Is “Cloud Slice” the Next “VM” for Hands-on Learning?
By Corey Hynes
We are on the cusp of major transformation in the hands-on training industry. Virtual machines revolutionized the delivery of software training. The Lab on Demand Cloud Slice will do the same. It’s time to understand what the Cloud Slice means, where it comes from, and how it works.
The Good Old Days
Anyone who has been in this industry for the last 20-25 years knows what it was like to spend your Sunday morning running complex setup routines to build classrooms in which each student had a unique system, the instructor had shared resources, and everyone played a unique role. Setup took a long time. Resetting a room was impossible, and sharing a room was difficult. I remember one training center using detachable hard drives to swap class images, hoping they did not break between classes.
A New Era
The use of a virtual machine changed all that. Back in 2001, I was part of a small team tasked by what was then Microsoft Learning, to “transform” a few high demand courses to run from the traditional distributed deployment model to leverage virtual machines as a tool for delivery. At the time, we used the newly acquired Connectix Virtual PC and rewrote a few courses to be delivered on virtual machines.
Instructors, training centers, and students loved it, and the move toward virtual machines as the delivery vehicle for software training was in full swing. The following years saw great creativity on the part of content authors. With the ability to build lab specific images, the ability to reset broken labs, and the simple deployment of a file copy, the overall quality of the learning experience went up. Large events adopted this model, and we saw an explosion of new content. Virtual machines made it easy to put very complex software in front of a student.
Companies invested in rich UI experiences, simplifying the ability to access libraries of hundreds of locally installed labs. Our own company reinvented the way labs were delivered at conferences with the creation of our holLaunchPad platform.
But it was not all without challenge. For all the good brought on by the movement to virtual machines, there were some big challenges.
- How do you license them? A VM is portable, easily stolen, so students were getting production software. Evals expire, so they don’t work . . . phone activation? Remember that?
- As people got creative with content, the size and complexity also went up. 32GB RAM on a desktop? These systems are not cheap. Download 200GB the night before a class? Want to keep 5 classes on a student system, now you need lots of disk. The investment in hardware quickly offset the savings in setup time.
- Not everything can run in a VM. Nested virtualization, difficult, slow, buggy until very recently.
The race was on amongst lab developers to create the best “virtual learning environments.” Who could make the smallest, fastest, quickest VMs that were easy to redistribute and just worked well? A handful of authors excelled at this and quickly became the go-to choices for creating virtual environments. I can still list most of these folks by name and the products they specialized in. Some are still around, but most have faded away and moved on to other things.
VMs Go Central
It was inevitable that the required investment in desktop hardware would soon drive the industry to look for more cost-effective hardware sharing solutions. Enter the age of the hosted lab. Companies specialized in putting lab environments on central servers, either in their own datacenters, or at conventions, or in individual companies. Instead of investing in hardware, training companies could rent space with hosting providers and use simple browser based clients to access and manage virtual machines. Farms of servers running hundreds of students all at the same time, from datacenters around the globe.
Two platforms, the holSystems Platform and the Lab on Demand Platform dominated the industry. One in the event and conference space, the other in the commercial training space. It was from the combination of these two platforms that the OneLearn Platform was born.
The One Truth
This entire hosted lab architecture is built on the premise that you are learning some piece of software that is installed and configured in an operating system, and that operating system runs on a piece of hardware you own and manage. The operating system is the container for what you learn. The virtual machine is the delivery vehicle for that container.
Think about it. Exchange. SQL. Linux. Office. Windows. C#. Java. Every one of those technologies has traditionally required a server and/or a workstation with locally installed software.
This entire industry, billing model, service model, and software platforms were built around the notion of replicating something you put in your own office or datacenter using virtual machines and centralized hardware.
Software is Changing
Software today is no longer something you fundamentally own. It is something to which you subscribe. Software as a Service is the notion of paying for software as you use it, only when you use it, and only for how much you use it. Upgrades are automatic. Current versions are assumed. Change is constant. On top of this, much of this software has moved from being desktop based, to being browser based. Our once powerful laptops, with tons of disk space and lots of RAM, can be replaced with web browsers. Office 365. QuickBooks online. Salesforce. CRM. Even real time communications products are based on the notion of having a subscription, an account, and consuming the software in real time, from a browser.
