Overview of HPE dHCI - VMware VMworld 2019
Transcript
so welcome to the CTO dose and vol 2019 my name is Yuki scar I'm joined by breath from eight feet welcome thank you we're gonna talk about DHI from nimble and I gotta say I'm curious you know start with what is it because I think I know but I don't know for sure absolutely so we just recently announced HP nimble search DNC I back and discover this here so it's nothing but a disaggregated hit CI where we kind of give the you know the the flexibility
of a conversion for structure over the simplicity of image to pray and this is predominantly for you know for traditional hyper-converged infrastructure you know which runs on predictable workloads whereas with dity we can cater to all unprintable okay so dissipated to help me out absolutely so the prominent hyper-converged usually is hey I'm not able to scam a computer I might even just give them a story separately so we want to get away from that and say okay you can come you know scale that's required but
in this case you know we're gonna do that with the industry-leading soever is the HP proliant and with a simple management that nimble storage provides right see you can have independence over notes that you can scale you can also add nimbler it's awesome any requirement arose okay so this is let me give this red so this is existing you know or new HP servers per lines for us as well as nimble storage so how does this work you know I'm a customer I've got me running
in production either a storage or either the service maybe even both know how do I start using this Dhers stuff it's a great question so we give this often right so uniqueness about the DHS solution is we can do it pull from the field as well as wait for you right so let's say you're an existing customer using ProLiant servers today and you want to leverage the TFC our solution all you have to do is basically get the nimble array and we have what is called
as a stack manager which automatically detects the servers in environment and then we add it to the stack the only caveat is you need to have a ProLiant gen9 urgent n for you to add to the stuff but other that we basically add it's basically a simple as plugging it plug it in discover and income figure right and it's all done through a centralized console as well so that's pretty straightforward so help me out here because you know what I what I see is you know
this works for new customers as well as existing and of course you need support at our core but you know how does this work from a technical standpoint after you have to you have to install and configure this yourself it's automated and maybe even you know that the biggest question here is yeah so the our point of view in this is we want you to manage this from recenter and not even go in logging into your arrays or server hardware right so gone are those days
where you manage them as individual components so with full stack what happens is we discover we managed to create volumes and with V was coming the picture of storage we actually livery storage space policy based management which actually automates most of the Lum management as well right there's no more actually the more lunch with real ones so all of these capabilities put together along with we Center and the nimble plug-in itself it gives you a lot of flexibility in managing configure environment and the other cool
thing is we've added what is called a configuration checker so let's say you're in a brownfield environment and you're not confident if if this dity solution is going to work well we actually what we do is we do a pretty checkered environment with the config checkers with the best practices that we have and if you find any god shells in your normally we actually tell you to fix it before we go and deploy the stuff so I think that's that's pretty innovative in the way that
we do that and it's all automated through the full stack manager right so that gives you know a customer flexibility to use one there it adds an animation layer to configure it so what about like that today too and those operations does it keep the hardware and configuration in sync does it prevent a configuration drift maybe not preventing admins to change certain settings yeah so we don't give you the flexibility to change what do you have for example if you add a new servers as well
we you can configure whatever excuse you want we're not stopping you from doing it it's as flexible as you get right the other thing is but with Nimbo the simplicity of nimble is carried forward here and it's one pretty much a single flat SKU all-inclusive licensing right we don't charge for the software stack on top it comes with the product so this is this is free assuring correct you can call it that way as well so yeah the software stack basically comes with the product all
you have to buy is basically figure out which array you modely want and which server configuration you need right and then you're off to the races and so you know in terms of scalability I'm presuming a similar model you can you know you can skill compute you can scale storage correct separately correct so we go in a plot model where you can actually scale up to 22 servers as of today I mean single party can have a nibble array attached to it and then you can
secure multiple parts together right so a nimble array can have a group of nimble areas and before of them right can be managed together as one so if you can scare up to four plots which is 88 servers so I think that's that good enough for most environments but we are in the process of making a higher number of server compute nodes and so you know 88 servers in in a pod or in a configuration I should say yeah so you know that targets the smaller
customers out there I mean this is not going to be the big enterprises running thousands of physical servers yeah so you know with that in mind what the use cases are you naming this formal type of applications you know what are customers using your phone yeah I mean from from the the days we've released it we have seen a lot of use case for VDI based workloads you know it's because they want this simplicity for management along with scalability because VDS kind of unpredictable as well
in some cases so that's why I eventually you know it's this solution fits into most of the unpredictable water scheme up yeah so it plays well into those markets in and that's been one of example of a big driver for this type of solution right yeah video makes makes sense and I got me you're saying I'm in terms of predictability versus unpredictable I mean you can scale this in two separate axes mm-hmm so that means you can you know you can take on that unpredictable world
