Ask the expert session, this is a follow-up from our digital breakout session where we talked about new innovations, for azure database for mysql, and postgresql. To turbocharge, your app development. Today we're going to answer all of your questions that you may have about the flexible server preview announcement, we just made, but before we do that let's quickly go over some housekeeping, items. First off to interact with the team of subject matter experts here please use the chat to ask any of your questions. And feel free to upload any of the questions that you're interested, in hearing, being answered, on air, with us in the experts here. Please also note that this session may be recorded, so keep that in mind as you ask your questions. And please be polite to all of the other participants, in the chat as well as our moderators. I also want to. Point out that we are following the microsoft, code of conduct. And so we're all here today to answer all the questions that you may have about the flexible, server preview. And if you watch your digital breakout session you would have learned about all the great features, that we've released, along with this deployment, option. So before we jump into any questions that you may have i want to do a quick round of introductions. My name is andrea lam and i'm a senior pm, on the azure database for my sequel and maria db team, fun fact about me. Through this pandemic, i've really, gotten into collecting, plants so you can see a couple of them right behind me, um i've got about 30 plants now and i think it might be an addiction, but. Yeah we'll see uh what about you arun. Hi everyone. Nice to meet you all my name is arun, i'm a principal, program manager. With the azure open source databases, team. I, focus, on migrations, to open source databases. And i have over seven years of experience, in the migration, space. I'm super glad to be here today, and, on a lighter note, i'm.
Very Happy that i got a chance to. Go get a haircut. Yesterday, so i could be camera ready today. Nice to meet you all and over to you lucas for the introduction. Perfect uh good to meet you everybody um so i'm lucas fiddle i'm a pm on the yesterday's, proposed chris team, i turn microsoft through the site is data acquisition, so i've been working with you know postgres for many years now in the open source community. Um and i really you know care about making sure that microsoft, you know um works well with the community, and we've done you know some good work with the postal community here, um. On a fun more fun note um i did actually manage to get a bike right in today between ignite sessions, so between, this session and the previous session i was able to do a quick bike ride which to me is you know important. Like gotta keep working out and stay healthy so. That's for me uh manish. Hey, thanks lucas, uh, hi everyone i'm manish. I am a pm in the azure. Open source, mysql, and the mariadb. Managed service team. Um. I joined uh microsoft, couple of years ago, after. Working in, the mysql database back in oracle. Uh fun fact about me like everyone else i refined my cooking skills during this scovic time, and i'm, certain, it has. It has increased a lot, over the last six seven months that i have been stuck at home. I'll be the moderator, for. For for you guys today, um. So let's kick start uh, this conversation. Um. So we announced a bunch of things with, the mysql, database, and the postgresql. Database. Uh, all along the flexible, servers, so lucas could you help us understand, what exactly, flexible, server, is, and, why it's, interesting, uh for the microsoft microsoft, to invest in this area. Of course yeah, and for that maybe android can we go back a few slides, i think you had, the deployment, options, uh there earlier, perfect um so i think the important thing to understand right is what do we mean by deployment option, so deployment, option, um, you essentially already used it if you use the product which is if you use single server right essentially, a certain kind of architecture. In a certain way of deploying, the product right a certain, um essentially when you create a server like a database server, then you choose, a single server for example is today an option right you also have the choice of hyperscale, for postgres.
