Recording Day 1 Azure Stack HCI Days 2021

Show video

yeah hello hello and welcome to the second azure  stack hci this time days we have two days my name   is carsten rachfar and i'm joined by manfred  helber who is so nice to share his nice studio   with me so that we can do the next uh two days uh  azure stack hci days here together so manfred some   words about you yeah um welcome also from my side  i'm very happy that carson is here for the second   time for the asiastic hci days last year it was  the azure stack hci day and before we had the   just like hci day we met several times on your  cloud and data center conference uh what was a   yeah let's say in person event a real life event  and because of the situation we all know we carson   had to switch to a virtual event and i had already  this video studio here carson mentioned and so   i'm very happy that we are using this for carson's  azure stack hci days and also for several events   i'm doing all over the year and i'm really looking  forward to the many interesting tracks who you   have on your agenda so yes yes in the last days  yeah we will talk about all the tracks and the   speakers and of course the sponsors very soon but  first i want to say some words about the cloud and   data center management conference um unfortunately  we couldn't do it because of the pan pandemic   last year in 2020 this year also not possible and  i'm to be honest next year there will also be no   cloud and data center management conference we  are planning now for 2023 so we hope that then um   corona virus is something uh we get used to every  everybody or nearly everybody is vaccinated or had   the virus and is immune so next year not a cloud  and data center management conference but i hope   in 2023 and i hope we will do another azure  stack hci day in 2022 manfred i hope so at   least and you start with a new show on the first  of october i think you will mention it in uh in   your presentation yeah and for you yesterday now  because you do a new thing yeah there will be an   azure stack hci show this show is presented by  sven langenfeldt who is a microsoft employee   and my person and we have great guests there in  the asia state hci show for example karsten in the   first show and maybe some of you know eric burke  he is also microsoft mvp um he will be also in the   first show in the third show i think so we will  have cosmos darwin uh will present there so the   show as the name says is all about azure stack hci  and if you are interested and you want to watch it   you can find it on youtube when you search on  youtube for manfred helber so my name then you   will find my youtube channel and there you will  find the show um it will be online it will be on   the 1st of october from 12 to 12 o'clock to 1 pm  and the youtube link will avail be available on   monday so it will be listed there on monday but my  channel is there already and we also do a little   thing together the on-prem show the on-prem show  there was a large summer um how you call it pause   uh that's mainly my fault because i i was a little  bit in a corona down and uh didn't do webinars and   other stuff a lot so we will start again i think  also in october in october yeah i think in the   middle of october our next uh on-premise there is  a date yeah and it's also streamed on youtube okay   so let's uh let's switch to the slides and uh do  some some logistics um so um for your information   we will talk about this sessions very soon but  first the sessions always start at the full hour   so the first technical presentation after we talk  a bit about the azure stack hdi day here manfred   and i the first session will start at uh 1pm  or in in german it's it's 13 00 and the session   will last up to 55 minutes i talk to the speakers  so they have they are free how how much time they   use they the session should be at least 45  minutes and then we have some room for q a so   if you have questions you are live in the event  please ask your questions in the chat and a uh   manfred and i uh will answer them and we have  some other uh helping hands here and uh if   questions are still open we will ask them directly  to the speakers so they can answer them yeah so   we have two small issues we are using teams live  event you are successfully locked in so it works   but we have a delay in this live event so we are  talking here and you will see the content about   30 seconds later so if you write down a question  maybe it will take um about one or one and an half   a minute till we see your question in the q and  a area we already have a question from jan he's   asking if there's interactive chat yes you have  found it jan and he asks if yes where it's here   and what's not really there is the interactivity  because we always have to publish a question   we can publish it or we can answer it and this is  the reason why you create a question you send it   we will see it and it takes a while if you get  an answer directly because we take the question   here to the speakers or we answer it here carson  and myself or we will write down an answer if we   are not speaking actually and we find time to to  go into a question so it should be interactively   absolutely um but please keep in mind that there's  a small delay yeah and we will have some questions   for the presenter anyway because uh manfred  and i we have also questions so we will see   um then all sessions are recorded they will  be available later of course i will inform you   i hope most of you have ticked when you um when  you registered for the event that you get onto the   newsletter and if you're on the newsletter i can  send you a mail when the sessions are available   and they will also be available on my youtube  channel i'm not quite clear now if we do the   whole event or if i cut them up in in smaller  pieces hourly sessions and press your thumbs   this is a very long uh team's live event we start  now and it it takes up to up to midnight so that's   11 and a half hours never done such a long uh live  event right me neither so i hope we will have no   problems with uh with some technical issues you  never know we are dependent on the internet and   um in my in my area today the internet is very  scary scarce because today is a is a day of where   the provider is doing some work okay um then  at the last session the closing that we do uh   tomorrow evening i will have a raffle and um  we will raffle i asked the sponsors about uh   about prices but then i i thought  how do we get the prices to the spawn   to the winners and if you win from outside of  germany maybe in the us or wherever you are   it could it could be very hard to send them over  to your place so i decided let's do amazon gift   cards and we do amazon gift cards when you win the  ruffle of course i get your mail and then i send   you a gift card and you can you can use it for  whatever you want and we will have another prize   i do also an azure stack hci course a five-day  course i've also a storage basis direct course   and a hyper-v course but we have the azure stack  hdi days so i will give away a free participation   in my course and if you if you are only an english  speaker not a german one we will find a solution   for that usually i do my courses in german but  maybe we find some more people who are interested   and then i will do a five-day english course  and you can participate in the course over teams   or live in the nice town i live in hallenberg  and manfred was also there it's a nice town it's   a small town but it is nice yeah okay let's switch  over to our speakers and if you have questions for   this we will also answer the questions so we have  an amazing line of a lineup of speakers i think   i have four slides of