If you no longer have to install, deploy, or secure something in the traditional way, by first doing the same with an operating system, then how do you learn it? Do virtual machines and the platforms we have created to support them still have a role when we no longer need an operating system to contain the software?
The New Container
If the software we consume is no longer contained within the context of an operating system, where do we find it? The answer is simple, we find it in the cloud. Cloud providers are replacing the operating system that we traditionally install on hardware, with an operating system that spans billions of dollars in hardware, hundreds of datacenters, and enough cable to go to the sun and back. Amazon, Azure, Google. They all operate on the premise that you tell them which parts of an operating system your application needs, and they will provide you with exactly that and nothing more.
For example, do you need a website? The cloud will give you a web server and some storage, but hide the “operating system” from you. You don’t have to worry about the tasks of installing, administering, securing, and maintaining what an entire ITPro industry was built around back in the early part of the century.
This changes how we think about hands-on training. If the unit we deploy and manage to put software in front of a user is no longer a VM, then what is it?
The Cloud Slice – The Next Container for Hands-on Learning
A cloud slice is what it sounds like. It involves taking a cloud provider, carving out a small slice of the capacity in a highly-managed way, and serving it to a user so they can learn. The mechanics of this vary from provider to provider. In Google Cloud Platform, it’s a “project.” In Microsoft Azure, it’s a “resource group.” Regardless of the term, the concept is the same. A service provider, upon request, will reach into a cloud provider and give you the pieces you need to learn, in a highly-controlled way, shielding you from all the complexities of getting started.
Cloud Slice in Lab on Demand
The notion of cloud slicing is core to Lab on Demand. We have been building our cloud slice technology and expertise for approximately three years, dating back to 2014, when we ran our first event lab deliveries on Azure. We are focused on delivering this on Azure first, followed by AWS, and then Google Cloud. To that end, we include cloud slicing as part of our Azure Virtual Datacenter strategy.
The Azure Virtual Datacenter
The Azure Virtual Datacenter is a suite of capabilities core to Lab on Demand. They include:
- The ability to replace a Hyper-V host with Azure as the primary hosting fabric.
- The ability to create a resource group, isolated to a specific user, and grant that user access to the resource group in a delegate role. This is cloud slicing.
- The ability to create an on-demand nested Hyper-V hosting server in any Azure datacenter to run traditional labs.
Azure as a Fabric
Below is a screenshot of a normal lab experience using a virtual machine. The difference is that this VM is running in Azure. It is a standard Azure IaaS virtual machine, running in an Azure datacenter of our choosing, and providing the user the same rich HTML based console they have access to when they use our Hyper-V fabric. Only the most observant student would even know the difference. There is no access to the Azure portal or any need to access the portal. This extends the rich experience of Lab on Demand to any Azure regional datacenter, putting users close to the VMs they access, with low latency and high response times. Additionally, users who want to invest more can choose to deploy with larger VM sizes, more memory and more CPUs. Lab on Demand helps control costs by shutting down VMs not in use and deleting everything when the lab is over. Furthermore, it is not limited to only our subscription. Customers can leverage their own subscriptions for 100% control over costs for hosting, or choose one of our managed subscriptions to simplify the business.
Azure as a fabric still fundamentally relies on the old paradigm of the virtual machine as the container. This notion will not go away fully in the near future, especially in the Windows ecosystem. More and more, the virtual machine is playing the role of a creator’s workstation with tools such as Visual Studio.
Azure Cloud Slicing
Here the real power of Lab on Demand comes to bear. A lab in which the user is expected to use a cloud platform such as Azure to learn software might require the user to have a user account, logon to a subscription, and create a series of objects. They typically then administer those objects through the Azure portal or using tools such as those installed on a creator’s workstation.
Lab on Demand understands the notion of a “lab without a virtual machine” in which the content of the lab is a Cloud Slice. A cloud slice might contain any number of the following:
- A set of pre-created user credentials with delegated access.
- A JSON based ARM template describing what is needed to start the lab.
- A PowerShell or CLI script to configure the lab environment.
- A scoring script to run when completed.
Below is a screenshot of a lab in which the user has been given a “slice” of Azure to work in, managed by Lab on Demand.
Interestingly, cloud slices can be combined with Azure as a Fabric to provide 100% cloud only lab experiences; combining a cloud slice with a managed VM creator workstation.