And So essentially with the new flexible server option, we're introducing a new architecture, and a new way to deploy postgres, in my sql into azure. Um really flexible server is about you know giving you more control, and giving you a simplified, developer, experience. That's that's great to know lucas. Um, one question, that i'm getting from the attendees, is, um, what's the difference between, the, already, existing. Single server. And the new flexible. Server. Uh andrea could you help uh clarify, that question. Yeah happy to let me just switch over to one of our slides really quick so. The cool thing about flexible server. It is it supports, a bunch of new features that aren't really available, in single server and this includes a bunch of really great things around. Security, including, veena integration. As well as more. Server parameters that we're exposing, so for any of you postgres, or mysql. Experts. Who really want to control, particular, server parameters. Flexible server allows you to do so, and i also want to mention custom maintenance, windows and essentially this allows you to pick the deployment, time in which you're going to, receive some of the maintenance, for your particular, postgres or mysql, um, engine and then on top of that it also supports, zone redundant, h.a which is one of the really big announcements, that we've made as part of ignite, and this gives you the ability, to do synchronous replication, between availability, zones so really exciting for those of you that need zone, uh zone resiliency, for your applications. And then, a big chunk of our investments, have gone into cost optimization. And this includes. Investing, in the burstable, compute, tier and that's our lowest pricing tier available for flexible, server. Um allows you to bank up cpu credits to spike up to 100, very similar to what you see in azure virtual machines. And then, we've also introduced stop start capabilities. Allowing you to stop your server for up to seven days and allowing you to save costs in that way as well. Uh that's, great to know and i'm certain that this is one of the most appealing product for the developers.
In That regard there's a question that i saw, from the attendees. How do we. Migrate, our existing, workloads. To, the, the new flexible, server. Arun could you uh, help answer this question. Absolutely. I think that's a great question, andrea could you pop over to the migration slide please. Thank you very much. So as you can see there are multiple, options, for you to. Migrate, to, open source databases, on azure. Given these. Databases. Are open source. You have the option, of using. Native tools, and multiple. Community, based tools to, enhance, your migration, experience. And keep the entire migration, process as predictable, as possible. So as you can see there are three main stages, uh the pre-migration. Phase where you go and uh, evaluate. Your workload. For, workload, fit, uh from a performance, and usability, standpoint, in azure. Platform as a service. You discover, everything, you migrate the schema, and you start preparing for your migration. And, when you're ready to do the actual data migration. You could use, a migration, tool, varied, across, three different. Sub buckets. So depending, on your use case you could either go in for, an offline, migration, tool. That ships with the database, engine by itself, like dump and restore, basically, like my sequel dump and my sequel. And and restore. Or, even pg dump and pg restore, for postgres. Uh you could also use uh community, based, uh, tools that are, designed, specifically. To perform. Super fast, parallel, thread based migrations. Including my dumper and my loader. And, if you require. Online, migrations, to reduce, your, available, downtime. You have, a combination. Of various, native. And. Fully managed services. Namely from a native, logical, replication, standpoint. You could use, the bin log replication. Or data and replication. Into azure db for mysql. Or, the newly. Supported. Logical, replication, capabilities. Where you're able to use the publisher, subscriber, capabilities, that got introduced. In post s10. To be able to perform, near zero downtime, migrations. And. In order for you to get that programmability. Edge, if you want repeated, migrations, and repeat them over multiple, workloads. You could also choose to use, our, online, migration, tool called the azure database, migration, service which is fully managed. It uses, some of the concepts, that you see from the logical replication, tool.
