Deep Dive: Modernize your Apps in the Cloud with Azure SQL Managed Instance | Data Exposed Live

[Music] hi i'm anna hoffman and welcome to this special edition of data exposed where we are live that's correct we are streaming live from actually all over the world to multiple channels we're so excited to have you here today we're going to be talking about managed instance we brought i think six or seven people on today so it's gonna be a great show again super excited to have you all if you have questions or comments throughout the stream feel free to post them on learntv twitter or youtube or twitch or wherever you're streaming from and we'll try to answer them either via chat or live on the show um so with without further ado i want to bring on borko uh who is going to be helping me host this episode uh borgo thanks so much for joining us today can you tell us a little bit about you and your team thank you anna for having us in this show my name is borko novakovic and i'm leading the team of program managers who are responsible for designing and developing azure sql managed instance i'm with microsoft for more than nine years and now also i've been part of sql managed instance team for since it's beginning for four years now uh uh prior to microsoft i work as a.net developer and technical leader and also sql developers so sql is really in my blood and i'm happy to have also a couple of members of my team here live awesome thanks borko uh we're going to go through introductions for all of our awesome speakers today uh the first one i'm gonna bring up is moladen thanks anna and hello everyone my name is vladimir i am a program manager in azure sql team for the last four years with 15 more years of a previous experience in the financial industry uh within azure sql managed instance i'm focusing on the sql server compatibility and business continuity awesome thanks maladin and we're going to learn lots more about the compatibility and bcdr in this session next we have sasha hello everyone i've been in sql team for 10 years and i've joined joined the program management team this summer i'm working on distributed transactions and on data mobility scenarios thank you awesome we're happy to have you and i think everyone's gonna be looking forward to what you have to say and show us related to dtc uh next we have danny hi everyone my name is danny lapava i've been with the manage instance team for the past three years i'm program manager within team before microsoft i have about 15 years experience working in the industry from the silicon valley us my focus is on hybrid data mobility high availability backup restore and intelligence and thank you for having me thanks for being here and you work on just a few things it seems like next we have certain hello anna hello folks uh glad to be here with you tonight uh and uh my name is sergeant bojowicz i'm program manager with sql managers team uh focused on networking and security aspects and i would be glad to share some things around that uh during today's session awesome yeah i think everyone's gonna be looking to you to learn about the security topics and i have a few questions myself so we'll see if we can squeeze those in and uh finally last but certainly not least we have urosh here uh welcome hi anna thank you uh i am urash milanovic i joined this super team like mid last year and my focus area is instastrad and instance management operations awesome so as you all watching can see we have a ton of great resources here today the folks that are actually building the mansions product uh so i think you're really going to enjoy today's show so be sure to uh stay with us and learn from each of these program managers throughout the session we're going to do is we're going to actually take a step back and borko and i are going to discuss the azure sql family because maybe you've been with us and hearing a lot about azure sql for a while but maybe you're kind of new and wondering what manage incense actually is so borgo do you think you could help us understand what the azure sql family is and where manage instance falls into that sure anna yes so basically uh on this very slide you can see uh the place and position of managed instance within a wider azure sql portfolio and as you seek we really represent a portfolio of sql products running in azure offered in two big or major deployment and control models infrastructure as a service and platform as a service right and in infrastructure as a service we do have certain virtual machines whereas in the platform as a service we do have two uh deployment models uh managed instance being one of the of them and azure sql database now of course with azure sql our sql customers can do pretty much anything starting from lifting and shifting their application modernizing existing application in the cloud or developing new applications uh in in the cloud as you can see uh on this diagram manage managed instance is really centered between the virtual machines and azure sql database and there is a reason for that because managed instance was designed to be the best place for sql customers who are looking to migrate and modernize existing application in the cloud and as such circle manage the instance try to bring best of the both worlds uh uh and in one in one hand it's it is really a uh rich uh surface area uh of sql server and high application compatibility and on the other hand it is a really set of very uh powerful platform as a service capabilities that can help our customers to get rid of the database management and focus their time and energy on high quality tasks uh what is also important uh being a part of azure sql family managed instance actually shares uh with them the set of tools and apis and basically if you were using sql database or if you're using sql server until now you'll find yourself familiar with manage the instance now uh typically our customers ask the question okay in azure sql there are many options uh which of these options i should use and particularly when i should use sql manage the instance so being a