The Database Trends that are Transforming Your Database Infrastructure Forever

Peter Zaitsev19 minutes

Today's companies face dozens of technologies, hundreds or even thousands of individual database instances, and petabytes of data. Explore how this complexity is changing the way we look at systems and operations.

Share this

Video Slides

Video Transcript

My name is Peter Zaitsev and I am the founder at Percona. Today we are going to talk about trends with databases with particular focus on open source databases. In the first part of presentation we’ll talk about the innovation which has been happening in the last few years.

the first very important topic is what we are moving from a single node databases like think MySQL poses to many of a distributed databases which are designed from a ground up to be distributed right that’s you can think about ScyllaDB Cassandra mongodb new SQL databases like hugobyte or cockroach DB are all designed to be distributed at heart the other interesting Trend I think is separation of storage and compute which is coming to us from the analytical databases you know I think uh the snowflake as a great example but now we I see even some uh operational databases like neon is uh building on this uh new idea of separation storage and compute we also have arrays of a serverless databases well of course the serverless databases in particular means there are still servers which run the database Technologies but as a user you don’t think about provisioning database service or instances you are just thinking in terms of what kind of capacity and performance I am buying which is much more friendly we also see over the last few years a lot of different database models have uh came to existence I remember when I was getting in the database ecosystem in the late 90s SQL or relational database was pretty much it vast majority dominating and SQL and relational database are still very popular but we also have a lot of general purpose databases time series database graph databases document data structures and so on and so forth right the database particularly built for specific purpose even more so we have at random some databases become a multi-modal databases so single database can support multiple data data language data models and often support uh different language or uh protocols we also see as a part of this is some form of emulation where newcomers to the market we use some popular languages and uh protocols to simplify adoption for example if you look at a click house it can speak both postgresql and MySQL protocols Victoria Matrix has in flux DB and graphite API ferrydb uses mongodb uh for front-end try to uh to talk uh to replace mongodb applications with a fourth grade scale back back end another pause the gray solution is Babel Fish allows you to talk to postgresql as if it were Microsoft SQL comparable database right so there’s a lot of uh those Solutions in their industry now if you think about the open source what is the biggest Factor impacting open source right now and has been for the last uh few years perhaps five or more years and it is the cloud because Cloud changes a lot in terms of how uh open source is adopted monetized and so on and so forth and in database space in particular we can see this interesting outcome for number of years we had a databases over years by DB engines uh the technology very open source databases right and DB engines rating manages databases by uh in this case by how much growth vargation and this here here or actually well 2021 it was there a snowflake with very much property cloud-based database so really cloud impacts our ecosystem a lot now how does cloud impact an open source and database in particular one it really allows to maximize and simplify adoption because in the cloud it’s actually all the databases are easy right wherever complicated setup you need to do whatever that can be simplified by the cloud vendors so you can deploy the most complicated database in a couple of clicks at the same time it changes opportunities for monetization right and now that is very a lot of conflicts in or with a cloud and open source I think develops in this case I want to uh remember rewards of Martin Mikus who was a CEO of MySQL AV when I was there who was always saying to us what hey open source is not a business model you can think about open source as development model right even maybe marketing model but you know what open source itself is not a business model and can support many different business models right and obviously Cloud changes a lot in this regard for example for MySQL in particular we have a big impact of a cloud was what cloud allows you to hijack GPL right if if you’re not trying in the cloud you could say hey if you are modifying my software and making commercial version of that well you cannot use advantage of GPL license you have to buy the Enterprise license and MySQL used to make a lot of money this way but guess what it doesn’t work in the cloud and Amazon can modify MySQL with Amazon Aurora right then monetize it without paying to Oracle anything for that this is why especially in the cloud it’s important to understand how the open source project you are talking about is own and governed right and in this case I want to separate open source in two different buckets one is Community Driven Foundation multiple vendors uh you know more uh like some ecosystem uh based right and then there is single vendor driven and that is where a reality of a cloud is quite different if you think about Foundation based open source they’re very much focused on cloud helps to accelerate adoption and yes Cloud changes who captures uh the value but you know what as long as the ecosystem is stronger Foundation is stronger right and we can see this very well with false gray squirrel as an example folks really love what the cloud really have for postgresql to become that arguably number one open source database has become in the last uh last few years and yes top dogs in a false gray squirrel ecosystem is no more Enterprise DB but the folks as um Amazon the Microsoft Google sort of our usual uh Heroes aspects but ecosystem itself is a very strong and flourishing now you think about the single vendor open source which tends to be Venture finder the public companies they are really focused on maximizing for uh and making that particular vendor successful so they really fearful of a competition of the cloud vendors because guess what if Cloud vendor makes money instead of I’m making money that is not good for them and their investors and that has caused a lot of those folks fully or partially abandoned in open source licenses right and this is just like an incomplete list of those vendors now what is interesting in this case is what exactly is a primary goal of that of that license change well is creating Monopoly on database resource markets right those folks they say well you know what if you just want to run a software yourself maybe serve some services to your customers to based on how open source software doesn’t matter right well not open source but our source available software uh it doesn’t matter but if you really want to build database as a service that is going to be a problem you’re not going to allow us according to the new license terms so what is the deal with database as a service why there is so much focus on on that well it really provides you state of art simplicity right that allows you application developers to focus on the application and not about that may seem complicated database kitchen which is actually requires very different skills what application developments often have right it allows a high level of automation as well so what does that mean the problem is if that Monopoly on database as a service Market well I think that really makes software not any different from property software because using database as a service is very different skill compared to Rolling your own database setup and maintaining that right and that means many of those companies just have no practical alternative because we can adjust uh implemented in-house if they if they choose to but in this regard I think not always lost and I think actually we have seen something similar to that before right I think the open source open source Solstice tends to be slow but inevitable right it doesn’t have such a large budgets and moves uh so quickly

