Drut s Composability Webinar 090823

Show video

Greetings everybody Simon McCormack here from  Drut. Today's webinar is our second webinar   this is about Drut composition, I'm going to  do composition of a node and you'll probably   wonder what that is about well throughout this  session I'm going to explain it to you and show   you all about it it's a word that's been used  in the industry but it's actually very important   particularly for the Drut technology now last  time a couple of weeks ago I did a session where   we did a dynamic attach reattach and detach of of  resources to a previously composed unit so this is   kind of the precursor to that where actually we're  going to set it up first of all and it's important   because there's a lot of very important technology  that I'm going to show you that we have within the   Drut environment that allows you to build a  multi-vendor data center environment into a   situation where you can basically do that Dynamic  attachment and Detachment there's some dry parts   of this demo where I'm going to go into building  machines so you can watch that your leisure I'll   try not to make it too boring hopefully it'll be  interesting enough I'm going to start off with   a few slides just to try and explain a little bit  more about why this composition piece is actually   important and interesting to you so we start off  with a little slide on on droot itself and some of   the messaging we've been using about environment  and I think one of the key ideas and thoughts I   have that seems to resonate well with customers is  basically better use of your existing Data Center   so what we often see is when a customer is looking  at a new data center pod or Rack or or whatever   they're often buying brand new servers nothing  wrong with that we love you to buy new service   what we have the opportunity with  Drut though is to actually use   existing machines as well as new machines  we don't mind go either way you want I think   that's the core value disaggregation  with Drut is actually better reuse   better use and reuse so anything new you  buy you can use and make better use of it   what you bought last year there's nothing worse  in my mind is when you know your your environment   you deployed some machines last year and then  this year everyone's talking about gpus and AI   imagine that you can't use the machines you  deployed last year so well they're no they're   no good to us that's a terrible situation to me  actually with root technology we can change that   we can actually repurpose those machines so we  build a lot of Hardware capabilities into the   solution plus software capabilities that allows  you to do this uh Dynamic distribution of the   environment to create it and make them more  efficient and so this composition piece that   I'm going to show you today is the core founding  fundamental uh way that you can turn your machines   into these composed nodes and then you can add  the right resources that you need to and we have   a lot of options with our software so we've built  it in a way that you can API Drive everything so   you can automate everything and we've layered the  software and Hardware so there's a lot of choices   in what you do so what we show you today there's  a lot of options for what can use but we consider   ourselves a complete system company at this point  because we actually have all of the software   pieces in place for you to create this brand new  data Center private Cloud environment using all   of our software and hardware and it can augment  your existing processes it can replace some of the   existing stuff it can sit side by side with the  existing software as well so a lot of capabilities   there now this next side I always put this one in  because there's generally three key things I want   to say and sometimes I forget them so it's good to  have a slide sometimes I say them all in the first   slide so so then I don't need to say them again  but I think it's good to reiterate The Three core   tenants that I put in this slide maintaining fluid  resources software and Hardware that's the key   thing I was just talking about make better use of  what you build new or existing so I don't continue   to throw old Hardware away be able to better use  what you had before and make better use of what   you're going to deploy tomorrow and think of it  in terms of The Sovereign Hardware that you can   put on it and droot software and Hardware can help  you to go down this path to to making a what is a   better environment and one key part of this and is  the standards-based approach what I just said to   you there has to be a multi-vendor I've never seen  a data center that's not multi-vendor both compute   storage and networking they're always multi-vendor  and so she's only got one single vendor platform   to do it makes it challenging right so parallels  like Drut is great because we're what I'll show   you today is a multi-vendor environment  composing a node both new and old systems   all right let's dive deeper a little bit I'm going  to look at a rack when Drut we're always drawing   racks of equipment because there's servers so that  the Block in the middle that's a bunch of servers   there I think there's networking on the top we're  not going to go into networking today too much   a little bit uh I've got eight servers there and  the eight servers have got the drute ific in them   so that they're connected to the photonic fabric  so there's just eight standard type servers at   the bottom there a couple of resource units with  a bunch of different resources it's a big double   wide gpus some little gpus some smart mix send  MDB and nvme drives a whole bunch of different   resources that you can build in to that that  gives you a lot of advantages um when you build   this kind of server so that's what you can do that  little server there on the left when you compose   it the demo we're going to show you can compose  it add all these resources suddenly that little   server becomes a big server very powerful server  on demand and tomorrow you can change it make a   different server add the different resources and  that gives you these four key values on the left   you know no stranded resources