product group came up with a relatively simple model and we think that only two questions our customers should be asking themselves uh when they deciding which option to use the first question is that what is the level of control that our customer need to have over the database layer right and if customers prefer to manage directly their databases and have the full control including the operating system level access strict uh control over the sql version uh controlling the patching and upgrade cycles then obviously the right answer for our customers should be sql server and azure vms but yes with the control uh there is also comes a lot of uh uh responsibility and we've seen that many of our customers are actually willing to delegate management of their databases to microsoft and then uh use the their time and focus to improve their application their business and their workloads and if that's the case then in microsoft managed world the next question our customers should be asking are themselves is hey what your application require what capabilities of sql are needed and if your application is really uh relying on a rich surface area of sql server including the instance level capabilities and we put a list here on this slide like native restore and log shipping and press database transactions and queries or net integration or transaction replication and such then really yes you should be going with azure sql managed instance but if your application dependency on the database feature set is much more narrow or simpler or scoped to the database then sql database should be uh an option good to go for you from the day one i hope this quantified clarified a question that our customers might have awesome thanks marco i really like this because we're trying to make it really simple and you know when i talk to bob ward about managed instance he always says uh manage instance is going to allow you to get the best of sql server plus the best of like a managed platform as a service offering so i think that's really cool and i think for our sql server folks uh this is great news for them because they can kind of remove some of that management overhead and focus on the things that they actually want to uh spend their time on and today you know i think there's been a lot of announcements over the past few months so we're going to spend a lot of time diving into the new capabilities i even saw some capabilities i didn't know existed managing since today in your previous slide uh so looking forward to to learning a little bit more about some of the things that were flashed there like a sneak preview um so i thought it'd be great if we started with uh things around compatibility and things that make migration easier and for that borko i thought we'd bring uh bladden on to share with us exactly yeah thank you thank you so there were so many things to show and tell that we decided to split the overview into multiple areas or value streams how we call them and we are starting as anna said from the sql server compatibility and seamless migration now uh before i start let me just quickly explain the frame that we are going to use through the rest of the of the session so what you can see on the screen is a vertical timeline divided into three parts so the past got the available section containing some of the capabilities uh that we introduced during calendar year 2020 announcing section that's the present time right uh or features being announced these weeks and coming soon section covers upcoming capabilities that are going to be introduced over the uh couple next couple of of uh months uh to uh and and like where we are going to invest more time explaining are the green items because those are the ones that we introduced uh very recently so let's let's let's start going to these items i'm starting from the machine learning services the preview has been announced back in september so with machine learning services you can run both r and python scripts to perform data preparation or general data processing or you can train your machine learning models within the database where your data data actually resides or if you have your existing machine learning models uh what you can do is basically to deploy them to the manage the instance in form of a stored procedure and then run run them there on the existing data and diminish the instance in the future with the additional integration capabilities with our other azure services you will be able also to reach to the data in blob storage data lake and so on the next feature is related to migration of solutions involving sql server reporting services structure if you are preferring fully managed software as a service solutions then your best pick would be to move your existing sql server uh reports to the power bi premium service and you know stop stop uh stop taking care of any of the managed management activities however if you prefer to steal uh host your uh reporting services in azure for example in your azure vm uh now what we introduced recently is that you can host uh this catalog databases to very important databases that ssrs runs on within managed instance uh just to explain a bit more about that sql server 2019 introduced native support for hosting this catalog databases of ssrs we didn't manage the instance however we now enabled also managed instance to host these databases for the older version versions of sql server so basically if you have a a a little amount of time and you have to migrate quickly and you don't have time to migrate to sql server 2019 for reporting services you can now stay on your previous version like 2016 or 2017 uh host this ssrs in a vm in azure and then host this catalog databases in managed instance and of course use the data and query the data from managed instances uh from these reports the next feature that i would like to to quickly touch upon is our distributed transactions so these distributed transactions transactions now available in managed instance are a great example of a feature that at the same time unblocks the migration of the existing