clock much easier than in a completely integrated solution right so tell me a little about that predictability over predictability how that's carrying words you know with real customer plans yeah so the other big thing about intractability and making it predictable is with info side right HP info site so we have we've had traditional integrations with VMware environments with what we call this plastic analytics where we can actually pretty much tell you you know latency in iOS where it's coming from the storage array but since we've
built an integrator HP info site into servers specifically provide in this case we have much more data that's coming in so now we can actually pinpoint truchaut information right we've all discussed our recommendation we can actually tell you where the pain points are or where the problem is so we have a like color-coded like let's say if the VM is sure it'll be blue with if it's storage issue screen right so easily we call it implicit recommendations and we actually do the root cause these are
the steps that you need to take for you to even you know you go back to a pristine environment so those kind of integrations makes intractable work as well yes that's a good point I'm assuming this you know has some integration in terms of tearing your storage as well so in imple the tearing is a little different because we actually don't do tearing it's all flash and the castle algorithm itself manages that for you because we have built-in QoS Auto QoS and within folks like what
happens is in facade actually gives you let's say you wanted a man okay us emphasizing information the last 24 hours based on what your usages and we give you recommendations based on that so it's a very powerful tool itself and so the the point of being is it has to be low touch yeah we don't want the user to go in and we want everything to be automated so that's what this should provide yeah so what about integration with other cloud volumes are offloading it to
the cloud absolutely so everything that nimble has today in terms of integrations in the ecosystem plays into this solution as well and one of that is Cloud Data Services follow Patrick Williams so let's say you're running on premises water with THC I you can always move workloads or flood volumes it's basically replicating the card ones right and then you can attach it to any public cloud of your choice that'll be Amazon ec2 or your VM we also recently announced Google Cloud integration so you need leverage
that the other unique thing with the solution is unlike other public cloud vendors you can bring back the data right you can replicate the data back on premises when say does the cloud solution doesn't work for you you always have an opt-out in our case yeah so there's not there's not much ingress charges as the other vendors charge you so that's another cool thing that you can have the service solution yeah all right so let's switch gears a little bit you know one of the advantages
of doing HEI is kind of that single throat to strangle right a single channel for support single so with this Dhai I'm assuming the support will you know be simplified as well absolutely so it is going to follow the same support model and nimble as leverage so I don't know if you guys have you know had experience animal storage but is one of the world class storages support system that we have for for our storage infrastructure so a DAT I would play into the same model
right it's one support vendor and if you call our support engineers well pretty much level three engineers and you can pretty much call them and somebody would pick up a call within the next ten to fifteen seconds right and that person would be able to help you troubleshoot the entire staff let let not be HP servers and storage you could also get in a provision where we would be able to troubleshoot the issue with that there so yeah it's pretty simple it's against flat support SKU
that we use for nimble and it's that's pretty much the same for THC as well yeah that make sense yeah nice right so what about the operational experience rep so you mentioned this is simple this is a single SKU single support there's gotta be some story about simple operations as well absolutely and that's the constant undercurrent in this solution right so as a VI admin you really don't want to spend time building infrastructure right when you can do something else cater to your day loves or
your engineering team and that's what we want the solution to be for them right and that's why you know all the we have Ben's are very familiar with VMware vCenter and we Center is pretty much in 90% of the market right in terms of one prime infrastructure so we are basically using that as a focal point say okay since you're familiar with this tool we now want introduce a new tool for you leverage your V Center and manage your staff from the V sound right and
some of the things that we've built into this pretty cool like you know you just have to attach your rack and stack your servers in your network we are about to discover those servers for you whether the recenter plugin that we have so that's that's how unique the solution is and leveraging existing infrastructure tools that we don't see every day so no going back and forth between different consoles you know we're setting it up for configuring it then you know managing it from vCenter it's all
from within a single kind of class that's correct that's correct yes and so talk to me about the you know the the architecture of that single management interface so what we do is when you first install the DNC eye solution we ship the staff manager with the array so the intelligence is built into the nebula itself right so what happens there is let's say you're you're adding a new server in in the case of Greenfield we actually should be server with OVH built into it so
we actually take you to the extend of installing ESX I would be 65 could be 60 70 support both it's automatically installing all the server's all that we require from the customer is a set of IP addresses right once you give me our basically IP go servers install ESX on then we even go to the extent of creating volumes on the array and creating initiated groups and attaching those hosts to those volumes and you're off and running right so from stacking to the entire wizard and
set up takes 15 minutes and you're ready to go so that's how simple this tag manager makes it that sounds it sounds great all right I've learned in a bunch of about DHEA and and you know for me from the city of ours I think now this is it I think we we learned a lot about the issue and we'll see you in the next interview all right thank you very much
