Virtualization Explained on a 'Napkin'

Wondering what all the fuss is about concerning 'the Cloud' and 'Virtualization' in cyber-space? I thought I'd provide a simple 'bloglanation'. Think of it as a Virtualization 101 or a napkin explanation of Virtualization...a very simple one.

Many people think Virtualization and Cloud computing are synonymous. Actually, they're not. To generalize, "virtualized" computing really is a broad term that basically describes an approach or design philosophy to computing and IT data center resources. It really came into it's own during the late 90's (with the advent of VMware) and now pretty much dominates our IT server infrastructure. You often hear or read about lavish references to the term "virtualized servers". Why should you give a care? Per my previous posts here on Cloud Computing and our Virtual World, virtualization is a key software technology enabler to our digital world and internet hosted services (i.e., applications) we all enjoy so much today. And while virtualization and cloud computing terms are not synonymous, virtualization is a core/key enabler of Cloud computing.

In simplest terms, a virtualized server or 'computing platform' separates the underlying computer hardware (i.e., 'physical server') from the operating system and application software. This basically sets the stage for installing multiple and even mixed operating systems on a single server (ex., Microsoft Server, Linux, etc.) with a set of applications installed on top the O/Ss. Thus, you can maximize your underlying server hardware resources and utilization rates across more 'compute instances'. Those of you happening to be using a Mac (hardware) running MS Windows are already experiencing a basic form of virtualization and separation of O/S from your underlying CPU/computer hardware 'platform'.

No longer is it a one-to-one relationship (like in the old mainframe and mini-computer days) when a single hardware computer (and CPU) was installed with a single operating system and resident application software. Virtualization software has disconnected these resources from one another and allowed them to be installed 'virtually' on multiple servers -- as in a local server 'cluster' comprised of multiple physical/hardware servers (each with multiple/separate CPU cores) all connected together by a LAN (usually Ethernet). And unlike the mainframe and minicomputer days when if the server went down you lost access to your email, other applications, data or files, virtualized servers builds in redundancy and high availability (in a cluster) so that loss of a particular server or 'crashed' operating system or software application doesn't disrupt access to that cluster -- or what you were doing there in the first place.

The diagram below is a good picture of what virtualization is all about.

Per above, a virtualization layer created by software basically sits on top a physical computer (i.e., server) and -- in turn -- multiple operating system VMs (images called 'instances' or 'virtual machines') can be installed independently on the physical computer shown.

This is cool stuff because now you can create a cluster of multiple virtualized servers that in effect back each other up (i.e., replicate) if one goes 'off line'. How? Well, that gets further under the hood, but it's much easier (as you probably well know) to "cut and paste" software from one environment to another without having to worry about needing to re-install or migrate an entire physical server O/S, utility software, applications and -- yes -- data. In the old days, it used to take a day or so to bring a new server up online or restore operation on one that crashed. Every thing had to be migrated off it and 'ported' or re-installed again. Remember those LONG interminable 'off-line' or 'unavailable' IT messages you used to get when you were trying to access your mail (Outlook Exchange server) or some database? Not fun. Major service disruptions.

Now in addition to 'virtualization layer' and 'virtualized server clusters', you may have heard or seen a lot of references to those VMs. Again, VMs (and SVMs for storage virtual machines) are virtual machines that install on top the virtualization layer. The virtualization layer itself is mostly created by specific O/S (or platform) hypervisors. Hypervisors are essentially stream-lined, cut down operating system 'subsets'. VMware has, for example, ESXi as a hypervisor. Microsoft has Hyper-V. There's also a KVM hypervisor for the Linux crowd. Basically, hypervisors allow you to install multiple VMs and Applications on top of their 'virtualization layer' for each physical server. Again...all these hypervisors, VMs, applications and -- yes, even data -- can be replicated or copied to other virtualized servers in a cluster to ensure redundancy and continuous, uninterrupted operation and access to 'the server'.

And you can even aggregate (or unify) data protected virtualized storage across those participating virtualized host servers simply be linking together their local 'direct attached storage' (DAS) devices/capacity to form one big data protected storage pool across the cluster. VMware's VSAN (virtual storage area network) is one such example. Dell EMC's ScaleIO is another 'virtual' or 'server SAN' example (based a simple software defined storage license install). In fact, these SDS based virtual server SANs are the latest/greatest thing now in the IT storage world and virtualized computing.