speakers there are 30s 30  speakers different speakers actually they are now   29 because one of our speakers got ill he informed  me on i think on monday and i had to shift the   the agenda a little bit so um i'm the next  speaker today usually it would be jan torre   from norway but he is not up to presenting  because he is a little bit ill so let's   talk about the different speakers you see here  fabian uh bosca is um from the company fujitsu   i hope that's correct for hitsu manfred is that  correctly pronounced i hope so yes he will present   today um about the great solutions fujitsu has for  azure stack hci and they are especially talking   about stretch cluster scenarios the title of this  presentation is i have it here deep dive building   stretch clusters with prime plugs for a microsoft  azure stack hci then we have two presenters from   dell another sponsor michael wells and lisa clarke  and both of them are also microsoft azure mvps so   as manfred and i we are also mvps and their  session will be turning hybrid cloud dreams into   reality with their technical technologies  and i like i like uh turning hybrid cloud   dreams so i like the title very much i'm looking  forward to that session then we have rob hindman   rob is from uh redmond and i know rob and  manfred also for uh quite some years he is   also responsible for the cloud and data center  management mvp program where where i and manfred   is in and rob is our mvp contact yes let's say and  he's presenting i have let's have a look i think   tomorrow tomorrow yes tomorrow um the last  one rebort in seconds with current soft reboot   yes with christina collette we have we have  her later oh she's she's here on the slide   christina coletti she's also a program manager uh  from redmond um they will talk about uh rebooting   in or reboot in seconds with a kernel soft  reboot i'm looking forward to that because it   it will shorten the reboot time of an azure stack  hdi cluster a lot right absolutely sorry for   interrupting you carson we have one message in the  chat if there's already audio or not um i think   because of the answer of yan we can be very sure  that there is audio because he heard our message   that there is a chat but we have to publish  the questions and so on and he replied to this   uh information we gave via audio but maybe you  can send some um yeah q and a messages to us if   audio is okay for you if you can hear everything  if you can if you can see everything um because if   there are any issues we can use this first minutes  to optimize it till the first session starts but   i think this it was an anonymous so i don't know  who is behind this message who has issues with   the audio i assume this is an individual problem  on this specific machine but it would be perfect   if some of you could give us feedback so we  can ensure that there's everything yeah and uh   jan also asked or said it would be perfect if  the attendees could chat to each other that's   unfortunately not possible with microsoft teams  team live events that's only only with teams   right yes yes yes and then we wouldn't see all the  questions because usually in the chat is so much   going on you don't see the questions anymore right  okay then andrew hansen has a session uh with   tina wu she is on on another slide he's also a  program manager in redmond and for um yeah storage   and files is a storage and file system team and we  will have here in that session about refs and the   syn provisioning for storage spaces direct then we  have roy casserby i think roy is where's roy he's   uh he's one of the presenters about secure core  servers and david shot both also live from redmond   is presenting about um yeah sdn um software  defined storage let's go through a bit quicker   because we have only uh 50 minutes to the first  session so then we have alvin morales alvin george   prasit and zakib and payman are also um and trung  and mike are also live presenting from redmond or   from the microsoft product groups alvin is doing  with other presenters gpus for high available   availability or available vms i'm looking  forward to that session a lot george is george is   i think uh doing an sdn session uh prasit is  doing is in two sessions one about gpus and uh   about secure core servers uh zakib is um also in  the secure uh course session payment is in the   gpu session jan torreira unfortunately i i didn't  want to remove him from the speaker list because   he was speaker until monday he got ill so i hope  you get better soon yantore we miss you here uh   trung is um doing a session um let's have a look  he's his co-presenter with truang where's trunk trunk is network atc network atc  is doing with dan uh on the next so   um we will see how you can automatically configure  your network in azure stack hci tina we already   mentioned she is doing uh this session uh about  thin provisioning in storage basis direct with   andrew hansen thomas morrow he's a former mvp we  know him very well he's from switzerland so he is   i think the only microsoft presenter who is not  from redmond and thomas is talking about arc   the the new thing for measure where  you can manage everything right   um jagomir casper jagomir is an xm microsoft  employee he is very well known for ws lab   and now it's called ms lab and we will show us how  to deploy azure stack hdi on hardware i think with   uh ms lab dave is an mvp fellow from canada and  he will talk about defender in azure stack hci   another mvp a friend of us from belgium he will  talk about smb and quick a new technology where   we don't use tcp ip for smb and other protocols  but the new quick standard and that's now new   in windows server 2022 the azure edition right  yeah it's in the azure edition yeah let's see   manfred you already see in the picture manfred  also mvp in the cloud and data center management   group manfred will talk about how we deploy uh  azure stack hdi with windows admin center and   his session is after mine at two o'clock i'm  presenting about uh stretch cluster right   at one o'clock where a young jan torres session  was and then we have helmut another mvp from   austria and he will also talk talk about from smb  to data center the scalability of azure stack hci   and now we are on the on the i think the  fourth speaker slide and we have udo vibra udo   is also known to you if you were a participant  in the last azure stack hci day in november udo   is from lenovo and he will do the lenovo session  let me see making azure stack hci solutions easy   lenovo think agilemx and then we have  cosmos cosmos darwin of course cosmos   will present about the azure stack hci roadmap  today at five and i'm really looking forward   to that session what cosmos will tell us about  the the future features or the future roadmap   uh of the product uh we are talking about hall  two days and jason jason also program manager   from uh redmond he will talk about the new  vm fleet i am a huge vm fleet fan and uh i do   all my installations or also petra another  employee in my company we do all our azure stack   hdi and storage basis direct installation we test  with vm fleet and there is a new vm fleet coming   um very interesting and he will show us all about  it and then we have dan cuomo dennis uh from the   network team in redmond and he will present with  truck trung trung about azure atc so automate   automatically deploying your network and then we  have matt mcspirit um he will tell us a lot about   azure kubernetes services uh with a colleague of  of a colleague of him has another session mike   we had him on the first slide mike  kosternitz he will tell us about the archit architecture of azure kubernetes  service and then we have jeff woolsey last   but not least