I Tools i just talked about. And. Enables, you to perform, near zero downtime, online, migrations. And once the data is moved you go into the post migration, phase you test things out, ensure, your apps are ready to go in the azure, platform. And boom your workload, is running in azure. Giving you all of the advantages. That come with keeping a leg in the cloud. Back to you manish. Thanks a lot aaron i think, that's that's a great set of tools that the customers, can use. To bring their. Existing, workloads, to the flexible, server. One question that i, see from, one of the mysql. Developers. Is. What are some of the flexible, pricing, options, that are uh available in flexible servers. Andrea could you uh. Help answer that question. Yeah so like i mentioned, earlier. The, burstable, compute, here is going to be the lowest. Cost here that we have but the way that pricing essentially, works in the pricing model. Let me just pull up our documentation. Really quickly for you. And essentially, what we do is we break up pricing, based on compute, and storage, so i'm going to switch over to one of our docs here really quickly. Compute and storage here. And essentially, what you do is you select, the. We basically the virtual machine, size or compute size in our service, and this is. Dependent, on the. Compute tier available so burstable, general purpose or memory optimize, and each of these is run on either the b series, d series or e series machines. That allows you to also scale up the v cores and with that the memory associated, with the machine. And then you can also scale up storage, and so the pricing is based on that so, if you guys do want to check out more details, in terms of pricing, definitely check out our pricing, page. And i'll pop it up here on the screen as well, and you can also use the pricing, calculator. In. Pricing calculator available in the pricing details, to get a better understanding, oops i have a different link open but it's flexible, server pricing. To get a better idea, of what pricing looks like per region. And so this is flexible, server here. Um. Uh would you like to touch base on the start stop capability, that customers. And developers, can use, yeah great call meni so. Stop start allows you to basically, stop your server for up to seven days and essentially, what we're doing in the back end is stopping the compute, for your server. And so during this time, you know you'd still have to pay for storage because we still have to keep your data around so that eventually, when you restart your server it actually has data files to connect to, um but stop starts a great idea for anybody. Who you know wants to take. Cost optimization. Options. Uh maybe you want to stop your server over the weekend maybe you don't need it. Or dev test scenarios, that's that type of thing so definitely, a lot of different scenarios, that can take advantage of stop start capabilities. Thanks for that uh andrea. So lucas, we have a hyper scale. Cytist, lover. The question, is uh, is cytos, hyperscale. Still limited to a single database. Uh. They are interested in dynamically, scaling, postgresql. Performance. But have a completely. Segmented. Separate database, architecture, for separate. Reasons. Yeah i think that's a great question monique um, so.
Let Me take a step back and describe, a bit more how hyperscale, works right so first of all the thing to understand, is that the cytos extension is open source by the way so you can you know go go look at that in github. And, the. The limitation, around you know using a single database called cytos, as a main database, um. Really comes from the way that extensions work in postgres, right so extensions, are per database. Level in postgres, which means that you have to create them. In each database. And so from a kind of management. Layer perspective, that you know introduces, a lot of complexities. If you support multiple databases. Um so a recommended, multi-tenant, model. Is essentially, to, use what we call kind of share table multi-tenancy. You can also use role of security, if you need additional, kind of you know security barriers. Um so to answer the question specifically, um today we don't support multiple databases, and it's it's technically a bit difficult, because of the inherent kind of architecture of post extensions. It is something that we know is important, to you know, some of our customers, um, and so we are certainly considering, to add in the future but i think at this point we don't, have an eta for that. Okay, thanks, thanks for that lucas. Uh. Another follow-up, question, um. From uh. From one of the attendees, is. Um. Is the presence of, zone return in ha, which is a key differentiator. That we called out in one of the previous slides. Could you touch base on that, and how that is beneficial, to. Modern, enterprise, applications. For sure yeah and let's maybe switch back to the slides for that and take a look at the, aj architecture. So or actually we can also start well. Eight of those two works, um. So the um. The ha architecture, here right like um that we show this is a flexible server right so just important context switch here we're not talking about flexible server again, um and so on the left side here you can see essentially the availability, zone one, on the right side you can see availability zone two so an azure an availability, zone is essentially a set of data centers that are you know independent, from each other so that means if there's a power outage or a fiber cut then you know the essentially, like if the one zone goes down the other side would still be available.