Solving Credit Card Denial of Service
The biggest business risk in the world of cloud slicing is the notion that users can create things which cost a lot of money and have those charges go on the providers’ credit card. Solving this problem is key to providing a robust cloud slicing service. Lab on Demand incorporates a set of technologies that allow customers to safely provide students with slices of Azure without the worry of unexpected charges.
- Students can be given roles ranging from read-only to having the ability to create objects.
- The types AND quantity of objects they can create is tightly controlled, and any offending objects are removed.
- The user is restricted from accessing the cloud slice outside of the context of a managed lab environment.
- Customers can use a OneLearn managed subscription, or their own subscription. For larger activities, slices can be spread over multiple subscriptions.
- Customers and students can choose the region that is best suited for them.
- Costs for completing a lab are fixed and predictable.
This set of controls all but eliminates the need for users to create trial accounts, create Microsoft IDs, redeem Azure Passes, and perform many of the “initial setup” tasks that plague the “Monday Morning” of a typical training class. It brings the getting started experience of cloud on par with the getting started experience of a virtual machine, eliminating the last barrier to broad adoption of cloud slice as the de facto content container in the future.
It might be easy to overlook, but for cloud slicing to work as a solution for training, you have to keep a student coming back. With virtual machines, the only way for a student to access VMs in a lab is to launch the lab and use our managed interface. With cloud slices, the same must be true. You must not be able to access your slice outside of the Lab on Demand interface. Not only does this ensure that the student cannot avoid the proactive cost controls that are included in cloud slicing, but it ensures that training managers capture the appropriate level of consumption data.
One of the methods to achieve this is a dynamic cloud credential. This credential is issued by the Lab on Demand platform, scoped to only a cloud slice, and only active when the user is completing the lab using the Lab on Demand interface.
Make no mistake, cloud slicing is in its infancy, and Learn on Demand Systems is a pioneer, innovating and driving this concept forward. As cloud slicing matures, virtual machines will remain the standard container for hands-on learning. Linux and Windows continue to produce server class operating systems, and users still install desktop software.
With the investments that Microsoft is making in the network of Azure Datacenters, it would be foolhardy not to take advantage of this to offer our customers; yet, another option for how they would like labs delivered.
On-Demand Hyper-V represents the ability for Lab on Demand to deploy itself into any Azure Datacenter, ingest lab content, and then make that content available in traditional virtual machine hosting, just as it is in our own managed datacenters with Hyper-V running on Azure.
This is not for everyone. Nested Hyper-V on Azure is not yet released, and the costs and performance impacts are not known. It is anticipated to be two things. Very fast and very expensive. This capability unlocks the last doors to truly being able to deliver hands-on content anywhere, to anyone. It will now be possible to deliver any lab hosted today within Lab on Demand, natively, from any Azure Datacenter supporting the new nested class of virtual machine. Lab on Demand is being modified to instantiate Hyper-V hosts on the fly, provision labs to them, and then connect users to those lab instances, with the customer being able to specify which Azure datacenters host the lab.
We are actively building this capability and expect to release it in parallel with the release of nesting in Azure.
Evolving the UX to Match the Technology
If you apply the same principals of UI design to cloud slices as you do to virtual machines, you are bound to have a bad experience. Cloud slice relies not on a desktop, but on an often uncontrolled, unmanaged browser window. Instead of wrapping around a desktop, you really just want to get out of the way.
That’s where CloudUI comes in. CloudUI is a variant of the Lab on Demand user interface that is specifically tailored for cloud slice experiences. CloudUI is in development, and while we don’t have a release date, we can say a few things with certainty:
- CloudUI will feature the same IDL experience that VM based labs have today.
- CloudUI will have the same support functions.
- CloudUI will be non-intrusive and take up a minimum screen footprint.
- CloudUI will have “hooks” into the cloud slice, allowing some functions which require elevation of privilege beyond that of the user to be completed by interacting with CloudUI.
- CloudUI will be customizable.
Are You Ready?
The industry is ready to transform. This transformation will take place over the next two years as cloud capability matures, and cloud training provides innovation and introduces features and capabilities designed to simplify and commoditize learning cloud.
Learn on Demand Systems is ready. We have invested tremendously in this area and are committed to leading the industry with the richest and most relevant features, leading the industry in patterns and design of cloud slice training, and providing the same level of “simplification of consumption” that we have brought to the world of virtual machines.