and jeff is also well known  as a speaker in the community and he will talk about the new things in windows server 2022   and that was the speaker list it's it's quite  a long list yeah so we have 10 minutes left   here you see the agenda it's live on the azure  stack hci day site so you see here we are now   in the first session it's not really a session  it's a welcome to all of you and we will start at   one with my stretch cluster session then manfred  then dda because uh then we have our dell session   or the sponsor session and then it's eight in  the morning in redmond and the redmond speakers   will join us so starting with um cosmos um about  the roadmap and then we have dave from canada   windows defender advanced red protection then  we have the networking session today we have the   sponsor session from fujitsu where we learn a lot  about the primeflex servers i hope we will also   learn a bit about the new server generations  that are coming um also at the dell session   and then we have sim provisioning for storage  spaces direct and rfs improvements um and then   something i i think many of  you are looking forward to   to gpus for high available vms um they will  talk about what's coming for gpu support   in azure stack hdi 21 h2 and 22h2 and then we  close the day with a secure core server session   and then it's nearly midnight we will close up  and tomorrow we will start at one o'clock with   jaromir he will show us how to deploy azure  stack hci21h2 with ms lab and then we have helmut   which is from smb to data center session then  we learn about azure arc and what we can do on   premises i hope uh from thomas mauer then we have  udo with the lenovo session make azure stack hdi   solution easy and then we have a round table and  i'm looking really forward to the round table   where of course manfred and i didier jeff woolsey  helmut otto jagomir kasper and matt mcspirit are   here to answer your questions and i hope we get  a lot of questions you can ask anything azure   stack hdi related and of course server 2022  uh about all the things and uh i hope that   we will have a very live discussion about  uh these things and then we have after   that we have jeff woolsey with a session about  windows server 2022 and then two sessions about   azure kubernetes services something  i really look forward to because i'm   i'm to be honest i i'm playing around with  azure kubernetes services on premises but   there's still something missing in my head so  i hope i learn a lot here and then we have uh   what's new in software-defined networking session  then the vm fleet session with json and then   the soft kernel reboot reboot in seconds not in  minutes in seconds and then we have the closing uh   with the raffle and then everybody can sleep  and i think we both are very done then right   i assume so yeah yeah i'm really looking forward  to this reboot uh session because the reboot will   faster than the duration of the delay we have  here but yeah yeah i hope and do you think it's   under 30 seconds i i don't think so i assume with  vmcs but not with hardware we will see we'll see   yes okay and then of course i have to thank our  sponsors because without the sponsors you know   um these free events are not really possible so  a big thank uh thanks to dell technology who is   supporting the azure stack hdi days puitsu and  lenovo and of course a big thank you to manfred   that we can do that here with his nice equipment  and i think so far is everything okay huh or do we   have yes and we had a lot of feedback that audio  and video is fine so many wrote that this is fine   and so my my information is always that and maybe  it's an individual issue of the audio settings at   the end point because teams works in the cloud and  so if one person can receive our audio and video   then everybody should and we have a few questions  already in the chat and one of this question is   very strategic so i'm not sure if it was desired  for us to answer it if or if somebody wanted to   ask the audience maybe this anonymous can add  this information because the question is do you   see azure stack hdi as a virtualization platform  a cloud platform ignoring azure arc or both   so if the intention was to ask the audience please  resend this question with the information then   we can publish it and we can try to get some  information from the audience if this question   was for us i think it's perfect to discuss it  tomorrow in the round table so we can take this uh   uh one because now we have only a few minutes left  so we should take it till tomorrow and then we i   have seen a question already about gpu support and  my recommendation would be we have later a session   so we can put this gpu question to the session  where we talk about gpu later and we have um one   question about uh where can i watch que uh  sessions i miss uh and they mention carson   everything is recorded and it will be published in  the next days so you cannot watch it immediately   but someone later i think next week yeah i hope  to to get it up next week uh and uh the session   could only can only be record only recorded we  do it with teams live event so we will get a   recording when teams has not not a problem so if  there are sessions that are very important to you   watch them live if it's possible for you i  know the time the time span is very large from   midday in europe to midnight but if  it's important to you and you want to   get your live question questions in watch them  live usually there are no problems with the   recording with team live events but if there are  and we wouldn't have a recording that would be   terrible so life is better of course than watching  the recordings and this is the reason why we   only have one track so we have two days with  one track instead of one day with two tracks   so you can really watch every session and maybe  there's something not for you so you have to   to to eat something and so on so uh remember  the sessions start at always the full hour   and if there should be any technical issues if  something doesn't work or the screen freezes or   something like this then stay on the call we will  come back we have a lot of backup equipment here   so um stay on the session um we will do the best  and then you can receive the information yeah okay you can also uh questions in german yes and  i we are we are capable of german very well   yes i hope our english is okay i will not talk  german here because if i switch to german in my   head i have problems to do english again and we  can translate this uh questions i did this with   some of the all your questions before yes yeah   so we have two minutes to go then i will start  my stat cluster notes from the field session   and if you have questions for that very welcome  uh for me it's a little bit small to read   yeah i will you have you have to ask yeah you  have to ask the questions um to me so i think   this start went well so we have nearly done the  first half hour so only 11 hours to go today huh yeah and it seems for the attendees everything is  fine and again to mention this if you want to have   something discussed uh in the um in the audience  then we can publish the question i didn't publish   any of the questions now because this was feedback  about audio and questions we will take for the   later sessions but if you want to discuss this in  the audience add the information that it would be   great if we can publish it maybe we cannot publish  every question but we will do our best and as soon   as i publish the question everybody sees it and  then you can interact with each other so manfred   i have a technical question do i switch no i use  this microphone right yes you use this microphone   because you