Maybe the SDDC isn't Vaporware - Tech Talk 44
Transcript
hi this is Keith town from virtualized geek.com and you're watching episode number 44 of virtualized Geeks Tech talk so today we're going to talk about the software defined data center yet again but this time we're going to take it from Theory to a practical application so one of the challenges that we see in the Enterprise all the time is developers and the battle between developers and available resources within the infrastructure so specifically we go through development cycle in the infrastructure or uh within the infrastructure where
developers require more resources or resources quicker than the infrastructure team can provide them so this is a real business challenge that cloud computing normally meets the ideal of elasticity and Agility the ability to give the business owner the application developer the resources that they need to react to business events so it can be uh you have a external facing application that serves your customers and there's a unexpected uh advertisement or sale that gives away uh 880,000 widgets for the price of 40,000 widgets and now you
must meet the demand of the your application is functionally not able to meet the demand of the um promotion so you you need to go through the these rapid test and uh development Cycles when you need to expand the application so real business challenge uh that the infrastructure today typically typically can't meet so from a generic perspective this is pretty simple problem you have an application that's comprised of x86 compute via some hypervisor uh storage that sits on some storage array and a virtu or physical
Network that's needs to be replicated into your cloud service provider whether that's AWS or Google the problem is that Amazon and Google weren't designed for legacy applications so when you want to test a legacy application Within These environments you have to go through and modify the application whether it's the type of the way that it consumes storage the way that uh the network is designed some change some basic change to the application needs to be be made this presents two challenges one it's a technical hurdle
and then two you need to give up Cycles precious Cycles from your developers to make these changes to the application so you create a overhead for when you want to move the application from production on-site production or on-site test to the cloud so this is kind of the Hidden cost of the cloud but this is where our the concept of the software defined data center comes into play the whole ideal is that your application and your data center in general is abstracted from the underlying Hardware
it it doesn't matter the type of hypervisor the type of storage or the type of network that your physical infrastructure runs under in your production data center the ideal of the abstracted software defined data center is that I can take this construct and move to whatever service provider that best fits my need for that particular uh cycle of testing uh AWS if they have a better pricing or better uh type of performance for the workloads that you want to test we'll test it on AWS if
Google has better pricing or better uh infrastructure or infrastructure in the area of the country or in world that I want to test my application then we move to Google so in order for this to happen we need some type type of translation layer there's a few different ways to do this there's a we can use uh apis from providers such as right skill to build that type of capability ourselves so we can create a portal that says you know what uh pick a provider and
we can write to generic applications layer we can uh write to generic apis the challenge with that is again we have to change the application at some point if we're looking to solve this problem from infrastructure pfic uh view of the world Rell systems has a really Nifty solution Rell kind of abstracts this layer from us and provides a nested hypervisor they'll run on either AWS Google and I believe they also uh support rack spaces or hp's Cloud as well the ideal is that you can
take your application wholesale you can take the V DK in The Logical Network design import that into AWS or import that rather into Revell and then they can run that using their nastic virtualization software on any service provider that best fits your needs or best cost depending on which one gives them the better cost one of the cool things about this solution is that you don't even have to have an account with AWS or Google you have the account with Rell Rell basically resales the underly
Cloud solution more importantly this gives us the ingredients for the software defined data center because this communication from data center to Data Center it can be done either through rest apis or through some type of admin console that Rell uh provides the rest apis is very important the reason why it's important because I can take openstack vmware's uh vcloud automation Center vcat and create a self-service portal that allows developers to basically say you know what I want to take these 8 VMS that compromise that sorry
that compose my application export those to uh Rell rll will automatically build the network logic based off the metadata of those VMS and then provide the developer a complete mock environment of the production system with the same IP addressing and I can have four five six seven different developers all do the same activity basically at the same time within rll Rell can uh make snapshots what they call blueprints of your application and allow you to uh basically replicate the application as many times integrate that into
your configuration management or your uh software development platform uhof development Change Control platform to allow you to test and build the application in as many containers as you need it to be so this is pretty exciting not just because from a practical sense we have a uh a a solution for a very common business problem this is important because I've in the past called software defined data center the software defined data center a great concept that was to a point vaporware this takes it beyond the
bware perspective and gives an example of a real business challenge that is solved with a solution that's on the market today and being leveraged today so that's it for this episode of virtualized Geeks Tech talk we'll talk to you next episode again like the YouTube video visit us on LinkedIn or uh via virtualized geek.com thanks a lot