applications from the on-premises uh however it also unlocks possibilities for creating a brand new modern applications using these capabilities but i i want to go more into the details because my colleague sasha will talk about this in in in a couple of minutes uh i would also would like to mention the uh log replay api so uh that's something that is coming very soon danny will talk more about that in a minute i just wanted to to say uh briefly about that so this is something that we already have been using internally in our migration tools for quite some time and we are now exposing that to our customers but more about that later and finally i would like to talk a little bit about cross instant service broker so series broker is really powerful component of sql server uh that that basically brings native and built-in capabilities of reliable messaging and queuing now regardless of whether you already have some applications using service broker on-premises for example something with a multiple self-contained services exchanging messages between the databases behind these services or if you are planning to build some modern solutions uh service broker is really a very powerful component that will help you in both of these cases now uh managed instance previously supported uh exchange of these messages within the same day uh within databases of the same instance however we now do support exchange of messages across the instances and thanks to the uh certificate based security uh these instances can be placed like virtually anywhere in the world in any of azure regions within the different subscriptions different regions as i mentioned and even within different azure aed tenants so another example of a building block that could help you build powerful distributed applications now speaking of distributed i would like to hand over to my colleague sasha to tell us more about distributed transactions hello everyone so distributor transactions for azure sql managed instance have been in public preview since october 2020 so it's a new capability this capability is useful it unlocks migration to asia for apps that already use this on-premises and it also enables new uh development scenarios so modern applications um they often need to utilize multiple managed instances and databases and partition data horizontally or vertically so in all these cases distributed transactions are a useful and needed feature managed instances can uh reside in different regions and in different virtual networks we have cross cross systems collaboration secured with server trust group entity and we are starting this preview with managed instance as the only supported transaction participant and with net and t sql as supported application layers a bit later we will see the demo um i would like to hand over to anna and danny awesome thanks sasha you know this is something that folks have been asking for for a long time uh so it's great to hear that it's coming in preview uh or it has landed in preview and we'll look forward to seeing it in action uh in a bit uh danny i'd love if you could take us a little bit deeper into the log replay service i know this is something that uh leg mod and mentioned has kind of been helping us internally i wasn't aware that we were planning on kind of making this available what what what does this mean for our customers certainly if you can maybe look at my slide i'd like to present a log replay service something they've been using internally so far to power data migrations using dms from sql server to manage instance and now we're exposing this technology externally because we've listened to the customers and we like to make it possible for customers to customize their migration experience so this technology is based on the existing sql log shipping which is well known in the industry it is database scoped meaning that you can move database per database so you're full in control of the entire migration process and you can cut over when ready so how does it work as the first step you would use your source which is your sql server hosted anywhere it could be on-prem in a data center in azure vm or in other clouds you would set up the log shipping from your sql server to azure blob storage as a continuous job so this would run every couple of minutes and would ship the backup logs to azure blob storage and from there you would start the log replay service which would pick up those logs and replicate them onto the managed instance and essentially this way you would have synchronization of data from the sql server to managed instance during this replication time the database is being replicated however it is not visible to customers and when you're ready to cut over when you make sure that everything is working fine then the last step using log replace service is to initiate a cut over and your db will be visible in the cloud on mi and this is how we are empowering our customers to build their own migration solutions awesome thanks danny i think this is really cool because we're saying hey this technology worked really well for us uh we're going to actually release it so you can kind of build it into your custom workflows to help you in migrating and this is something i hear a lot of requests for so really exciting uh to hear it thanks danny and and next danny i wanna to kind of shift gears a little bit you know this next set of improvements so believe it or not viewers like that was just one set of improvements i think we have like four or five more sets of improvements uh but this next set of improvements is what bob ward and connor cunningham refer to as the meat and potatoes of sql server and uh you know when they say that they're talking about performance security and availability so uh the manager instance team is doing a lot in all of these spaces but i thought you know danny can we start by talking about some of the exciting recent performance updates exactly and before i jump into the performance i just like to mention that uh and it all comes into performance as well that we have considerably improved