booth it's. A database you can trust because it's reliable, not gonna lose your data it has a long history of being reliable developed. In a good slow steady fashion. It. Can reduce costs if you've got a you know expensive, proprietary database already in-house and. Now you don't have to worry about scaling, and. Then you can actually also run this anywhere so you can run this on a sure but. Hyper scale is one of the options for as your arc so, if you want to take and run it on your own cluster so you have you know parity, with your on-premise, stuff that, you're still running for whatever reason I need to scale along with your, your stuff on Azure you can manage it all in the exact same way so you can still look up on your phone and manage this for your on-premise stuff now with ARC. So. Quick. Thanks and then a couple other kind of call-outs and notes, so. One, if you like Postgres if, you like reading about it learning about it I encourage you to sign up for our newsletter. It's, once a month it's generally, kind of highly curated content and interesting, breeding it's. Interesting, blog posts, things that we see about post press that are exciting, that, sort of thing so if you're you're curious, and want to stay in touch with post press encourage you to to sign up and subscribe. Please. Send, us feedback let. Us know what you found interesting in this session let us know what you would like to see more of. Hopefully. We see you here next year and we, have, more sessions about Postgres and can dive deep. Into all of the internals, and that sort of thing so let, us know what, you want to hear and.
I Think we probably have about. Five. Minutes or so so for questions so thank. You and if there's any questions happy to take them. Yep. It. Works. What. Are the chances of getting a service offering, similar to the azure sequel offering, that's out. Serverless. Is definitely something we're looking at because. Of the nature of post press like it probably takes a lot of changes to the engine that we're exploring, the. Unfortunate answer is it's not an immediate quick, thing on our roadmap so definitely. Something that we're looking at and like the feedback is helpful that, you know you would like it. Any. Other questions. All. Right thank you very much.