Running Azure On-Premises!

Video Statistics and Information

Video
Captions Word Cloud
Reddit Comments
Captions
hey everyone in this video i want to answer the question how do i have azure on premises um and before we can answer that question we really have to understand well what exactly are you asking because there are many different aspects to azure but first as always if this is useful a like subscribe comment and share would be appreciated so i want azure on premises so what is azure so we think about well we have azure this cloud service and i can think firstly azure is capacity now that capacity could be cpu network storage various types of things but fundamentally azure is offering capacity now the way azure offers that capacity remember is we have this great big microsoft backbone network and then what we have are regions so this region is defined as this two millisecond round trip latency envelope multiple physical facilities in that connected to that microsoft backbone and there are lots of these regions all throughout the world so i have multiple regions available to me now those regions offer a whole set of different capabilities which i'll get back to in a second now in a different in addition to what we consider kind of these major regions one of the things we start to hear about are well these azure edge zones and here i can think about these smaller facilities that it may be closer to certain metropolitan sort of cities etc connected to that same microsoft backbone network uh still microsoft owned but these are these azure edge zones that are going to offer a subset of services and then you'll hear about well there are these edge zones with carrier so here on the carrier's network that carrier network does connect to the microsoft backbone as well but here in the carrier facilities they're offering these kind of microsoft services and the benefit of these edge zones with the carriers is that carrier network is kind of one hop away from things like the 5g networks if i was operating some kind of 5g network in order to super low latency to certain azure services i could use that capacity in those edge zones with the carriers to actually provide that now a few times i've said services so this is capacity this is places where there's some capacity to run stuff now the stuff i'm actually running that capacity is exposed as services now when we think about kind of the azure regions well there's a whole bunch of different resource or resource providers and we think about very common ones like well there's compute there's network there's storage there's a whole bunch of these that offer essentially a huge range of different services like compute we can think about vms vm scale sets disks here we can have virtual networks here we can have kind of storage accounts huge range of these now if i actually go and look in the azure portal we can actually take a quick look at this so if i actually just go and search here in the portal and i just type in resource explorer by looking at that resource explorer it's going to show me all of the different resource providers and there are a huge number of them and then as i kind of alluded to well i could look at something like compute and then here we can see on the compute where's availability sets cloud services disks galleries and if i keep scrolling down i'll see virtual machines here virtual machine scale sets we have all these different types of service provided through those resource providers now those main azure regions i alluded to well there's just a huge number of those all throughout the world so we have those capabilities that offer most of those different types of resources now what we're also now seeing are these kind of edge zones this is preview but here we can see certain metro areas that are coming online for edge zones offered by microsoft and you're going to see edge zones with carriers for example at t right now are offering these in kind of atlanta dallas and los angeles so if i think about well services hey i have the full set in these azure regions now when i start looking at things like well the edge zones it's really a subset i'm typically going to want these built around kind of iot maybe containers on kubernetes and kind of vms so it's a much smaller subset but they're the sorts of things i would really want there now in addition when we think about all those different types of services a huge number of those i can almost think of as supporting type services i could think about yes we have all these different vms and vm scale sets and all of those great things but there's also great supporting services i could use for many different things things like azure site recovery for replication things like backup things like monitoring update management so there's a whole number of these that bring additional sets of capability that help those services actually running in there okay so i have azure can be thought of as capacity and that capacity is exposed as services which are available through these different means but to actually do things well there's a management plan so now i can think about okay we have this azure resource manager or arm this is the control plane of azure it is how i do all of my management now through that azure resource manager we do basic things like crud operations create read update delete i create things i update them all of that stuff on top of that we have metadata about all of those resources so of course we have things like tagging through those i can do things like search i have things like inventory i can do things like change tracking then i have things like monitoring and i alluded to that up there but that monitoring things like log analytics really builds into many other types of things where azure monitor we have kind of azure monitor logs which can then go and bead in build into services like azure