are on my camera here so um yeah you  you only you only lean back and start with stretch   cluster notes from the field so i have a question  for the audience do you like to have the speaker   at the side of the slides or do you want  to have the full view of the presentation because i have to start my presentation now  so question is if we do demos of course yeah   yeah i will switch to the full screen  and if the feedback is that full screen   of the slides and presentation is always  better then i will stay on the uh on the   full screen layout of the presentation so  we can i will take this feedback and um   react based on the feedback yeah okay so it's uh  1 p.m in germany so i will start with my session   good morning to everyone out there first  something about me you are in the session   stretch cluster notes from the field my name  is carson rachfar i'm one of the cloud and   data center management mvps in fact for 11 years  now i got my 11th award some days ago and i had   i hadn't even the time to tweet and blog about it  and i'm also an azure mvp now in the third year   i like to to be in both categories i always  has cloud and data center management is the old   on-premise world and azure is a new microsoft  world and i'm between that a bit more on-premise   still than in azure but uh both and then i'm also  being beam one guard so why i'm doing this session   um in the azure stack hdi implementations  i'm doing uh the stretch cluster is really   the feature that most of the customers want uh and  if they don't don't do storage spaces direct they   and do azure stack hdi it's mainly because of  the stretch cluster feature because in germany   and i think it's the same for for austria and  switzerland we use a lot of stretched clusters   maybe if you look at the amount of people  who live there most stretched clusters   in how you call that uh per head of people are  maybe in in germany so why use a stretch cluster   uh what are the reasons for stretch cluster  because it's more complex of course than a normal   cluster or even standalone nodes so the data  in the vms is today very valuable and you can't   lose hours of work or even minutes of works in  some environments and there are a lot of those so   um we have companies today who have um have  a 24 by seven so they work the whole 24 hours   seven days so including saturday and  sunday and even some of them 365 days   a year i have i'm actually actually at a  customer who has this requirement uh who   where the it has really to work 365 days  and you need really full availability   and a stretch cluster or um yeah a  stretch cluster can help with that   so we want to protect protect ourselves against  local disasters so a stretch cluster could be   if we think about let's say a natural catastrophe  like like a volcano uh breaks out like we have now   in las palmas or you have a a thunderstorm or a  hurricane you have earthquakes and so on so you   have maybe a stretch cluster that is uh whether  where the two sides are are large there is a large   distance between both but the stretch clusters  also if you have two rooms uh with one wall   and one of the room you want to prepare that this  room has no power or can burn or whatever so a   stretch cluster the distance can be very small  two rooms uh directly connected the rooms are   on the same compass maybe 100 meter 500 meter but  it can also be a very long distance and you want   to you want to protect yourself against the the  total um not availability of one of these sites   so if if one side is hit vms starts on  the other side that's very important   it's not something where you have constantly  replication of the vms and the memory of the vms   it's a replication of the data so if a vm runs in  site a the data is also replicated to site b and   if you have a disaster and say site a the vms can  be started inside b so you have a slight outage   of your application but it will be there maybe  after some minutes and that on the other side   so there are some requirements for a stretch  cluster we have to have at a minimum four nodes   two on each side so the the smallest microsoft  azure stack hdi stretch cluster has to have   four nodes not two four nodes two on each side we  have to have a network between the sites that can   constantly replicate the change that churn so if  we have a long distance stretch cluster we will   see on the next slide an example from microsoft  where they have one site in london and one side   in paris that's very far away uh then you have to  have a connection between the two where you can   constantly replicate your churn and if if  you have vms on one side that do a lot of   rights you have to get them to the other side  so with a 100 megabit internet connection   i think you that will not help you you have  to have gigabit or even 10 gigabit and that's   that's something that is not so easy especially  in long distance stretch clusters on a campus   shouldn't be a problem to have 10 gigabit between  the two sides then what's very important is   the witness placement we have to have a third  side or azure i will talk about the witness place   placement a little bit later because this is  really important we have four nodes and four nodes   are equal number of nodes so the cluster must  have a witness and the witness can't be in the   first side or in the second side this is very  important and many stretch cluster projects um   don't have this cert site or don't have the right  connection to azure azure is also an option you   can choose and then uh azure stack hdi is an azure  product but you have to have it every host has to   be integrated in an active directory and i'm  not talking about azure active directory i'm   talking about the old active directory that can be  on-premise on-premises of course it can be hosted   in vms in the cloud but you the the notes have to  be in an active directory for example today for   the live migration moving one vm from one node to  another you need you need tickets from the active   directory and that that is a must today otherwise  you can't live migrate so here is an example of   microsoft microsoft always thinks big because  the azure data centers so when you when you   have your redundancy there are i think at least  300 miles apart so the london paris example is   is something microsoft thinks of the cluster is  stretched over a large distance but that that that   is not the requirement you can also have one room  london could be and the name of one of your sites   the room is called london and the other room on  the same campus can also be called paris but we   can also uh have a cluster stretch over  both cities both european cities yeah so   when we have this cluster we have vms running  in the london part and these two nodes   we have a storage pool where these vms live  so their data is spread over these two nodes   yeah and then when we create a replication a  replicated volume every every data that one of   those vms is written to the csv so the cluster  shared volume is replicated synchronously or   with such a distance asynchronously to the other  side and then we write it here in a volume that is   um it's offline it's not accessible so it's uh  but when a disaster hits this volume will be   brought online so if we have an outage here  in london for example fire or power laws   these two nodes here because this is one cluster  so we have one cluster stretched over two sides   these two nodes will notice that the other  nodes are not available over the heartbeat   and then they will bring up this csv where the  data is and start the vms that are in these csv   yeah this looks a bit like an active passive  design that's one of the possibilities but   more often we have an active