so you don't have  GPU sitting machines that can't be used anymore   uh we have plenty of resources so now I have  a machine that maybe I only had one or two   GPS before now I can make one machine have many  many resources and build the impossible servers   uh power options so I've got a lot more better  options with power because the resources like   the GPU the power hungry bid is actually  not inside the server now so it gives me   a lot more flexibility in terms of how I power  them and savings and Power and then finally I   got these interesting peripheral options so in  the past I say once you've plugged a GPU into a   server you're never going to open that server up  and move the change the GPU to a newer one it's   just never going to happen but when you put them  in a resource unit actually it becomes possible I   can now fit in New versions of gpus new types of  GPU or new types or peripheral into the resource   units and then dynamically attach them so it gives  my data center a much more fluid accessibility to   those resources going forward now this is all made  possible with the software so there's a software   slate as I'm calling it and it's not empty it's  going to be filled up soon there's a software   slate that makes all this possible and some of  this I'm going to demo today so start off with   a BMC bios and ipmi now I know whenever I do a  webinar I mentioned these three terms half the   audience leaves but actually it's not that bad  in this case it's going to be it's important for   us you as a user you don't actually have to get  into it until we basically in that layer from you   but it's important for composing a machine with  resources obviously we need to be at the BIOS ipmi   level and there's a lot of different ways to do  it and you'll see in the software we have actually   can be multi-bender I'm just trying to show we we  need to understand at that level uh in the in the   environment to be able to compose these nodes  so that's step one that I'll show in the demo   and it's cool and it sounds actually secondly  you've got to put an operating system on your   server and there's obviously in your data  centers there's many ways to do this you've   figured out many ways to do it uh and if it's  a multivendron right you have you may have   many ways to do it in your own environment so we  can give you a single multi-vendor way to deploy   your system OS and I'll show you that as part of  the demo as well and then there's there's more   physical networking on the server not going  outside the server today just the network on the   server when you bring up a new server you do have  to do some networking configurations vlans IPS   DHCP we can do all that for you with this software  as well so you see I'm building up this layer of   software here that is vital to your environment  that we can offer as part of this solution   and then the final one for today will compose our  system so we'll build you a composed system that   has its operating system its networking setup and  we can control it at the um The BMC level uh to   be able to manipulate the resources what do you  do next with that system is going to come later   there is more just a bare metal machine with  an OS is not the end it's the beginning of a   data center what comes next is a virtual platform  virtual Network virtual storage virtual machines   and container type environment to be able to run  your workloads we've also built that into our   software suite we're going to cover that in future  webinars what I'm going to take what we've done   today in building this composed environment and  we can go into the virtualized software level and   the reason we want to do that because we want to  go to the next level which is the software tools   I put AI workloads in there any workload it's  possible the exciting one is the AI workloads   but once you've composed a droop system with  gpus deployed the virtual platform software   the next thing you wanted to start to do is is  AI workload work so that's that will show you   that Journey you can go on with droot you  can go on a journey with us taking you all   the way to application deployment a single  front-ended portal Suite with a lot of uh   droot built software using the best of Open  Source all integrated into a single platform   comes a very exciting system level solution that  we can offer you so today I'm going to do these   these five things I think you know we'll we'll  start with in this environment I'm starting with   a an existing data center I've taken some servers  I put the droop fit card into those servers and   connected them to our photonic fabric so it's new  or existing I think this is the the key message   of the day that that we like to think about now  what servers you might ask are compatible with   the Vic 1000 that's a good question well most  of your modern servers have been built with   the ability to take a GPU and some and those  gpus are big Power hungry wide modules the Fig   1000 is not as big as them so generally if you  can put a GPU and a server you can put the pic   1000 in to replace it or augment it you don't  have to replace it so a lot of your existing   data center is already be compatible with this  module and as it will show you the software we   have today it's multi-dender so it'll work with  that environment and I'm not against or for any   vendor in that what I'm saying there is every  data center I've seen is multi-vendor doesn't   matter where it is even if they are a big fan  of Dell big fan of Super Micro big fan of hpe   you've still got some service you bought last  year from a different vendor so they all become   multi-vendor which is a I find that being one  of the biggest challenges when I talk to Data   Center operators is dealing with the multivender  environment they have and the tools to use it   so it's critical for us to be able to work in the  environment the second piece servers discovering   the servers commissioning a BMC and RPM this one  I'm going to demo very shortly order automatic or   manual New or Old or and old manual new and old  that's kind of what it is so we're going to show   you an automatic method manual method new and old  bring them bringing servers into the