sentinel so we take the logs we look at the logs we apply machine learning to that and we can deduce things we can spot certain types of things we have governance requirements so we have things like azure policy so i can both track compliance and enforce things that's both tracking and enforcement we have things like azure security center um i have things like update management as i kind of alluded to up there as well there's all these extensions that i can add for things like virtual machines custom script extensions anti-malware with defender and one of the other nice things i get in here is kind of this managed identity that actually i can use to do r back to other resources so i can kind of control those things and so if i think really what is azure well azure is this it's capacity with services with a control plane and so when you ask the question i want azure on premises which bit do you care about do i care about i want to extend the management the control plane of azure to my resources be they on premises i'll be there maybe other clouds or do i actually want to start getting into well there's certain azure services i want to be able to run on my equipment or i want to add in additional capacity in an azure consistent manner so i have to be able to answer that question to really identify what is the right solution for me but let's start kind of from the bottom going up let's say it's the management plane is what i care about now if it's the management plane i care about what the answers really going to boil down to to this is arc so we have azure arc and what arc is going to do there's different flavors of azure arc but fundamentally it's extending the control plane of that azure resource manager and making it available to other types of things now the most simple basic unit we can think about often is a virtual machine so i could think about hey over here i just have my vm now that that virtual machine this could actually be now i'm saying a vm i should actually change that i'm going to say instead an os instance because yes it can absolutely be a vm but it could also be bare metal it could be windows it could be linux uh it really doesn't care and again i'm drawing it in green to kind of hint that hey this is kind of this on-premises thing but it could also just as easily be in another cloud it could be in aws it could be in gcp it really doesn't matter cloud now the way this works is obviously i'm dealing with an os instance so what we have with arc is we have an agent so it's going to install on here the arc agent now this is the azure connected machine agent and again there's version for windows for linux there's ways to do a large scale deployment of this but once i get this arc agent deployed what it's basically now doing is it's going to go and check in with that azure resource manager this virtual machine now actually has kind of a little avatar a projection of itself into my azure subscription so i do stand up kind of an arc management plane in a particular subscription so this will now light up and all my multiple vms would now show inside there so now this is a known entity it's the agent goes and connects to azure so i'm not opening up firewall ports or anything like that and this is in ga today so the arc4 servers so again there's different flavors of this so this is kind of arc for servers now this connection it could be over the internet if i had express route it could be over express route and they're working on kind of actually doing that over a private endpoint as well now an important part this is not creating os instances today this is i have it there already be a vm or whatever i'm putting the agent on it then it lights up the os to arc today it is not creating vms on hyper-v or vsphere or anything like that it has to be there already and then it can actually take and do the management of it now what we get is well we get things like the monitoring through this agent i can now actually get the monitor and through monitoring i get things like well there's azure security center through those logs again now we can send things to things like sentinel i can get things like app insights even other types of insights like sql i get those capabilities we get policy now the policy is primarily in guest there are some types of management policy that would apply for example i could say hey i want it to have certain tags but primarily the policy is going to be in guest so i'm kind of thinking about well for windows that's going to be that desired state configuration set of capability and for linux it's going to be inspect but i mean dsc is available for linux as well so i wouldn't be surprised if that changes i can apply things like tags now i can search i can do an azure resource graph and i would find that thing i have that kind of change management change tracking and my inventory i have that available to me i have things like the update management so all these actually tons of capabilities one of the most interesting ones is i also get a managed identity for that virtual machine now through that managed identity i can then use that to go and talk to other services in azure like key vault well key vault could then store things like certificates and then the other thing i can do here is extensions so there are certain extensions for vms that apply to this and there's one of them about certificate update management so through the identity i can get permission to a key vault to get the certs to apply to that os instance so it's actually a very cool kind of thing i can bring all of this together so this is just my os instances bare metal vm other clouds doesn't matter i'm putting this agent on them and then through the agent we now light up all of these capabilities and again kind of some of those extensions will kind of shine through on there if i actually quickly jump over see if i go and