active design so we  have other vms running on this side in another   volume this volume can also be stretched  to this side and then if this side fails   the vms that we're running here will be  started here so it can be an active active   or an active passive construct so um normally  i will have i would have here an eight-minute   demo how to install an azure stack hdi stretch  cluster uh in the next session uh manfred will   show that in detail i will only show show you the  important part what's different with the stretch   cluster to a normal cluster that will manfred  show so um we what you see is windows admin   center on and in windows admin center this  is a the actual one the 2103 dot 2. um we can   add here under add i will do that very soon  we can add a cluster and i do that now add   then we have create new server cluster there are  other possibilities we have and then we have the   possibility to install a windows cluster or  an azure stack hdi and here you see we have   all servers on one side or servers in two sites  and that's the difference we have when you see   manfred's demo soon um let's stop here um there he  will not choose two sides he will uh install the   two nodes in one side so and then everything is  the same until minute six here after the cluster   creation yeah and then here in minute six when  we when we continue here after the cluster   creation because we choose we chose two sides  at the beginning we have to assign after the   cluster creation two names for the sites here  i choose east and west that are the two sides   and when then we have to assign the nodes to the  individual sites so two nodes are added to east   and two nodes are added to west and then in the  storage part that you that you will see in an in   manfred's section it will create a storage pool  inside a and a storage pool inside b and that's   the only time as far as i know where we can  have more than one storage pool in a azure stack   hdi cluster and storage basis direct does not  support two pools so i go back to my presentation and continue with the possible designs we talked a  little bit about already so we have two possible   designs we have an active passive design so we  have a stretch cluster over two sides we have   side one inside two and two servers minimum inside  one two servers inside two and we have all our   active volumes our stretched volumes inside one  and they are replicated with storage replication   two sides two but here they are offline and  we have only vms running inside one so these   blue and red uh it's not really blue it's purple  also they are running in side one and site two is   passive so it's it's just waiting  that side one fails and then it will   bring up the volumes automatically and also  start the vms but most of the time to be honest   i have never installed at a customer site  an active passive design they always so far   so far they always have active active designs  so in an active active design we have one   um stretched volume here that is replicated to  from side one to side two and we have one active   volume here from site two replicating to site  one and of course in one volumes are actively vms   running and on the other hosts are also actively  vms running and now each side is uh watching over   the other side if there is a failure then the  volume from the side that fails will be brought   online on the other side and the vms are started  and of course you can have multiple volumes you   don't have have to have one volume on one side  and one volume on the other side there can be   multiple stretched volumes on each side and we can  have also the possibility in a in a stretched uh   azure stack hdi cluster you will see that in  the demo later to create a volume that is only   presented on one side it it it must not it it  must not be stretched you can also have volumes   that are only presented on one side for example  if you have applications that have um redundancy   built in the application like an exchange d80 so  in a dag or a and sql act um sql how it's called   always on cluster they have already redundancy  in the application so you you put the the vms in   a volume that is only represented on one side  and another way i'm in a volume that is only   represented on that side and then you use the  in application replication so um stretched azure   stack hdi possible nodes in two sides we already  learned we can have four nodes so two nodes in   each side here in this picture we have a site in  frankfurt and the site in hana that are two german   cities that are not so far away let's say maybe 20  kilometers or 15 kilometers so we can do that and   of course you can have a six node cluster three  nodes in each side and you can have an eight node   cluster four node in each side and you can have  ten node twelve node fourteen nodes sixty node   and then five nodes six nodes seven eight aimnode  eighth note in each side what you can't do is a   two node cluster stretched or three node cluster  switch or having an odd number of nodes so   you can't do a seven node cluster three in one  side and four in the other side they are always   the same number of nodes in both sides and if you  look at this we have also always an even number   so we have to have a witness because we need a tie  breaker for our cluster design and for that that's   our next topic the witness design so we have again  our picture from before we have an active active   cluster we have two nodes here and two notes  here and we need a tie breaker so now let's   have a witness the green one here in site one  if we place our witness inside one we have now   five um five votes yeah and if now something  happens to our site two we have now uh   three votes from five votes so in total it  would be five but two are gone so we have   three from five that is more than half of the  votes are available so what will happen the vms   the volume will be brought up the replicated  volume will be brought up inside one the volume   from side two and the vms are started here  everything is fine so let's do the other side   site one fails and of course then our witness  is also gone so we have two nodes running inside   two they are perfectly fine but they have  only two votes now from five so there are   below half the votes so and they are in  the minority they are not at the mirror   majority they are in the minority so what do  they do they shut down the workloads they have and we have no vms running but you you build a  stretch cluster to have all the vms running in a   disaster not none running because the cluster is  in the minority it will shut down the workloads   so how do we do that correctly we have to place  our witness in a third side so and what's very   important each side can communicate with the  witness without the other side so we have to   have a direct connection from side two to the  witness and also from side one to the witness   not going through side one to the witness so if  side two can only reach the witness to through   side one because all the internet connections are  inside one that's not a not a design you can't   build a stretch cluster because if site one goes  down remember the the slide before your witness   is also not reachable it's running in your third  site or in azure but you can't communicate with   it and that's like down so the cluster will  go down so we have to have the witness in a   separate site and each side can communicate with  it now site two fails and what will happen the   cluster will bring online the vms inside one and  the other way around is the same we have three   from five uh votes so still the majority and  our vms will be brought up on the other side so what witness options do we have with  an stretched