And So in order to kind of you know, like understand the benefit here, it's really about you know making sure that, if you have mission critical workloads so you know need to be up and running all the time, the, zone resilient option we're introducing, the flexible server, is key to that right and so the way this works is that you can um get essentially your primary server and zone one here, and then use synchronous streaming replication, and postgres, or mysql. Um, to, essentially, stream all the changes over to that second standby. Um the reason you know benefit from synchronous, of course is that it, doesn't have any data loss right so essentially, it always makes sure that when a transaction, comes in it also gets committed on the second zone, um and so that means if the zone one goes down, um you, automatically. Switch over to the second zone and you have no data loss the application, doesn't need to do anything it automatically. Essentially, happens when you reconnect. Okay. Thanks thanks a lot uh lucas for that explanation. And just to add to that, uh my sequel, uh similar to, postgres, my sequel uses, the vin log replication. Uh which is uh the native replication, provided by mysql. And it is again synchronously, appreciated, across the two zones. To bring this, data, resiliency. Uh in case of a zone failure. Uh, moving forward. Arun one question that i'm getting uh, is, where can um. Where can, folks go to learn more about. Uh the migration. Processes, and what are the best practices, for that for migration. Absolutely. Happy to answer that andrea, could you open up the data, migration, guide for me please. Thank you very much. So uh from my experience. And based, on. Multiple, customers, we've helped, onboard, the azure. Azure, platform. For both mysql, and postgres. Migrations. Are multi-step. Processes, and it almost, always. Helps you to have a framework, to go after, use best practices. For each. Step, and sub step in that framework, to be able to. Uh, make sure you. Have successful, migrations. That. Don't, add on a, lot of risks as you move your data from one place to another. So at azure, we have the cloud adoption, framework, which is roughly, split, into, three. Big steps, that make up the entire migration, umbrella. We talked about it previously, which it is the, uh one as the pre-migration. Phase, the migration, phase and the post-migration. Phase. And to be able to talk to all of these best practices. Per cloud adoption, frameworks, sub-step. We've come up with this database, migration, guide that you're looking at on the screen. For all of the sources, and targets, we support from a migration, standpoint, today so, you will see everything, including. Sql server, mongodb, to cosmos, db and everything in here, but specific, to migrations, to open source databases. Uh andrea if you could just uh pop into, one one of the migration, scenarios, there and click click into that, so you will see that this is this is the one one example for a postgres, to azure db for process migration. You will clearly. Look at all of the migration, best practices. From an overview, standpoint. What tool to use when what the supported, versions, are. What are the recommendations. And prerequisites. You need to take care of before you start a migration. With full focus on successful, migrations. And, it will help you go from one stage to another. Helps you with those process, best practices. Ensures, you don't, miss out on any important, step. That comes as part of this, important, migration. Uh the best part it is. You could do this yourself, if you are very familiar, with mysql, or postdress. Or. On the right hand side of this page, you will see a list of our partners, who are well adept, at, uh, helping our customers, migrate, hundreds and thousands of workloads, into azure database for posters, and azure database, base for mysql, so you could reach out to them also. And lastly. I definitely, encourage, reaching out to us through one of our deals that that andrea is going to put up on one of our slides today. We'd be happy to uh help you with any sort of questions from a migration, standpoint. Or even put you in touch with one of the partners you're looking at on the screen here, back to you manish.
Thanks Thanks a lot arun, i think that certainly, answers the question. Um. So andrea, one question that i have here from the attendees. Is, what are some of the new parameters. That uh, flexible, server, has made available, for customers, to tune. Um, could you share some insights on that. Yeah so we have a bunch of nodb. Buffer pool, um. Buffer pool. Parameters, that are available. We've also have a bunch of. Other parameters, like thread pools. And. There's a lot really but, one of the other really cool things that we've done is a lot of customers, are actually curious about okay, you know there's some parameters, that i can't, tune, and. You know in our current experience for single server you're not actually able to tell what you, can't tune you can see what you can, so one of the improvements we've actually made within the azure portal is to also expose, parameters, that you can't tune just so you're, aware, of what the values, are, um because otherwise you have to go query, your server individually. But this is just a quick way for you to just get a better grasp, of all up. What the server parameter, values are for both the ones that you can and cannot tune. But i also recommend, going to check out the documentation. To get a better understanding, of how you can change the server parameters. And then also take a