look at my kind of environment here one of the things i can actually see is if i go home and i'll go to my arc there's my azure arc and we can see there's different types of arc we can see there's there's servers there's kubernetes there's data services there's azure stack hci but if i go to servers i have this is a windows server 2019 virtual machine registered and running the agent and we can start to see the things it has now it does have access control i didn't really stress the r back because honestly the r back is more about what i can do to the object in azure i can't do things like change who can log on to that os so i think that's less of a feature today but you'll see i can have extensions there are various extensions i can actually add things like that desired configuration and there's actually a complete list if i look at this list it actually shows me the available extensions so things like defender log analytics agent key vault then there's ones for linux so we have those extensions i can now apply actually to my arc known virtual machine to actually leverage that i can see i can apply policies and right now even though i've done nothing specific well it's going to have things like the regular policies i have applied to my subscription like azure security center um it's going to show me where i am i could light up things like update management inventory change tracking i can get my insights there are certain tasks i can use i can light up all these different things actually for this instance now in terms of the ark service if i think about well for things just like registering it and forgetting that tagging the searching that that's free now when i start to look at things like the change in and this kind of desired stake sets of capability that's where i start paying money also if i was doing defender defender is a separate kind of item there's kind of dollars for that as well so you need to go and check what bits of the functionality you want to use but this is pretty significant so if your question was hey i want azure on-prem and it was really based around have a whole bunch of vms i want to start doing some management i want to be able to run scripts like those extensions custom script extension is one of those if you want to be able to search them and tag them i can do all of that stuff my virtual machine in here i've got tags and there's a certain set of kind of common tags you use but you can have like the city the country data center any tag i want well i can add those tags to these resources to make them really fit in with everything i'm doing so arc for servers might be your answer and you can kind of stop that might be exactly what you want but maybe that's not exactly what you're doing let's kind of go over to the other side now maybe i'm a little bit more advanced maybe i'm getting into these more modern applications and maybe what i'm actually doing we're going to kind of change color here maybe i'm all on the container train and what i'm using is kubernetes so kubernetes is really kind of i think one out as that standard for the orchestrator then that kubernetes kind of management plane manages a bunch of worker nodes that actually run the containers themselves now as you would expect once again there is the ability to light this up for arc so this i'm drawing it here it's really a bunch of containers that we run in the environment but now what i'm getting is kind of arc kind of for that kubernetes for my container environment and once again this adds a whole bunch of really very nice features that are going to be key to how i think about using those container environments now once again it is not deploying kubernetes clusters the cluster has to be there already but also this is not just you might think about some aks deployment somewhere it's nothing like that as long as this kubernetes environment is cncf standard so that yes it's aks but it's open shift there's a whole bunch most of them around this it will be able to manage that kubernetes environment and what i can think about well once again we have kind of uh inventory which i'll do in this color i have my inventory i can do things like my monitoring so my azure monitor for containers will apply here i can do my standard tag search rbac all of that is going to apply i can have my governance so azure policy for kubernetes guess what that's going to work here as well things like the azure security center is going to work to understand those environments and maybe one of the biggest things this is going to help light up and manage is git ops skid ops is based around the idea of well we're going to have this flux component so that's going to get kind of deployed as well and what we have is we have a get repository so i have a git repository out here now in that git repository it doesn't have to be github doesn't have to be azure device repos it could be a bit bucket it does not matter my point is this git repo has a whole bunch of yaml files so remember yaml files are attractive because these are those declarative files uh i conversion control them they're observable i can see what's in it i can see what's changed and what i'm essentially going to do is via the get ops configure that cluster to say hey i want you to go and sync with this particular repo so now we'll actually go and look and if there's a change to that that repo it will pull down the updated yaml files and apply them now maybe it's referencing an image so at the same time i can have kind of a container registry which has got my images and so if i'm referencing an image in that deployment file that declarative configuration will it will also go and kind of pull down the image it needs to deploy those container those pods to my kubernetes environment and and so that's really what's coming together when i think about the arc iv kubernetes so it gives me the insight