azure stack hci cluster   or even with an azure stack hdi cluster we have  the same witness options with as we had in storage   spaces direct we can we can choose a file share  witness where we have a share somewhere i think   at least an smb2 share not smb1 s b2share or we  can have the cloud witness we have another picture   cloud witness is an offering in azure you have a  blob that is your cloud width is very cheap and   i i heard from a lot of customers in the past  that that have done or that are doing storage   spaces direct no no cloud witness we don't want  to be connected to the cloud with azure stack   hdi you have to have you have to be connected to  the cloud because azure stack hdi has at least um   inform azure every 30 days how much cores are  running because azure stack hdi is built through   azure so you have to have a connection to the  cloud and then you can also use a cloud witness   if you don't have um a third site handy with  a file share witness of course you can use a   cloud witness and it usually it's much cheaper  than to build a third site just for the witness   and what's not supported in azure stack hdi  and storage basis direct even if they are   not stretched is a disk witness so in azure stack  hdi stretched or not stretched and storage space   direct only the file show witness and the cloud  witness are supported not a disk witness so let's   do a small demo how we install the cloud witness  where is my there it is so i go to an another node oh no i do it here the other already have a  witness so if we go to settings here i'm now in   an azure stack hci cluster it's installed um yes we talk questions at the end that's a very  good we we talked about your questions at the end   so um i have already installed it it's registered  in azure and now i go to the settings and manfred   will show you how you register your azure stack  hdi in the cloud right okay so uh if we go here in   the settings of the azure stack hci cluster and we  go to witness you have the possibilities we have   no witness or we could choose a cloud witness or a  file sharewitness i will add a file share witness   and then we have to give it a share and here i  have a share we see there are already two clusters   registers under this uh directory under this share  and i will do now the third so i go back here   and post paste it in and click on  save and now we have our witness   set and the witness of course has to be on a  place third side or in the cloud so now our   cluster has a witness so if something  fails it has the possibility to have   a quorum we call it a quorum cluster as a  quorum he has more votes than the other side   and here you see it's all all so registered in  azure and manfred will show that in the next   session so let's go back to my slides this this  was a very small demo so um for the replication   for the stretched volumes microsoft user feature  that is available in windows with windows server   2016 it's called storage replica so you you can  use storage replica if you have a data center windows server 2016 2019 2022 but it's included  of course in azure stack hci the new operating   system and it's a feature for disaster  recovery so prevent you from disasters   so here we see where the storage replication is  embedded in the io stream to the disk or the i o   pass from an application to a disk device and  it's agnostic to to file system filters to the   file system we use to uh snapshots bitlocker  everything in fact it's between a vault the   volume manager and the petition manager so uh  in fact we replicate partitions not volumes   and here every io to the disk is replicated  to the destination site and then also written   to a disk so it's very easy it's block based  everything that is written to the volume is then   transmitted over smb3 to the other side there are  three scenarios we can do with storage replica   there is a server-to-server scenario so if you  have an hyper-v server with local volume you can   replicate it to another hyper-v server the volume  would be offline so if the a disaster strikes here   you have to bring manually this volume online  and register all the vms that are in the volume   because this is not a cluster server 2 doesn't  know about your vms in that volume it doesn't even   know the volume so this is the manual task we have  another scenario where we have two clusters using   storage replica and it's the same we have one  cluster here one cluster here if this volume fails   you have manually to bring up the volume  and register the vms in this cluster because   this is the second cluster and this is the  first this cluster doesn't know the roles   only this scenario and by the way this you can  do with storage spaces direct with two separate   storage bases direct clusters but what we want  is a stretch cluster where we have one cluster um   including both sides and we have our volumes that  are replicated so if this site fails the cluster   already knows this volume and knows that he can  bring it up and he will bring it up automatically   and he also also knows all the virtual machines  because they are this is the same cluster he knows   every node knows every role and the status of  every row so he brings up the volume and he   starts the roles automatically so how does it work  the synchronous replication we have two types of   replication we have the synchronous replication  we have an application for example our vm that   is writing a block to our source node and in  azure stack hci we have a source node and the   destination node that is replicating the volume  so it the application the vm writes a block to   the source node and the source node is writing the  block into a separate log volume i see also a csv   and in the same time transporting the data  to the destination node over smb3 so smp3   is used as a transport and then it writes on the  destination side also in a lock volume and then it   acknowledged the right the the successful write of  the data in an in a non-volatile volume so the log   volume is also on on flash and if you write it to  the log volume it's even if you have a power out   of the server the data is there then the data  is acknowledged and then the application gets   acknowledgement so now um the data is written  and of course how long it takes from the right   uh until the acknowledgement depends on how far  away are your both the both servers if they are   next room it's very fast if there are 300  kilometers between the sides you have maybe 20 30   milliseconds of delay from your acknowledgement  and that's too much for a lot of applications   after the acknowledgement is done the data is  copied to the original destination so as you   see here we have four volumes for a stretched data  volume so four volumes instead of one for stretch   dwelling and we have some requirements the latency  between the site should be below five milliseconds   and the lock volumes have to be on fl  flash storage if we don't um if we can't   promise these requirements if the delay is longer  than five milliseconds or you have also the   possibility to use asynchronous replication how  does that work we have our application it writes   a block the block is written to the log and the  the write is acknowledged and then the data is   transferred to the destination node written to  the log acknowledged and then it is written to   the data but we don't wait on the other side so  the uh synchronous replication is immediately you   don't have the delay but you have also your data  on the other side it's not there immediately but   it's maybe a delay of two seconds or 30 seconds  or whatever so it's a continuous replication   but you don't have the guarantee that every  right is on the other side