backup options available in managed instance in terms of the improved compression so there's nothing that you need to do the com the better compression will be enabled automatically and we have enabled some backup cost optimizations that i will talk about uh further in our in our presentation um another cool improvement that we made is global v-net peering as you're perhaps aware managed instance is deployed in a secure v-net and it is um now possible for you to connect v-nets to separate v-nets using azure backbone earlier you had to do it through gateways through routing through your own um on-prem or data center but now we have enabled this connection to go directly through azure making your failover group scenarios much easier finally big improvements in terms of performance there has been improvements on both gp and bc service tier in gp per instance we're having a significant increase from 22 megabits per second to 120 megabits per second for increased log rate and then in this table you see per dip per database new limits in order to get this tweaked you need to increase the size of your files and you get all this for free so no additional cost um on bc um again for the same cost getting better bang for the buck we've increased the log rate from from 48 to 96 megabytes per second and 2.5 k from in iops from 2.5 k to 4k um what i'm going to show lastly are the test results so uh running the industry standard tpcc and dpch tests we see two to three x uh significant improvements on on both service tiers um if you pay attention to the link uh at the uh in the lower left corner and and because this is gonna get recorded you can also rewind uh that's where you can find out more information about this particular improvement awesome danny this one seems like pretty big it seems like you know at no additional cost uh you can just kind of be faster which is huge uh you know i do have a question like what do customers need to do you know if they have an existing mi to take advantage of these capabilities so literally uh they just continue using it the only thing that they might consider you doing on gp is to increase their file their db file sizes to get their respective speeds as indicated in this little table that's it awesome so apart from that small tweak that they might need to make they're just going to wake up one day or maybe they've already woken up they already have it it's already out there yeah yeah it's only awesome cool that's awesome danny uh you know you talked up briefly about some of the backup compression and backup cost optimization can you tell us a little more about that certainly so what we've done at the back end and there's nothing you need to do about it we made a better backup compression about 30 but this means that your cost for the access backup usage is going to be lower and that's all automatically provided to you further we have put in some additional knobs for you to fine tune your backup retention so earlier you could retain your backups your short term backups from 7 to 35 days and now we're making it possible to fine-tune your need for having backups or active databases from 1 to 35 days and in addition we having included a new knob uh the ability to set the retention uh period for deleted database so once you delete a database you can set up the retention period from 0 to 35 days zero would essentially mean that you do not need that backup and that might be useful let's say for large databases doing analytical workloads um and essentially data that you do not need but by not keeping the backup that's how you optimize your backup costs and finally we have made this a very significant improvement in terms of uh giving empowering the customers to choose the type of storage for their backup needs so earlier we only had grs which is global redundant storage that means that your backups are replicated in two regions so globally for the safety reasons however and and that type of backup is backup storage is of course the most expensive however now we're making it possible for customers to choose at the time of provisioning manage instance zone redundant storage and locally redundant storage so zone would contain uh would contain multiple backups within a uh within a zone that your manage instance resides and local redundant storage means that the backups would reside only on the local mi this is not only important from the backup cost optimization but also from the data residency because you control where your data is in terms of the backup and lastly if you look at the link i've provided below that's where you can find more information awesome thanks danny uh again a lot of things that you know customers have been asking for and we're you know putting right back into the product so it's really exactly exactly that um i think next we're going to take a look at a few of the capabilities uh we've looked at so far with sasha i'm sorry in a demo i think i have that right sasha is it demo time or yes it's right it's demo time so um as i mentioned distributed transactions are useful for uh migrations of existing applications to asia and also uh for the new development scenarios so modern applications often need to split their data either to overcome limitations of individual instances for example storage size limitation or they need to do distributed data across geographies to achieve better latency and performance we call these applications and databases sharded applications uh and in this demo uh we will use uh net library uh called elastic scale uh and with this library this library what it does is it sets up the sharding and assigns different tenants to to the shards uh and also uh when applications need to connect to a certain tenant they use this library to to get the right uh connection so uh we will show several scenarios where distributed transactions play important role uh let's uh briefly explain the architecture of our demo application this app deals with customer orders and product deliveries across the globe we have several shards these are instance one and instance two in