it's a single pane of glass and that's kind of key these environments here will show up kind of next to my aks environments in the azure portal so i'm really getting this great set of single pane of glass standard set of policy standard monitoring inventory it's really bringing them all together which is likely what i want so that was kind of the the arc for kubernetes side of things but if you think about it once i'm managing the kubernetes environment again this is all still the management plane really right now i'm still focused on this it is not doing lifestyle life cycle management today of kubernetes it is not upgrading the kubernetes environment today it's not doing that but now i am kind of been able to do configuration management of that kubernetes the last part of arc is arc for data services so now we're crossing the line we have until this point been focused on the control plane and when we think about that control plane the control plane really is what i interface with things like hey i can use the portal to interface with that i can interface that with rest apis i can interface with that through power shell through the cli in a very standard way and i can use that across all of these different things but now we're going to start going up into actual services now i want some of the services from azure on-premises well now because arc can manage this kubernetes environment today and again this is preview as well so this arc for kubernetes is in preview and also we get arc for data services so once again now through arc it can also deploy a sql managed instance and it can also deploy a kind of postgres sql hyperscale they're going to be evergreen so it's its job to take care of updating them to take care of hey the scaling i want to do so now we're actually kind of with this arc data services now we're actually bringing down services we can leverage as part of our business functionality up until this point art for servers art for kubernetes was around the configuration and management now i'm using that configuration management to now drive actually deploying things to that kubernetes environment it's managing to actually bring true data services so that's arc now you might look at this at this point be like oh that's what i wanted yeah great i've got vms or i've got containers that's exactly what i wanted to do i've already got the hardware i've got kubernetes i'm good stop talking john great stop the video you're good to go but what if you need a bit more what if that isn't kind of the whole answer there's actually hey i want some capacity or i need some additional services so the first service i'm going to think about i'm actually going to zoom out a little bit let's go over here we had azure stack now azure stack was really kind of its own thing um there was no other stacks around but it's been rebranded we now think of it as azure stack hub so i have an azure stack hub so i have azure stack hub now an azure stack hub is this kind of turnkey appliance that's delivered to you this could come from kind of a dell a lenovo it could be hpe there are many others but it's provided by a vendor and i probably pay them money to to get this stack now that stack has kind of multiple network racks then there's a kind of hardware life cycle blade it's all the cabling is native and then i get a bunch of nodes now the exact number varies but it's between kind of four to sixteen and these each have kind of cpu and they have storage so there's not a separate sand it's kind of using a hyper converged methodology behind the scenes but i get this azure stack hub so i get it's 4 to 16 nodes providing this capability now what does it actually give me now what it's giving me is always azure consistent api now the api version is going to lagger a little bit behind but it has its own kind of portals and there's both a user kind of the regular portal and an admin i have to create plans and offers that people can consume i can use rest i can use things like powershell i can use cli but it is a different endpoint i i am not going to portal.azure.com it is a local endpoint for that particular azure stack if i had multiple azure stacks just to be very clear each one of them is an island i can't kind of join them together they're each their own little island they each have their own endpoints and so if i had five azure stack hubs i'd have five different urls endpoints to go to to manage them i am doing the management so you manage once a month for example i might go into the admin portal pull down the latest update package hit apply and it will go and update the different nodes all the different parts of the azure stack to do that but this different local endpoint is kind of a big deal it is not being managed by azure at this point it is really not connected to azure it it can be and i'm going to talk about that but it is its own set of management endpoints now it's consistent where it has the functionality so it does have a number of resource providers there's kind of a core set and again we can think about from a core set perspective obviously we have things like compute so again we can have vms and vm scale sets um it actually uses the aks engine so the aks engine spits out a json file that then mixes with config and vms to actually deploy kind of a kubernetes environment it has certain networking so it's using software-defined networking it has certain storage capabilities it has key vault now key vault in azure is kind of the shared hardware it uses host security modules there is no hsm in it in an azure stack hub so this is software based but it's on your equipment it's in your data center probably not such a big deal then there are optional resource providers at this point i can do things like the app service plan type features functions web api i can have things like event hubs i can have iot so i can have iot hub