Here is a block diagram of what a VMware virtualization platform looks like compared to the traditional architecture (and bad old days) explained above. It comes straight from VMware. Note that within the VMware layer shown, both their ESXi hypervisor and a virtualized management layer (such as vSphere) is assumed. Hypervisors in of themselves can't manage a virtualized server platform or session/instances. You need to have something like vSphere installed so you can actually manage and control all this virtualization.

OK. So you could give two cares, right? You're not into data centers, technology, abstract virtualization stuff or servers. Understood. But the reason why I wanted blog-off on virtualizaton technology/architecture is this...the decoupling of hardware from O/Ss and applications into separate, portable 'atomic units' or entities has enabled the whole "Cloud Computing" (and even Data Lakes) 'style of computing' scene.

It's important to note, by the way, that not all internet surfing or computing you do is necessarily Cloud based computing. We do much of our online shopping or other web-based computing (filling out forms, looking at pictures, accessing specialized services, etc.) via web pages or -- more specifically -- web applications that are specifically coded in JAVA Script, HTML, and other web app programming languages. Cloud computing is different. Virtualized environment software, services and application servers are not coded in web based scripts or languages. These are literally the same computing/application sessions you'd have with your friendly/local data center or server(s) -- though virtualized, of course, in the form of your own 'instances'or 'VMs'. You hardly would be able to tell the difference accessing a Cloud based virtual server or your local/friendly on-premise IT one...except if your internet response got slow or that particular off-premise Cloud service went down (i.e., the dreaded Cloud service "outage" or service "disruption").

Enter the Public Cloud. What is it? It's not some esoteric, ephemeral vapor substance in the Datasphere ... but really just "internet computing". For the average user (mobile or PC desktop based), it means you don't need to own or need specific, expensive compute and storage hardware -- as well as actual applications -- all in your home, office or company's data center. Hosted Cloud service providers (like Amazon Web Services, Microsoft Azure and many more) have all that necessary hardware and software resources in their data centers (gargantuan ones I might add) for you to access. You can also store your personal emails there (like you already do with Google, Yahoo, Hotmail, et al), along with all types of data including your photos, tunes, videos, TurboTax returns, etc. Hey, maybe even Hillary stored a bunch of her "personal emails" on some Cloud service provider or 'hosted service' (application) site.

How do "they" do it? Using virtualized server clusters and virtualized computing layers/software, users like you and me can create 'an instance' on a Public Cloud site...that can consist of computing (i.e., AWS's "EC2" elastic compute services), storage (i.e., AWS S3 or EBS cloud data storage) or even specific application resources that you've subscribed to (Hosted Services) or have uploaded to your Cloud account. These service or hosted service providers do the heavy lifting by ensuring they have enough processor/CPU/server power, storage and data transfer rates (network/comm) resources to ensure fast, reliable and continuous availability when you long into your account. Obviously, you don't need to worry or care about all of that. You just want the up-time and access to whatever you've subscribed to over the internet. And today, so many hosted services, clouds and compute/storage resources are available to you beyond 'simple' email and database storage.

Most of these 'generic' services are free (like gmail). But most public cloud service providers do charge for your compute and storage/data resources that you use. Typically, everything you do or use can show up on monthly Cloud utility bills. Depending on what you do 'in the Cloud', these bills can get pricey. But the whole idea is that the eliminate the need for small businesses, professional users or larger organizations and corporations from having to 'own' or expand their compute and storage physical hardware resources, etc.

<----- On Premise ------- Combination --- Off Premise/Cloud Service ---->

Finally, you might be wondering what the difference is between a Private and Public Cloud is...or even a Hybrid Cloud. Again, in simplest terms, a private cloud is simply a virtualized server cluster (usually with appropriately sized compute/CPU, cores and physical storage resources) that is located, managed and controlled on site/on premises (i.e., privately) by their owner. The figure above illustrates these different "Cloud types"....

Public Clouds, after all, are subject to internet or data center outages (AWS and Azure famously have had a few) or from being hacked or compromised. Private Clouds give you more control, security and data integrity. Hybrid Clouds? These are virtualized server clusters and software that combine both Private Cloud use with access to a Public Cloud for remote data copy, back-up or Disaster Recovery capabilities -- or even for 'seasonal' needs for supplemental Public Cloud resources or expanded demand/usage. Lately, Hybrid Cloud computing is growing in popularity among SMB (small/medium businesses), ROBOs (retail or regional business office) and larger Enterprise organizations as well.

So there you have it. A napkin drawing of what virtualization is all about and how it's a key -- if not the -- technology approach and architecture enabling Cloud computing. Remember, "in Cyber Space nobody can hear you scream". But with "the Cloud" you're always a Click away from getting more resource help..and all the stored digitized "stuff" you've put there.