look at some, at the values in the azure portal today, we're working on exposing, more and more and so if you definitely have some server parameters, that you want to be able to configure, you should reach out to us over email. We we need that feedback to understand, what types of parameters, everyone likes the tune. Yeah thanks, thanks andrea and i think uh parameter, tuning is something that. Most of the. Uh developers, look for, uh, well they. Built their modern, mysql, applications. Um, so lucas, uh one of the attendees, uh, wants to better understand. The burstable. Compute type that we have introduced. And what kind of workloads, is more suitable, for that. For sure yeah and if you were in the session i think andrea had a good, graph on that um i can try to recapture it again, um so essentially, the the main benefit of burstable, is that it accumulates, credits, right so essentially the, like over time you're you're gathering credits as your cpu utilization, stays below baseline, so for example, you know baseline could be ten percent of cpu utilization. And so as long as your superiorization. Stays below that, it means that you're, gaining credits, and if you go above the, baseline, and then essentially you're using credits and that allows you to essentially go beyond, what the burstable. Server usually supports as a baseline. And lets you um peak up to 100, cp utilization. Um and that's really helpful for workloads, where you know, maybe. Let's take a school for example right like in the morning, um, like all the you know um, the pupils will sign in using the online, learning system for example, and you know you would have a peak essentially in the morning, but then over the day that will you know flatten off and during the night there probably won't be much activity. And so having something that essentially you know lets you support these peak and burstable workloads, is what the bristol is good for. It's also good if you just want to get started, you know you just want to provision a quick server, you do care about cost for example right you just want to have something to, maybe you know for development purposes, or to play around um and kind of get deadly, product, um and so one key change that we actually made to flexible server is that we now also support the vnet option and premium storage options, um so you know compared to the single server option where that was more limited on the basic tier, on the burstable, tiers essentially, a fully featured. Um kind of entry point uh entry price point, um so if you just want to get started that's also a great starting point there. Thanks thanks uh thanks lucas. Um. Along the similar line i have one more follow-up question for you is. Um, could you talk briefly, about the azure. Arcs, service that was announced today. Uh, and what uh what kind of workloads, is more suitable, for that. For sure yeah and and specifically. I assume you're referring to azure arc uh, azure arc enabled, postgres hyperscale. Exactly so today we had a couple of announcements, uh regarding azure arc which is you know our. Hybrid um story, um, so azure arc enabled, uh postgres hyperscale, is part of the azure, data services. And azure arc is really about you know running, the databases.
In, Environments that are not in azure but they're for example, on-premise. They might even be in gcp, or aws, right and so essentially you're using the azure technologies, such as postgres hyperscale. In the environment of your choice, um so it's really about giving you that flexibility, giving you the ability to run, the azure technology, in, other locations. Um and suppose because hyperscale, is you know as we talked about earlier, based on the stylus extension, and so if you want to run the size extension, on premise, and have it fully supported by microsoft. Then that's a great uh choice to use the azure arc option, um azure arc is in public preview starting today so you can just you know start, try it out um it's pretty easy to scale, um the other thing to know about the azure arc data services, they are actually built in kubernetes. So that means when we look at a postcards hyperscale, for example, it actually, on azure arc, scales out using kubernetes, nodes, and so you're essentially first setting up the kubernetes, environment and then you just like seamlessly, scale out seamlessly, add more nodes, um to scale out the postgres hyperscale. Cluster. Sure, thanks thanks. I'm going to make a really, quick. Um note here so on the screen here you'll see, the travis's, session, that's coming up. Tomorrow morning pacific time, which covers, everything, related to the azure arc announcements, that. Came out this week for ignite so definitely check that out if you are interested in learning more about parsers, sql hyperscale. With arc enabled. Thanks for the, thanks for that information. Andrea. I have one question for you. On the control. Element of flexible server that we are introducing. Uh could you talk a bit about the. Uh the scheduled, maintenance. And the custom schedule maintenance window that we have introduced, on, the flexible server. Yeah for sure so i'm going to switch over to the azure portal just because we haven't shown the azure portal just yet. Um let me just pick one of these servers so this is a existing, mysql, flexible, server that i provisioned. You can see here over on the left side of the screen. You can click on the maintenance, page. And what this does is it pulls up, the ability, for you to choose if you want system managed, or custom schedule, and so system manage is basically.