i'm going to write sql but it's kind of like a sql broker it's not azure sql database it's really this different thing but obviously i'm getting far more services here i'm getting a lot more azure consistent services and again it's a different endpoint but i can use the same so i'm doing all these portals rest and of course a big one might be arm templates so if i have an arm template creating these types of resources i could apply that to the azure endpoint to create the public cloud i could deploy it to my azure stack stack to deploy it here as well so that's kind of one of the big things i'm actually getting through azure stack hub is it's more services and of course it's capacity so what does it do with azure so imagine now okay so we've got the public azure what is it doing here now when i actually get the azure stack hub delivered one of the things it actually does is obviously i have subscriptions in azure it has to register because i drew the idea that hey i pay money for the appliance but i i have to pay money for the services be it vms or storage or app services whatever that might be and there's really two kind of there's kind of pay as you go or i can actually do this kind of per core buyout and if i do the core buyout well that actually lets me run it in a disconnected mode so if i'm connected i have internet connectivity i can connect to azure i can do pay as you go or i can do pcor i have a choice if i'm running in a disconnected mode then i have to do the per core buyout and this is one of the very attractive things i can run this air gap i can run this disconnected so if i had a cruise ship or an oil rig or a submarine or something like that this doesn't have to have connectivity for the authentication yes it can use azure id but i can also use things like adfs to talk to a local identity provider so this is the big appeal of azure stack hub as well this is the option that can run fully air gapped i don't have to have that connection to azure there so i can disconnect this it has all those it's all local management endpoints and again i just buy out the cause so if i now think about getting azure services well azure stack hub gives me a huge range of those services um and it's building they're adding more and more resource providers to this but realize so where does arc fit into this well funnily enough there is no arc it is not compatible it will break if you actually try and put arc on a vm running on azure stack hub the only place you can use arc is here if you have a kubernetes deployment that then i can get the arc capability so i can do kind of the git ops deployment the policy but for regular vms cannot do it so generally it's not arc compatible so that's azure stack hub hey i buy this appliance i pay for the services i get a whole bunch of different services available but i'm not managing it through azure there is essentially no real connection there's a few things i can do by sending some telemetry and logs up to an azure to get insight but it is local management endpoints i'm doing the management i'm having to go and download the update package and deploy it to the hub now what i'm talking about hub and again remember each one is an island you may have heard of this thing called the azure modular data center it's basically this gigantic shipping container um so this is kind of this azure modular data center and you'll hear it kind of talk about the mdc it has kind of the kind of ac kind of controls up there and essentially today you can kind of think about it racks and racks of most likely hub i just think hubs put inside that thing it has obviously network connections but it has this cool ability if it needs to i can't draw a satellite dish but it can go and talk satellite it can have external kind of power unit so i can go and connect up to it that's going to be a very very specific use case uh the average company is not going to be using this thing but it's there um if i'm actually going to open the url quickly get a picture of what one of these things looks like and it would not fit in my garden but that's kind of this azure modular data sending you see kind of hey look we've got the airflow controls in there we've got the exterior powers we can have but it's the idea that again in a disconnected mode if i need it i can have a set of consistent azure services i'm available in a very large scale to really augment and it talks about hey look satellite communications options i mentioned and really just this big thing on wheels that i can roll around and i guess while i'm in here just from that pricing perspective this is kind of the basic price sheet if you don't buy it out the core you can see hey look you pay a certain amount for the virtual machines certain amount of storage managed disks app services event hubs etc and realize the reason you're still paying that is yes you're paying a vendor for the physical equipment but then you're still using microsoft's kind of investment and research and intellectual property for the services you run on top which is why you kind of have that pay as you go you're still using um that part that's more than just hardware so that's azure stack hub and that was kind of the original azure stack and then there was this thing called a azure databox edge and it was really a storage gateway well this is this idea of this i think it's kind of a one u but it's also kind of a mini pc version and this has been rebranded so this is now azure stack edge this also can have things like an fpga it can have a gpu i think it's these the test tensor cores and this is giving me a subset of the features available now this is all about things like well i want azure services like the iot edge iot edge remember it's basically containers i can have kubernetes i can have vms so it's a subset of the services and it can be a storage gateway so hey i want to use azure