in a Innovation as some focused property companies can right so for example in early 2000s I remember we had this very polished Microsoft and I remember folks would laugh at US Open Source dude who would use VI to code something in parallel and PHP but well if you think about now we have a fantastic tools for developing ecosystem which are completely open source now though we have in the cloud vendors like Amazon provides us a very good and Polished uh Cloud experience which is kind of uh a lot more polished than the one of what the open source ecosystem provides another example of course would be something like cellaris versus Linux I remember when I started in in this echo system uh there have been a lot of people using uh salaries and laughing at Linux when I started with Linux it was 32-bit operating system and didn’t handle multiple CPUs completely you couldn’t even create file more than two gigabytes inside it was a joke well and where is that joke now compared to uh to Solaris so I believe that is what is happening with open source right now right and is going to continue uh continue happen What open source catches up and will continue to catch up and I think there are a lot of innovation in this case come from the cloud native confusion foundation and and kubernetes right so what does that allow us all to do well I think if you look at the cloud you can really look at the cloud at the two different ways one is cloud as commodity you know like think about like internet there is a lot of providers of your internet out there but you know what there is a pretty much no difference which one you use right the same happens in the cloud right now if it’s infrastructure services like Network like storage like computer pretty much all the same different price different quality but they are the solid building blocks and then there are those highly differentiated Services right which are property and which create the uh uh create login and I believe if open source we can let the cloud to be that commodity infrastructure provider where we can create most of the value through the open source software what is interesting in this regard Amazon tends to agree with me that image is actually from Amazon presentation from you know decade or more ago when they compare cloud computing to electricity services and guess what electricity services is commodity you can buy electricity from any provider and you know what your free jio TV will continue to work you know you don’t have to change any of your house appliances if you happen to change the electricity provider well in fact a little bit of philosophical discussion let’s talk if you do have open source database as a service implementation in the ecosystem well and as I know yeah not really not yet but we have uh taken the baby steps and we do have a very solid foundation what is that Foundation the first one is kubernetes which is a foundation that cloud native ecosystem I mentioned anything about the kubernetes is not so much what this is amazing uh piece of uh of software right well but but the main thing is what that is universally available and has such a huge community and uh and a mind share behind it right kubernetes is available at any uh oh public Cloud vendors uh worldwide it’s available for uh private Cloud providers you know I think like uh VMware openstack right you can use uh run kubernetes on all of those uh environment right and kubernetes while it started to focus on stateless application it’s getting better and better for uh State full applications as well and most open source databases now have a kubernetes operators available so you can run those operators out there so why operate kubernetes operators are important and how they are different let’s say from a bunch of uncivil or terraform scripts you may have already to deploy your databases well within of databases is day one is really only two of the iceberg most problem most complexity comes from a day two operations because guess what when you put the database in production it likely needs to operate for decades it needs to be started up most of the time in many cases right or at least the overtime having like a no downtime right or if a downtime has to be taken that has a significant business costs right and you cannot lose any data right that is why those day two is very important and the kubernetes Really allows you to have a very robust day two automation much better than many other uh Frameworks right and uh database uh as a service can be built on kubernetes kubernetes can be if it’s found they and kubernetes operators can be a fantastic building block some people may say well you know what we heard what Iranian or databases on kubernetes is impossible bad and so on and so forth but reality is if you look at majority of database as a service solutions which has been uh has been built over last few years we actually use kubernetes underneath right here are some of examples of a folks who mentioned to me or stated publicly what they are using uh using kubernetes operators so if you think about that whether you believe in databases kubernetes or not there are hundreds of thousands of database instances that constantly running on kubernetes right and quite successfully right because we don’t hear you know too much of oh my gosh it’s all Crash and Burn Right For by folks using those Solutions if you look at their Corner we are actually one of the companies which is built in an open source database as a service uh solution as a feature of our per Corner monitoring and management project right and uh we expect to be one uh one of any right as it often happens if open source first you see the proprietary Technologies come to Market they figure out what has to be built how exactly it has to look how it should be marketed and then the open source comes in and democratizes if you will right makes it more available uh low-cost solutions to a market and I think that will be happening in this case as well our solution is still a work in progress in better if you want to try it out give us some feedback or even contribute some code you can and you’re welcome to do so that is uh the power of Open Source software comes from and that’s all I had to say uh for today uh let’s keep a chat in touch and enjoy the rest of the conference [Applause]

Read More