but it's also not   bad because if you have a disaster on the source  side you maybe are missing let's say 30 seconds   of your data that's much better than restoring  a backup from the night that that includes maybe   does not include six eight hours of data you  have done you have already changed in the vm   okay so let's create a stretched volume do a demo  where we create a stretched volume so we go to uh   oh we can do it here we can go to volume in the  windows admin center and you see we have already   four total uh four volumes and these volumes are  the very important cluster performance history so   if you create a stretch cluster where is my mouse  here yeah it's very small so if you create if you   create an azure stack hdi cluster a stretched one  you get your cluster performance history already   stretched because it's important and we don't want  to miss all our data from all vms from all data   from all this from all volumes so microsoft  creates a stretched volume and here you see the   four four csvs we get with that so now we create  a new stretched volume click on create and here we it will gather some informations and we have now  the possibility to create a volume in a single   site for my exchange dag or my sql always on  cluster no we want to create a replica replicated   volume across two sites and then we can specify  where where is it online invest or east so if it's   in west it will create it will be created invest  and replicate it to east or the other way around   so maybe if you create multiple volumes you spread  them over the sides for an active active design   we just do it from east to west because east  is first here and then we can choose which   replication mode do we want an asynchronous or  asynchronous and of course you can change that   later so you can change from synchronous to  asynchronous or from asynchronous to synchronous   after the fact so it's not it's you can  do that later so i will call it stretch   volume and now we can choose the resiliency and  this is a four node cluster so we have only the   possibility to do a two-way mirror in each site  if you want to to to do something like a nested   mirror or a nested nested resiliency  we we have to do it in powershell um   but for the demo here we do a two-way mirror  and i say it's a 400 not terabyte it's a 400   giga gigabyte volume so then we have some advanced  options here and you see our volume is called   stretch volume so it will create replication  groups from the name with group and replication   group and a lock volume size and best practice it  turned out at least do 100 gig or better 200 gig   for your real world uh um stretched volume so the  the proposed 40 gigabits is much too less so or do   200 or even 500 gigabit if you have a 10 terabyte  volume and then we can do some other things here   but we can't do bitlocker and so on and now i  i go to create so i press create what will the   windows admin center do now it will  first create the stretched volume   in side east then it will create the stretched  volume log inside east then it will create the   stretched volume inside west and this is offline  and it will create a stretched volume um now it   will create a stretched volume replica in west  and it will create a stretched volume replica lock   in west and that will take a bit and it will also  install a running storage replica so we will come   back to this one and i will show you that in  a demo or i go to another another cluster that   is prepared for that and if you look here and  look on the volumes on the volumes we have eight here you see we have our cluster performance history  our stretched volume stretched volume lock   and storage volume replica in stretch volume  replica log and if we click on storage replica   we can look at the replication status of  the storage replica partnership and there   should be one partnership for the cluster  performance history and one partnership   for our stretched volume and it is continuous  replicating we can look into some details   here we see it's synchronous here it is  synchronous and we can if we want to can   go to settings and here we can modify the  partner settings so we could switches to   asynchronous or from asynchronous to synchronous  we can increase the lock size it's a 200 watt   gigabyte volume so i would advise to do more here  is it no it's i think here it's only only a 100   gig okay but i think you got the impression how  it works so let's go back to the presentation i've   still 10 minutes maybe i need a little bit more so  there are some power shell commandlets to help us   with uh the replication and to be honest in the  moment with windows admin center you can't delete   a replicated volume or at least you can't delete  it fully so you need some powershell command   let's for example to delete the partnership and i  hope it will be added in the windows admin center   module but in the moment you need some  powershell and there are plenty of powershell   commandlets we have for example for  the storage replica we have a test   sr topology where we can first test if  a replication would work especially in a   stretch cluster that is is a far distance stretch  cluster we can look at the replica status there   are replication groups you can with get storage  replica groups you can look at the status and   see them and there are the partnerships so you can  also get get as our partnership and there is a set   and the new and the remove of course then if your  synchronization stops for what reason or ever for   example your internet went down for two days then  of course your uh synchronization stop because   the logs are full and you maybe want to manually  resync it there is a powershell command for that   then a very important thing you have to specify  which network cards are used for the replications   microsoft has some very strict strict requirements  for the storage replication networks and you have   to for your storage replication groups you have to  specify which networks are used um very important   then we can delegate admin rights so you  don't have to be domain administrator if you   want to care about the storage replication  there are possibilities to grant rights and   revoke rights and we can limit the bandwidth how  many bandwidths the storage replication can use   bytes in bytes per second um usually we have to  have a very fast replication network where you   where you that you don't share with other stuff so  you don't have to limit the replication bandwidth   you have to guarantee that your replication can  always happen otherwise you have some problems   so let's let's move the vm and the volume to show  you that was a question when i first introduced a stretch cluster before it was available we we  knew last year that the stretch cluster scenario   will be available and people ask me in webinars  um can you move a vm around without a fail   yes you can do that and i want to show you that  so i go back to the demo here in this cluster we   have a running replication and i have deployed  one small benchmark vm into into this stretched   volume so if we look in the hyper-v settings  here and you see we have it in stretched in   the stretched volume there are our discs yeah in  c cluster storage stretched volume small benchmark   and now i saw i i missed something i didn't  include the vm in the cluster i will add that so i have to connect to the cluster i'm on two   it's the cluster is called  azhci to cluster so this is live and i open the cluster the good thing is all the  old tooling is still working and if i go to roles   there is nothing and i will configure a role just a virtual machine the cluster will look on the hypervisors which  roles are not included in the