storage i want to be able to kind of talk to it from this local box for the workloads running at my edge this provides an easy way to actually interface with them so now i think of hey there's azure this is managed by azure so all of the management is up here the deployments are via the azure management fabric this just extends some capacity and services into my edge network if i near need some maybe near um artificial intelligence some near processing maybe image processing etc but it's purely managed through azure it has to be connected i cannot run this in a disconnected scenario it has to be connected and i'm paying the money up here essentially i'm paying azure this per month charge to have this on my premises so this is a nice solution if hey i do have maybe i need some local azure consistent set of services for the edge of my business maybe in stores maybe in certain factories and i need to run these things near really low latency but i want it purely managed by azure and what's interesting is with this kubernetes it actually is using arc as well so kind of the arc for kubernetes is automatically just lit up when i use the azure stack edge so if i think about hey i've got those edge scenarios i need this one new box with fpga or gpus there's a whole set of these skus there's ruggedized ones again there's these mini ones there's there's pro skus i think with different combinations of the fpga and the gpu this is a great solution because it's just going to appear as another target for my deployments i perform against azure hey deploy this to that azure stack edge and obviously i could have multiple of them on my premises now one of the things i kind of drew earlier was this idea of edge zones so the microsoft edge zones and the edge zones with carriers and i talked about this iot there's kubernetes and virtual machines well that probably looks very very familiar and so the other thing that's actually powered by this is we actually have this ability to have azure private edge zones and essentially the azure private edge zones are built on azure stack edges i can have between one and four and then basically i'm gonna add some kind of antenna um or other equipment which is then gonna add like an lte or 5g locally so if i was in a factory and i had equipment bustling around maybe wi-fi is not that great it's not giving me the service i want so these azure private edge zones once again i bring all these services down to here but i can also deploy things like virtual network functions and those virtual network functions can then light up things like hey an lte a 5g service from a number of partners to give me that locally in the facility for all the equipment that connects that then leverage those services running on those boxes so again another type of capability that we're bringing on-prem so again hey i want azure on-premises well maybe that's a solution if it's my on-premises is a bunch of maybe stores or factories and hey i just need to run these work clothes but purely managed by azure great or hey i need to light up a local lte or 5g network and i want to be able to talk to these types of services great if i just had a massive app services or event hubs or vms i just wanted to run i needed new capacity that i want to manage locally well then that would be the hub scenario there's one more now i'm not saying i've kind of saved the best for last but i think i've saved maybe the most applicable to last if you are looking for capacity as well because again from a management perspective arc is that answer there's also aspects of that with things like here but what about if i do need some additional capacity so you may have heard there's kind of a third branch of azure stack and this is azure stack hci hyper converged infrastructure and as the name kind of suggests this is working off i have between 2 and 16 nodes once again they have kind of the cpu and storage locally and it's using that hyper converge to make that storage resilient and available now this is kind of the v2 there was an azure stack hci v1 that was just windows server hyper-v storage spaces direct um and windows admin center so that that's really still kind of the case except what's happened now is this solution it's built on server 2019 but it now kind of has its own branch it's kind of branched off and this is a separate os now you install i download and install hci so it's using server 2019 it's using hyper-v and it's software-defined networking stack it's using storage spaces direct which is that ability to have the local storage and then made highly available and replicated between the nodes and yes it's using windows admin center now it is a very controlled service so when i think about this this absolutely is kind of this hypervisor cluster i cannot run file and print on this thing it is hyper v a virtualization cluster only that is its job it is going to have vms on that thing the management today the management is here you are using windows admin center these are not azure i as virtual machines from the compute resource provider these are hypervms so that's what you're doing here so fundamentally i'm going to end up with kind of let's just different color i'm going to get vms now the value prop here is then i have the arc agent put on those virtual machines now i do want to stress hci is kind of doing some other stuff as well so if i think about well let's zoom this out for a second we have the idea of well there's azure so this hci does register so i'm going to kind of see the hci up here and it does more than just register so i obviously have to pay for this thing so this is kind of a per core per month charge now i do get 30 days free but after that it goes into a reduced functionality it won't delete things but i can't create new stuff it's really going to lock down i can still stop and start stuff but i couldn't create a new virtual