environment   so that we can then eventually compose the  service before I do that I'll get into the   provisioning the network and so I'll talk about  show DHCP VLAN IP address configurations that you   can do and you might say well don't you need  to do that after you deployed the OS actually   we can prep it with that software so you can prep  the networking options for the machine before   you've deployed the OS convenient you can change  it after you've deployed the OS nothing's stopping   you that but the software suite we have allows you  to basically build that that provision networking   piece as well and then you know that next step  will deploy the operating system onto one of the   systems and it can be operating system of your  choice or anything that you can use and you know   in any in any production environment you're going  to have a mix of operating systems depends on the   workloads that you have and then finally we'll  do a composition of a server now what I've talked   about here there's a lot of stuff to do this live  I mean do it live but I when I sat back and think   about how can I do this live because this involves  discovering machines booting machines deploying   operating systems we don't want to be here for  an hour do we so I've got a number of machines   in different stages that I'll show and hopefully  it'll make sense I think I'm on to demo time so just looking at the clock see how much time I  talked for I think I talked for more than I was   planning to talk for say lovely the recording  I'll cut it down to a much shorter time frame   that's the beauty of recordings we can do that  all right I've jumped onto this machine actually   it's not the machine I wanted to start I'm  going to start on this machine so this is our   um what do we call it this is  our DSP droot DSP software piece   and I've got a bunch of machines in there  we'll see this guy this is two actually I   didn't rename him last time I discovered him  this is three I'm going to delete them both   I've done this a few times and I want to delete  them because I want to ReDiscover them deleting   them from the system is that it's essentially  they'll be fresh to us so I'll start off with   um one of them let's go add Hardware so this is  the manual adding of a hardware machine I gotta   give it a name this is my dsp3 now I don't need  to touch most of the options I can just name it   give it a power type example ipmi where  you see this multi-vendor there's a lot   of different options there so it's an ipmi  system I need to give its IP address you   can see I did this earlier right this is my  browser filling this in for me and it's not   right and then I gave it the password for  the ipmi access that's all I need to give it   and it's going to go off and basically  discover and commission that machine   aside machine you see it's now commissioning  and powering on we can view what it's doing   I mean it's got to build a machine to go to  the machine it's going to take a little while   so I'll start it off I'll put it on the side and  then we'll we'll observe this is 130 right this   is this is the 130 console access and I'll  launch the console and I'll just leave it on   the side and we can observe it there's some  interesting times which I'll I'll jump into don't just have to log in of course and that'll that'll let me  see that and then we can as   we observe what's going on I can point to it and it should give us some interesting  some interesting points as we're   going through it and launching just  the console access to this machine   so we can see as it's booting and we should  see it's like uh it also gives me a bit of   confidence in a live demo that I know it's  doing something nothing worse when you're on   a live demo and doing things and you realize  you forgot to press a button say there it is   off going doing its commissioning on the side and  obviously our software gave us that value as well   now I have another machine as well I want to Auto  discover I want to show that difference because   the software we have here can as the machines  boot we can discover them as they boot so the   machine that didn't tell you anything about so my  127 it's this guy actually I'm going to go to him and I'm gonna power cycle him actually I'm  gonna lose I want to pixie boot him because   it'll detect him through a pixie boot  because that's a login right makes sense if I want to pixie boot him so when he picks he  boots it'll be picked up by the system and then   automatically brought into the environment chain  you can just you've got tons of machines that   you need to deploy machines are actually off but  I'll set up I'm gonna make him pixie boot one time   that pixie booting he'll be then  detected and we'll be able to bring him into the system as well power him on   and I'm just going to ignore him for now and I'll  carry on with the damage I've got two machines   churning away doing stuff you see we detected the  one that we manually added this to take a while   so this is going up and basically discovering the  machine it's actually booting right now once it's   booted it'll go through the ipmi commissioning  discovering the resources that that system have   putting in the inventory and then setting it  up available for us to do the next step and   these machines you see on the on the right here  have already been deployed in The Next Step they   went through that commissioning process and we  discovered all their resources I think that I   can click on one now and I can show you this the  environment that we have so it's learned about   the physical bare metal information inside This  Server knows it's a Dell server the cause memory   storage networking learned all that so you've got  a really powerful Inventory management tool for   your data center just with this software which  is actually something that's critical to a lot   of environments anyway and you maintain this  and this works for both bare metal and version   machine so it's an interesting side with virtual  machines we'll get into in a future setting now   if you remember the steps I do the next thing I  was talking about was so this machine's ready to   deploy I can deploy the OS on it but I want to  talk