these shards we typically have two types of data it is local data which in our case is customers and their orders and we have uh master data main data that is copied we have a copy of this data in each of the shards in each instance so our local clients for optimal performance they typically connect only to the local shards and use data in the local chart we have one example here and the other example here and we have set of admin applications these applications are in charge of maintaining the master data in all shards and they also maintain the shard map database so when we have a new client a new tenant admin application is assigning it to a certain chart so in order to have all this working typically uh globally distributed instances are on different virtual networks so we need to set up we net peering between them and we also need to set up a server trust group in order to have distributed transactions working so uh let's how let's see how this setup uh works awesome uh so uh i have uh one virtual network here uh and i will go to peerings in settings section and i will add new peering so i need to give it a name uh this workflow is automatically setting up uh appearing in the opposite direction so we give name for that one as well and we will choose uh the second virtual network name and we are ready to add this peering let me see if this will work okay just a moment awesome thank you okay and while this is happening i will go to manage distance to set up a server trust group okay so sasha while you're setting up the server trust group can you tell us a little more about uh how it relates in this scenario so in this scenario uh what uh server trust group is doing um it is um setting up uh certificate based trust uh between the instances so under the hood uh all instances in the group they exchange certificates uh and after that distributed intersections can work um in a in a secure fashion so uh here it is i gave it a name and i have chosen the the other instance and i will click save and it will work so in order to save time i will switch to um to the demo application that is already connected to another pair of instances where all this is already set up so um what we have here uh is let's quickly take a look we have local client application connected to west europe shard and here we see products that's the master data and we see customers and orders which are local data for this chart we have north europe application and this one has identical products and customers and orders are local for north europe and in the first um in the first demo what we are going to do we want to add a new product so in a single transaction we want to add it both to west europe and north europe chart so in order to do that this admin script will call start procedure create new product let's take a look at the code so what the store procedure does it begins distributed transaction performs insert into the local shard then uses linked server to perform insert in the remote shop remote chart and commits the transaction so let's see it in action we will execute this and we have new products in in both charts the second uh demo uh that we will see uh is a bit more complex so what we want to do here we want to use this admin application to move tenant customer 102 from west europe to north europe uh so what will happen here uh is we will move all the data related to this customer uh to the north europe so both data from customers table and from orders and within the single transaction we will also update the shard map so here we see the metadata uh from this chart map so we see that customers in the range of hundreds are connect all uh attached to the west europe uh and customers in the range of two hundreds uh they are mapped to north europe so when we run this move tenant in a single transaction and let's quickly take a look at the code snippet so in dot net we start we create new transaction scope it will map to distributed transaction we will move the tenant data we will delete the old mapping and create new mapping and with the scope complete we will mark transaction as successful so it will be able to commit so uh let's see that in action so from from west europe to north europe we are moving canon 102. so it has moved to north europe uh its order has moved and we see here that mapping is no longer in west europe but it's in north europe awesome wow that was pretty cool yeah and and with this i would wrap up the demo and hand over to yuana and danny again awesome thanks so much sasha you know it's really cool to see because you combined a few of the new technologies uh not just dtc but also this global v-net peering which is making it a little bit better for our customers in a few ways uh so thanks so much sasha uh we are moving onward and forward through the many updates that we have related to managed instance uh if you're just joining us live be sure to uh tweet at us use twitch use learn tv comment uh let us know your questions because we we can answer them in the chat and we can also answer them live near the end of this session okay so we've talked about some of the latest latest updates related to compatibility and migrations we also talked about some performance enhancements you just saw a really cool demo from sasha and now we're going to continue on talking about the meat and potatoes and talk about vcdr or business continuity and disaster recovery and for that i'm going to bring danny back up to talk us through uh some of the many updates that that we've been working on in that space thanks danny thank you thanks very much and if you want to look at have a look at my slide uh there are there is a group of features related to bcdr uh that we have delivered in in this period these are all g8 features so geo restores provides a capability for uh you to restore a db from another region to from one region to the other that's in case that region is down you can get back up and running in another region provided that using grs storage as mentioned earlier is one of your backup storage options we're enabling pitter for