We Pick. Um base we pick a time to deploy, and any maintenance, to to your server that includes, uh maybe like a minor version upgrade for your engine. Maybe we have new security, patches. Or we'll release a new functionality. You'll typically see some type of deployment, from our team monthly. And so if you choose system managed, this will essentially, roll out during, what we consider an off-peak. Business hour for your particular, region, and it really just depends, on, the time depends on the region. Itself. Um otherwise if you pick custom schedule, you can see here. I then get the ability, to choose the day of the week, that i want this uh, deployment, to occur, as well as the start time, and so we give you a time window in which you'll get that um maintenance so really great for, any of you who have you know like business critical applications, where you know, um and you need better control of when any potential, downtime, may be, um and allows you to actually get more. More fine-tuned, control, for anything that can happen to your server. Thanks thanks for that. Andrea. Uh lucas we are getting a bunch of uh. Postgres, questions. Um. So, one of the questions that we are having from one of the attendees, is, can we install, custom, postgresql. Extensions. Uh and full text search files. In flexible, server. Yeah i think that's a great question and you know with uh, my personal, um background on the post post-processing, of course you know i would love to. Allow all extensions, and allow you know anything that can be installed to be installed. Um, i think, unfortunately, you know the the reality of a managed service is a bit different um it is actually you know, important for us to verify, the security, and make sure that you know there's no risks essentially that come, um, to the database, itself from you know allowing any custom code to run, um so today, um, the the way that it works flexible server is pretty much similar to single server which is that you know it is a kind of you know list of allowed extensions, that we we manage. And we are planning to add a lot more extensions, on flexible server, um on the specific questions, of kind of text search and like other you know custom files that are not extensions. That's a very interesting use case and i think that's actually something that i, i personally didn't know about and i learned from, like a customer essentially asking us about this i think a couple of months ago, um and i think, this is something that we're exploring, we don't have a good way to do this today, again security, is a risky right like how do we make sure you can upload things and essentially it's safe to run on the database environment, but it's definitely something that we we are evaluating, we're trying to see if we can support in the future.
Okay Thanks thanks for that um a follow-up question, on uh. From one of the attendees, is around azure, arc. Um, azure, arc enabled data services, will allow, to run postgres, on, azure, stack. Uh yes um yes it is that's a great question um so yes if you have an azure stack deployment today, um or you're looking to purchase azure stack, you can run the azure arc data services on that. In fact, the azure arc team is collaborating, closely with the azure stack team, it's part of the test pipeline, to make sure that azure stack and azure arc work well together. And that includes running postgres hyperscale on azure stack, using, arc. Okay. Andrea. A question, for you. On the burstable, option. Is it on the roadmap. To, increase, we course option. For production. So. Um. Like maybe lucas or i mentioned earlier, um, part of the re-architecturing. That we've done for our, base. Platform for flexible, server, is to start using. Public compute, vms. And so some of the things that we're looking into, is, as the compute team lights up new vm skus, how we can potentially, adopt those into our service, so as stands right now. With the b series compute, we have support, for a one to two v cores, um and so potentially, if there is more. Uh support for larger v cores in the burstable, series, as it comes then yeah we'll definitely look into potentially lighting that up, in flexible server as well. Okay, so we have, last two minutes, uh. Uh a quick question, for um. For our own, uh do we support oracle to postgres, migration. Uh that's a great question, absolutely. It's, uh we've been helping, a, number of customers. Move their, oracle workloads, into postgres. In fact, we have a big motion going on going on around it, uh, the same data migration, guide you saw earlier. Uh, has, all of the best practices, that you could go look at. From, an. Oracle, to. Azure database, for postgres, migration. It will walk you through all of those steps we talked about. In fact. We have brilliant, field teams that are that help our customers. Who have come up with. Ways. And scripts. That can help, automate, different, parts, of a tough heterogeneous, migration, like oracle to postgres. So, uh if you have an oracle, workload. Uh you'd like to see it run on postgres. Please feel free to reach out to us please feel free to check this guide and let us know what you think. It walks you through some of the open source, tools available, out there best practices, around. How to go, go and install, and run them like order to pg, and go about using. These open source tools, and. Fully manage services, like azure database, migration, service to, move data, and, you know prepare, and fully, uh finish. Your. Migrations, to posters, from oracle. So that, there are several, uh, details, available, in this guide. And. If you have more questions please feel free to reach out to us. Thanks. And uh, thanks a lot uh, lucas, andrea. And everyone who joined this session, uh and the wonderful, questions that you had, thanks a lot.
2020-10-02