about a couple of things before I before   I go through the deployment first of all it's  a networking setup so it's discovered uh this   machine it discovered all the networking the  physical net interfaces that are plugged into it   now if we actually look on this machine on the  left we'll see what it's doing now it's it's   doing its pixie boot now so this machine that I  manually added it's gone through booted up and   then doing a pixie boot that the our DSP software  is supplying the Pixi boot software to it so we've   interrupted the the boot cycle and we're giving  it it's um base operating system for us to do   the commission so it's progressed through that so  the software we have here actually provides a lot   of capabilities one is you've just learned we can  pixie boot machines um back to my networking setup   so I've learned the next and you see previously  on this machine I configured some vlans onto it so   I basically said uh no dsvp but here's a VLAN uh  an IP address and subnet that you're provisioned   on and you can do this to all your machines this  is essentially before I've deployed the OS think   about that the OS has not been deployed but I can  do it before that so that when the OS is deployed   it's now configured with the networking setup  and they should be able to talk to the network   and your application should be able to much easier  to deploy them so it really simplifies that whole   networking piece when we set that up now within  the software itself this piece if I go into the   subnet configuration and I look up my 53 VLAN for  instance I can go into it and I can configure DHCP   so not only is this software a pixie boot server  it's also a dhp server and these are optional   you can do the dhp through different resources  but this software that we have as part of this   true package becomes a complete system software  that can use through a data center it's actually   really powerful software it's built along a lot  of Open Source so the DNS Master HTTP config   utilities that are often used in millions of  data centers around the world are part of this   raw thing you can manipulate manually as well  in this setup but here's a nice UI that lets you   manipulate it see I've got a reserved range  of addresses that I can apply when machines   attached to the dfcp you'll see them listed you  can statically reserve addresses for machines so   a comprehensive DHCP serving environment  and you not only use this for your bare   metal servers because the servers themselves  need addresses but also the virtual machine   layer will eventually deploy you can use this to  manipulate that as well making the entire system   working nicely within this network environment  government provided by the the Drew software   that was networking let's go back to my machine  because there's more you see a bunch of different   tabs along here another one is storage so  obviously the bare metal machine comes with   a bunch of discs and you've got to put file  systems on the desk so so when they boot they   can deploy the software and you can manipulate  that file system piece from this page as well   you see all my petitions that are available add  new file systems modify the disks modify the boot   disk and a lot of the The Better World battle  creativity that you need to do on a new system   again giving you really comprehensive bare metal  usage outside of the droop core value prop for   disaggregated PCI devices is in addition to that  you get all these capabilities with the solution   becoming very powerful our PCI devices uh this  is what we're going to go into more very shortly   because I'm going to actually compose one of  these systems into environment it gives you   the basic PCI structure and so on and so forth  through through the different configurations so we'll notice our um the machine I added My Demo  three so that was the one I went in and I said   it was a machine that was sitting there it  essentially had nothing on it imagine you   literally just brought it out of the uh box brand  new machine racked it and all you've done is given   it an ipmi address so we do need an ipmi address  so using whatever you have in your system for ipmi   it's got an IP address you gave that IP address  and credentials to fruit and that's what we did   and we said you know we then took control of that  machine brought it up into this state where we can   we can now basically deploy an operating system  on it and we can actually that actually this is   good time I can go to the network portion here  you'll see it's it's all blank because it's just   some physical networks it doesn't have the vlans  now I can start adding the VLAN to this machine   but actually I want to do the next step I want to  talk about the next step is that I've got a fully   commissioned machine that I can now deploy the  operating system on it and it does make sense to   do this staggered kind of approach uh it happens  in all your data centers through your servers this   data kind of approach there's usually a lap guys  done some low level uh thinking to to figure out   how to do it in your environment here you can  do it nicely in this one's one system so I go   into the deployment now for this machine that  I discovered earlier so you notice here in the   deployment now I get a set of uh let's go to  the bottom a set of operating system options   you can load your favorite operating system  to the environment I've landed centops on   Ubuntu onto the setup but you may have a Windows  server or or some other server that you use that   you like in your environment choose to release  kernel level and then some more options I can   actually do some Cloud init boot options so when  it boots up run some more procedures essentially   you you know you've got a a build that you want  that works for your applications you can make   this system then deploy it so we can just I won't  I'll just hit start deployment on this one so he's   going to go off and start deploying Ubuntu onto  that machine if I go back to my machines you see   them in the background he started he's noticed  someone started to do that didn't and you see   these machines at the bottom here I have some  previously deployed machines