pointing tom restore for deleted databases and that enables you to recover a database that you might or databases that you might have accidentally accidentally deleted accelerated database recovery is also provided all built in it enables the recovery of databases to be much faster that is instantaneous and there is no longer need to wait for a long roll actual box is instantaneous and finally danny sorry sorry to interrupt uh could you share your screen we seem to have lost it on our end um and i don't want folks to miss these new things certainly thank you awesome should be good now yeah yep so these are the things that i mentioned in bcdr space and then finally cross instance peter cross instance point in time restore provides the this amazing ability to move your dbs your backups from one instance uh to the other as we're making automated backups on every managed instance now it is possible within the same subscription to restore those backup on another mi really adding to this bcdr scenario significantly that's really cool i didn't know that cross instance uh point in time restore was a thing so it's cool to learn learn about it is there any other items in this space that you all have been working on i can't we can't disclose that at this time but there are gonna be some pretty cool things coming related to even greater data mobility awesome well we'll look forward to having you back on i'm gonna i'm gonna uh take note of this that you're gonna come back on when we have some some more updates that you can share with us in the meantime i know there are some security and compliance related things you all have been working on are there some things you can share with us in that space so this is i think uh surgeon is the guy for it oh awesome awesome okay sorry danny alright so we're gonna kick danny off we're going to bring certain on and we're going to talk about security and compliance uh welcome certain you know i was trying to pass off pass off all your hard work to danny but you know i think you're the better guy to do it well we work together really in a in a great way so um no worries about that uh and uh if you see my screen now uh there are these some of these uh improvements that you talked about listed here so basically from from the security perspective there is this uh tls 1.2 enforcement and uh we are actually enabling our customers to to choose uh to use the latest uh cypher suits when connecting to their managed instances making their environment uh overall more more secure and we plan to continue uh doing that in the in the future with next tls tls iterations this is not just a security thing but also it's important from the compliance perspective i would say ah because uh many regulated industries you require to to use um kind of the latest and strongest uh cypher uh cyber suits as well uh and uh when we are talking about uh this compliance thing uh i think that many of our users would be happy to learn that we finally have long-term backup retention uh in preview um because uh all of those uh customers they have this requirement like often like 10-year period you know to to keep their backups and uh until now they had to do this copy only backup so we provided option but this is this is really kind of platform as a service thing and uh it it it really looks great when we are kind of talking about the backups uh it's also we tried started supporting backup data residency uh danny had a chance to talk about it before but i would just kind of clarify that this choice is also from the compliance perspective very important uh because uh when we are talking about the azure pay regions they're sometimes kind of more paired from the technical perspective but uh from the geography perspective uh their requirements that uh data stays uh stays locally uh and uh on this front uh we also not not just kind of bring more compliance things uh to our audience but we also um try to ensure that uh uh their uh their configuration always is kind of compliant with the requirements of our service and that these compliance are as most restrictive as possible so that security footprint would be um would be best and uh with that we have this uh service aided submit configuration feature where we kind of um keep our customers uh subnet where managed instance is deployed always aligned with the kind of latest uh subnet security uh configuration there so so really really a good good thing uh one thing even better uh is that on top of this subnet configuration and service endpoints we bring uh service and point policies this is something that is uh that is coming soon uh and it's it's really great because it would enable our customers to define policy and uh basically uh define a set of uh storage accounts uh where um customers could connect from manage instance obviously over over this private type space so so for example um me as a customer could uh set up and say okay only accounts and this subscription uh should have access or only accounts in this resource group and up kind of single accounts uh could could be put there uh there as well um a quick question certainly you know what scenarios would these service endpoint policies be uh best for like what type of customers or customer scenarios are using those yeah well i'm glad that you asked that so so basically uh uh there are multiple standard areas uh really kind of where this sql is connected uh to storage um one of these scenarios would be um these uh with customers want to do copy only backups the other scenario is this exciting scenario that danny already talked about and this is um this log shipping scenario uh where actually these uh uh logs are on this temporary storage end then customers want to kind of lock lock down the access to the certain storages then auditing is well another example bulk insert another one so so really kind of when you're talking about sql and storage there there is really a tight integration and we want to keep these uh azure services uh very uh in a very closed loop uh basically so uh with that uh we're just kind of moving uh to