machine locally on that environment if it's disconnected for 30 days it will go into reduced functionality mode so it has to kind of check in because if you think about the functionality like sdn storage space is direct they are data center level features i'm not paying for buy not buying a software license so that's where that functionality comes from and now kind of doing this per month service to get those features so it has to kind of go and check in now in addition obviously there's all those other services in azure that this is going to take advantage of so this hci can talk into things like azure site recovery it can hook into things like backup you can hook into things like monitor update management and that cluster remember in azure we have storage accounts i can use it as a cloud witness so it's actually going to use that to keep my kind of cluster quorum um healthy so arc is absolute so just hear on its own is going to use those things but i'm then going to combine hci with ark to get the kind of the best experience because then it's going to start hooking in to those things now one of the things i can do via the windows admin center today is i can actually then add aks for hci again i'm not pushing that via azure today i'm in turning that on by the windows admin center so then on top of those virtual machines what it's basically giving me is a kubernetes environment and oh man i guess we could kind of say it's an aks kinda and once again it would put the arc agent in there and once again that would then have things like the flux agent in there so i can do get ops and pull down those configurations and from that central point so here what i'm adding is capacity and then the management through azure again this is just going to get stronger and stronger today the management is 95 almost to 100 by the local windows admin center i cannot deploy vms to azure and they're going to deploy here these are just hyper vms today but i think this is an investment area this is just going to get stronger and stronger more and more features will light up via hci more and more things going to get lit up for hci with arc to start doing things on i think that control plane both aks and kind of the the hci itself in terms of provisioning and other stuff don't know that that definitely seems to be where it's going one thing to remember though this hci is not giving you guest os license rights so if i'm running windows server in the vms i still need to go and make sure i've got windows server licenses this this license here the hci this per core per month fee i'm paying does not cover guest os you should bring your own so kind of really remember that point so you don't get into licensing trouble this really is kind of a bring your own license just to make sure you don't end up getting your wrist slapped so that's really kind of the the summary again this is a big investment area this aks deployment via the windows admin center today and i can upgrade it there's a powershell command to upgrade it to a newer version of kubernetes but then the actual configuration so i think there's two planes the kind of the provisioning control plane that's kind of hci and then the kind of data playing the configuration plane well that's arc they're getting stronger and stronger there's a lot of unison between them you even see in arc you see hcis these things are lighting up so if i if i was to summarize covered a lot of stuff here in this picture if my question is hey i want azure on premises if what i mean by azure is management nothing else that's arc it could be art for servers it could be up for kubernetes it could be both of them if my meaning is i want azure management and services well which services it could be the arc data services meet your requirements if it's really the services you want and kind of the management or maybe it's the azure stack edge maybe it's hci because these are both connected obviously that hdi is a connected model as well obviously hdi and these are both adding capacity now i can technically build hci on my own servers but the vendors may not support it there's actually vendor validated solutions for hcr you can go and buy and put in your data center if my i need azure on-prem is hey i want a whole bunch of maybe the more advanced services i want local management but not managed by the azure cloud endpoint then that's azure stack hub again that can run it can doesn't have to that can actually run air-gapped completely disconnected and it gives me a whole bunch those resource providers but it is not managed by that public endpoint every hub remember is its own island its own set of endpoints i would connect to but it's consistent i could use portal rest arm templates in a consistent manner as i would deploy them to the public cloud so that's it um hopefully that answered the question hey i want azure on premises it's for many people going to start with arc that's going to be the go-to i might then add some hci to add some capacity to that and again to be managed by arc for very specific scenarios hey edge i need some edge compute managed by azure and then i think for a smaller subset where i maybe just need those more advanced services but i don't want it managed actually by public cloud maybe i need egg apt you have azure stack hub so thank you for watching i hope it was useful until next time take care you
Info
Channel: John Savill's Technical Training
Views: 44,305
Rating: undefined out of 5
Keywords: azure, azure cloud, azure on-premises, azure stack, azure stack hub, azure stack edge, azure stack hci, arc, azure arc, azure management on-premises, azure in my datacenter
Id: PwZdeUUehEY
Channel Id: undefined
Length: 53min 35sec (3215 seconds)
Published: Tue Feb 02 2021
Related Videos
Note
Please note that this website is currently a work in progress! Lots of interesting data and statistics to come.