that I that I that   I created the deployed machines are ready to do  the composition and we do that in a moment before   I do that you see this guy's deploying Ubuntu  now in the meantime our we've discovered a new   one see it's marked as new this is actually my  it doesn't know the name so I'm going to tell it   demo dsp2 actually is his name so that's the  machine that I basically I deleted it at the   start and then I just that booted so that was like  a you think about this your operator in your data   center has got 10 new servers he's racked and  powered them on the IPM is the extract them   and powered them on he didn't even need to give  them an address at that point it's been learned   new by the software we didn't tell it anything it  automatically learned it so you can see there's   the options uh dynamic or dynamic or manual  automatical manual for discovering machines so machines once they're deployed they've got  their operating system on now they're ready   to to build into you know what we call the  composed node so that Drew can then do the   dynamic attachment Detachment of resources I'm  going to jump to this guy over here slightly   different setup so yeah that was my plan with  this is to have different machines in different   stages because I wasn't sure the timing actually  the timing almost worked right maybe I talk too   much in point where I actually got the point where  I could have composed it but I go to this setup   here and you'll see I've got a bunch of machines  here that are have been discovered these bare   metal machines and I go to this machine so this  is a super micro sm-30 and I go node composition   screen you'll see this machine is not part of  the Drew environment you have to compose it   first it's saying it's basically the bare metal  node it hasn't been composed I'm not going to   do it here I'm going to go to our node screen  because I want to show you all of the nodes   there's one node already completed so this is a  composed node s26 but I want to go compose node   and choose the Zone I give  it a name it's Simon's node   hopefully give it a better name in your  environment and then I go into compute   blocks that are available so you think about this  I've racked 10 new machines I've gone through this   commissioning process that I couldn't could have  done through the automatic manual so it's actually   pretty simple it's like you you think about that  the operators have racked the systems they turn   them on they walk away when you're admin for  the infrastructure arrives they're already there   within this system we've already automatically  discovered them they're already there prepped   ipmi has discovered we've discovered all of the  resources ready for you to deploy the OS and then   compose it all of this can be done through API so  it can all be fully automated it but it really is   the steps I'm showing actually become very simple  so you think about that with the composer nodes   when you want to compose you better say I want  to compose a node because you've now got you   think about this from the the application user  perspective he says I want a machine with 10   gpus and this much CPU and this many disks on it  and this networking configuration of smart Mick   something like that from this API call from here  you can see these are the available machines so   these nodes are available for us to compose into  a specific mock block we'll take the Super Micro   and we go next on the supermicro and now you've  got the option now you can just compose the mode   with nothing on it because you once you've  composed it you can dynamically attach some   resources you don't need to compose it again  in the essence composing is kind of a one-time   thing do this at start a day and the attachment  Detachment can be done later in this session and   I'm actually composing the node and then I'm  choosing some block storage resource blocks   to attach in this case I've got storage so I'll  grab a couple of storage blocks and attach them   it says basically I've taken this uh compute  block with super micro 110p server and I'm going   to compose it and in the same process I'm going to  add this storage to it and if I had gpus in this   setup I put gpus here so you can do it in all in  one go I don't need to dynamically attach it in   that case because obviously I've just done it in  one session and I finally say compose the system   and it then begins to go through the process to  come to compose it into environment with those   resources attached and it's really quick because  it's already done so this part is super quick so   if you think about that from the operational  point of your environment a lot of it can be   done automatically from machines being plugged  into the network and powered on figure them to   connect them to the network a lot of that can  be automatic through this software so they   automatically get into the point where you can  compose them the actual composing then is super   quick the adding and deleting resources is almost  instantaneous and happens within a few seconds   so that is the end of the demo that was  what I was going to show today the the   next session that we're going to do I think  we're going to get more into the the virtual   machine container storage layer to show where you  can go next posing machines are a little bit dry   it's actually very exciting to infrastructure  people but I think it gets more exciting as   you go by that stack so we want to do a little  bit more of that so in my um in summary today   what I've shown you I have done composition of  a node into the root environment but I prior to   doing that I basically showed you how easily  we can bring bare metal machines into this   environment and actually complement a lot of the  service you've already have in your data center   with the droop software really adding a lot more  value Beyond just the composition of the nodes   I don't know if there's any questions I need  to ask for the live you know no questions to   answer live that's why anybody interested  follow-up please feel free to contact us at   droop.io my name is Simon McCormack and it was  pleasure talking to you today thanks very much

2023-09-15

Show video