um ajd uh directory readers and uh i would have a separate slide on that this is i would say a long-weighted uh thing uh for most of our customers that uh would like to use azure id authentication as a form of a password less i would say more secure and more preferred way of authentication and and customers that operate in their regulated industries uh the thing here is that we kind of we heard a lot about this feature and we tried to really simplify it so uh basically uh before this feature landed uh it was required by global admin to approve each and every managed instance to use azure id authentication and now this is really a well simplified instead of approving each and every managed instance global admin could delegate that privilege basically to a group and then members of that group could add additional managed instances into this directory reader's role so which was once kind of tedious setup now uh it's it's pretty weight simplified it's kind of one-off engagement with tenant admin and from there from there everything really goes go smooth awesome thanks so much uh this is awesome to hear about uh personally this is something i've been hearing a lot about from customers so i know uh folks are gonna jump right on uh this opportunity so thanks for sharing it with us uh you know next we're gonna bring danny back up uh hopefully it's danny and not sure dan hopefully we're bringing danny back up there and sorry danny i put you on the spot earlier uh but we're gonna talk about uh monitoring and manageability and i think this is really gonna be important for our sql server people because uh you know they're used to having control that's a lot of the reasons why they land in management so would love to to hear from you what's new in that space yeah exactly and i hope that you can see my slides so we've done a significant number of improvements in the monitoring space we're making it now possible for folks who already have scom monitoring environment to monitor managed instances as well which is awesome for legacy uh applications and hybrid scenarios we're making it better now to uh make to actually uh set up alerting and manage them from portal it's right built in into the mi uh mi blade uh in azure portal and we're making it easier now from the azure portal to query log analytics for those folks who are using log analytics for alerting and monitoring further we're improving the manageability area by offering customers possibility to execute user-initiated failovers to test their application resiliency and i will talk about it more in one of the next slides and finally coming soon we've we've realized that we want to do better in terms of health resource health monitoring and we're going to be making some improvements in this space such that you are well aware when of your health state of managed instance with this i would like to briefly expand upon the scom management pack for mi we've the management pack is free of charge downloadable through our website for all those folks using this com environment install it plug it in and monitor your existing sql resources and mi at the same time it just works for more information please see the link shared below in terms of the other options available besides com to monitor manage instance for example sql analytics and other tools that we have available so your comprehensive click on the link below for your comprehensive guide on the mi monitoring finally i'd like to talk about a really really cool feature a cool scenario that we're making available and that is we're now enabling uh customers to manually initiate failovers on managed instance why is this important failovers are part of our service pass platform platform as a service because during the software updates which we do regularly uh on on managed instance for a brief amount of time there is a failover and for a brief amount of time of just a couple of seconds there is an availability to an application when we switch from one node uh one node to the other how that works i'm going to show for gp and bc service tiers for gp it's a single node mi as you can see in the diagram and upon a failover what we do is we take a one of the spare nodes from the pool of of of available available nodes and we switch over to a new node this failover is a built-in high availability of gp managed instance the failover is transparent meaning that the application does not need to change any connection strings or any uh or any dns settings is just uh auto magical as i like to say it for for an application um however in order for applications to be um to be cloud applications to work well in the cloud customers would need to implement a retry logic to be resilient these failovers so we're now making it possible for customers to test this on the mi bc tier because bc tier is a more robust it has always on availability implemented which we see with sql and prem so with mibc this is what you get you get always on in a single package an application can use the primary replica for reads and writes and could use secondary replicas for read only upon the failed over as in always on what happens is if we're failing over the primary the secondary and primary switch to place and we're also making it possible for applications for using read-only also to fail over the secondaries to test this fully um on the bc service tier so with this um it is a powershell command um please follow the link shown shown below um it is possible to fully test your application resiliency in the cloud and make sure that your application is going to be praying just fine awesome thanks danny uh i think this is a really awesome capability that again just goes along with like we're trying to give more control we're trying to make things easier uh we're taking customers feedback and actually implementing it right in the product so exactly exactly that's awesome please keep it coming yep and uh next uh we're going to bring out uh yurosh who has some new features and improvements around a few other topics and i know we're getting close to the end but you're going to