cluster yet   so i add the small benchmark vm  and then we have it in the cluster so here we are it's running on the second  on the first node you see that here and our   stretch cluster going to windows admin center  to show you clicked wrong to show you the   servers and under servers we should see under  inventory where each node is so we have an east   and the west side now we can't see that here right  so we go to windows admin center under nodes here   are the sides so the first node and the second  node are in east and the third and the fourth   node are in west so our our vm is running on node  1 in east in the volume called stretched volume and start stretched volume is also the owner  is the node 1 where also the vm is running so   now i want to move the vm to the other  side so let's do a live migration here move live migrate and select so if  i choose best possible node it will   move it automatically to node 2  because it's in the same site and if i   then again live market and let him choose  where to move it will move it back to one so   the vm will stay in the same side where the  storage is and if you if you let the vm move it to   the other side that's now the case i have now  running the vm on in the other side um it will   the cluster will move it back to the site where  the storage is so the vm is following the storage   after a while it will move it back automatically  so now we have the vm running in site west and the   storage is inside east and we can see that if we  look at storage replica here it's much easier in   i like the windows admin center and i always  use windows admin center unless i don't have   the feature i need like moving a vm on purpose  you can't do that in windows admin center in a   cluster in the moment so here you see we have  our replication from note 1 to node 3. and now   i click here let's see what what happened with the  vm it's now running on the other side and i was   i say here switch directions switch directions  and maybe he asked me for something so here we   see the vm it's switching directions and  we will see two things the i o will halt   on a hardware cluster this is a virtual machine  cluster on a hardware cluster to for five to six   seconds then we have we'll have i o again here  it maybe takes longer because this is a nested   virtualization cluster then the io will will be  there again and then it will halt for another five   to six seconds here unfortunately it takes longer  but in real world i've done this multiple times   so you will have an i o hold for five to  six seconds and another one but your vm is   running here you see the another one but after  let's say 30 seconds your vm the storage is on   the other side we have you have two small  five to six seven i o pauses and then the   vm is running on the other side and  now it will rebuild the replication   yeah so uh replicating some blocks back to the  other side okay let's go on i i'm nearly done so   um with the time at least so now some  important networking requirements   and microsoft was a little bit late to to document  the network requirements for a stretch cluster   so in in the side so here we see an eighth  note stretch cluster four notes inside a's   four not inside b in the sides you can do smb3  over smb direct so with rdma for the storage   replication between not storage replications uh  excuse me for the software software bus layer so   the storage bus when you have your extends your  vm writes into extend and you have a three-way   mirror the the data must be written also to other  nodes for for that we use smb direct preferably   um switchless is also supported in the site so you  can connect the four nodes with a lot of cables   without switches so for the replication network  between the sides we need a fast low latency   network that's important low latency because  if you do synchronous replication and you have   a high latency network it adds to your io  because it adds the milliseconds to the i o   microsoft does not support rdma for the  replication traffic indeed microsoft wants to have   separate adapters for the replication traffic with  no other traffic on them i have some support cases   where where that is clarified no other traffic  than the replication traffic on those adapters   and only smb over tcp and we have to have layer  3 routing microsoft does not support a layer 2   ip network between the sides even if there is  only one small wall between the two sides and   you can perfectly have a layer 2 network microsoft  requires a routing instance so you have to have   routing in your switches or in your firewall for  the replication traffic it's not supported to have   a layer 2 network no layer 2 network is supported  i'm i'm talking to microsoft because in a small   stretch cluster with with small distances it  would be much easier to have a layer 2 ip network   instead of doing routing in switches and so on  and we have to add our constrain network uh so   the storage replica constraint network uh for the  replication group otherwise the network is going   over another net and it does not prefer the routed  networks usually it uses the management network   or so you want to have your replication traffic  on the separate networks that's very important   i will maybe do another webinar about that  because i'm doing in the moment i've done the   last months some stretch clusters what about dot  dot in stretch cluster azure kubernetes services   in the moment azure kubernetes services are  not supported in a stretch cluster scenario   and that's a huge miss in the moment because if  we do azure kubernetes services on azure stack   hdi people want to stretch them they want  to have high availability or even disaster   recovery possibilities for the containers  software defined networking doesn't   work it's not supported in a stretch clusters  enough it's only supported in a non-stretched   cluster scenario bitlocker it works so i have  a customer where i implemented bitlocker on the   on the stretched volumes but you have to do it by  hand so with powershell and some commands you have   to do on the host in in a remote desktop session  on the host it is not working um remotely where   you use invoke command or something integrity  checksums i don't know actually i have some   questions for the product group integrity checksum  you can't configure them for the data part of your   of your stretch volume so there's no um in  admin center there's no possibility for that   manfred will show you that you can do that  very nicely bitlocker and integrity chocsums   if you create a volume in a not stretched  scenario deduplication same i don't know if it's   supported at least you can't configure it with  windows admin center but in the moment i would   i would say don't use deduplication  so now disaster demo we skip   i will because we have only some minutes left and  other questions a lot of questions okay then we go   to the lot of questions uh helmut did a great job  he already answered a lot of them but maybe it's   interesting to take them or some of them on audio  yeah so ask ask away i have we have let's say six   seven minutes and then uh okay so let's see which  questions we have here where we will start so   there are some questions about wino's admin center  i will take them in the next session i think this   is a good idea so we don't forget them there are  comments about windows admin center we will take   them uh later so you can sure we have them here um  so there was a question when you were presenting   carson can we change the resiliency from two-way  to three-way mirror that's a great one and uh   thank you for that question and uh it's um so  the the technical answer or the the marketing 

2021-10-04

Show video