want to stick around uh for this one so over to you euros thanks ana this is definitely one of the one another area that we have invested a lot so far and the thing i want to to start is something that was there for quite some time and one thing is instant storage scaling where your general purpose instant storage can be scaled in a time many times and another thing is switching between harder generations which was especially important for our customers that we're running my distance on gen 4 hardware which is now the deprecation path and that we advised uh to to to switch to gen 5. so these are just some new scenarios that we were adding in our uh scaling space another thing that we invest a lot and that we are still investing is the reliability and speed of instance management operations and as you speak we are working on publishing some new uh documentation pages and some updates related to the deployment deployment times so besides the scenarios and the reliability the next important thing is the management operations transparency where customers can uh fully monitor their deployment processes process their scaling process or if they are using uh some uh or they if they have some script scripts they can enhance them with these management operations api that is public publicly available including the powershell and the cli coverage one one more important thing in this area is the maintenance window picture that is coming soon in the preview mode and that my colleague madden will cover more in detail awesome thank you thanks zirosh this is a lot of stuff i know the transparency uh that's a huge one for our customers along with these other updates uh finally i'm gonna bring latin up to uh talk to us a little bit more about these maintenance windows thanks anna yes so we wanted to finish this session by announcing another upcoming feature a long-awaited feature i would say for azure sql this time bought for sql database and managed instance this is a maintenance window so let me quickly set up a stage for it so danny already did a great job uh sharing some of the best practices for applications running in cloud pertaining to the resiliency of these applications retry logic that every application should have within uh within its uh within it now uh now we are aware that there are also third-party and legacy solutions out there and we are aware that not all of the solutions can do that overnight and build build this uh resiliency into into applications so that's why we came up with this maintenance window solution and that set of predefined windows preset windows that we are offering and customers will be able to choose between those windows uh we are following uh the most common usage patterns like weekdays or weekends and uh of course even though it is based on the on the weekly cadence we are not going to use this each and every week basically these maintenance events uh with the causing failures happen once or twice a month now very quickly since this is a preview there is a couple of limitations uh like this is uh currently limited to the types of the subscriptions where typically customers would run production workloads like enterprise agreement cloud solution provider or pay as you go solutions and then we are starting with a subset of of regions now another important thing for people to know is that once you opt in for the maintenance window you can expect a brief failover so we can onboard you to the feature like this is completely the same experience as for example if you are resizing your instance now uh the last thing there are very rare exceptions possible still even do maintenance window is introduced like once in a year or even less frequently there might be some urgent security patches that are more important that we would have to to immediately apply and in those situations uh maintenance window might not apply so just to visualize uh this what we are starting with here are the two uh windows uh corresponding to the typical to the typical workload pattern so one of them is weekdays monday to thursday and another one is weekends so this is what we are starting preview with and uh we may shrink these windows over time and we might might be adding additional windows uh we will carefully analyze your feedback to uh to come up with an optimal set of maintenance uh windows so you you as the customers are really encouraged to share your use cases with us so that we can define it uh define the final shape ship with that uh i think that we are pretty close to the end and i would like to thank awesome thank you uh aladdin uh i think maintenance windows is huge for our customers so i know people are going to be really excited uh so they can kind of plan for when these updates might happen uh now in our final minute i want to bring everyone back on uh you know we've had a great show we've jammed in so many updates we're gonna share more resources with you after this live session on learntv our youtube channel on twitter so make sure you're following us on all those platforms uh borco do you have any closing words to close off this awesome session on manage instance oh sorry so thank you anna again for for having us and and i just wanna tell all the folks who attended this is not an end of mi journey uh uh an exciting year is ahead of us so please stay tuned for our new improvements and new announcements and we hope to get again to this show and talk about these things awesome well we're gonna take you back up on that borco and the rest of the team thanks so much for joining us today uh for our viewers uh thank you for tuning in whether live or recorded we're happy to have you on the show we're gonna be starting to do these live shows every wednesday starting in january 2021 uh with that i want to thank you all again for joining us and we hope to see you next time on data thank you thank you thank you [Music] you

Channel: Azure SQL

Views: 1,841

Keywords: azure sql, azure sql database, Azure SQL Managed Instance, Azure SQL MI

Id: UBwy5Z3Z3JY

Length: 61min 30sec (3690 seconds)

Published: Wed Dec 09 2020