If one was to believe the statistics presented by some vendors, you’d be forgiven for thinking the vast majority of organisations are already running extensive levels of virtualisation across their production environments. In reality, however, many organisations have actually only virtualised the low hanging fruit or are still is the process of piloting their virtual infrastructures.  One of the first barriers to virtualisation roll-out is often a lack of understanding of the differences between a Proof of Concept (POC) and a Pilot.

A POC is typically a partial and often standalone solution used to establish that a concept or system satisfies some aspect of the requirements for the complete solution. The proof of concept implementation will not affect business operational data although it may integrate with existing business systems to some extent. In many environments pilots are actually more like POCs, but unfortunately the pressure to reduce cost and rapidly deliver new services has forced the POC infrastructure to become integrated with production bypassing the wider scope of planning that should be undertaken.

The purpose of a pilot project is to test, usually in a production environment, whether the system is working as it was designed while limiting business exposure. The transition from running a pilot to virtualising the wider environment shouldn’t be a leap of faith because sufficient design, development and planning should have been undertaken, and here lies another barrier. The design and planning required for the pilot should in effect be treated exactly the same as deploying the production environment. When a successful pilot has been completed, more often than not, it will simply be rebadged as production and expand accordingly.

At a high level, technology and operations are both key aspects that need to be planned and tested carefully in order to ensure the transition from pilot to production is a strategic success. Virtualisation pilots often tend to be very technology orientated when, in fact, there should be just as much focus on the operational elements associated with successfully managing the virtual infrastructure. These operational processes, such as change management, capacity planning, virtual machine (VM) lifecycle management and chargeback, have to be in place during the pilot and have the ability to scale into production. From a technology point of view, organisations must look beyond the hypervisor and address all the components of the infrastructure that virtualisation has an impact on; such as networks, backup, storage and disaster recovery. Many pilots simply test the smaller, easy to virtualise, candidates and only focus on performance at the application and operating system layer which often produces unrealistic results.

Organisations must have test strategies that include the full range of potential configurations. This will ensure the infrastructure has the capacity to scale in order to meet the demands of larger workloads as and when they are virtualised. The classic example of this being storage input/output, where cheaper storage technologies are implemented and VMs perform as expected during the pilot but performance can decline significantly once the infrastructure is loaded or VMs with heavier workloads are introduced.

With an operational strategy and ‘bigger picture’ approach to virtualisation technology planning organisations won’t need to making a leap of faith, they can just cross the bridge to a better place.

-Tom Brand, GlassHouse Technologies (UK) Practice Lead

With tightened budgets, businesses are constantly looking for way to see a rapid return on investment (ROI). This has increased interest in the adoption of pay-as-you-go cloud services and virtualisation technologies where the ROI can be very attractive. In recent months, Tom Brand, virtualisation practice lead at GlassHouse Technologies has found himself frequently answering the question: “What is the difference between cloud computing and virtualisation?” In Tom’s latest blog post he gives his view…

In order to answer this question it is first important to clarify what the two terms, cloud computing and virtualisation, actually mean: According to the National Institute of Standards and Technology (NIST), cloud computing is a pay-per-use model for enabling available, convenient, on-demand network access to a shared pool of configurable computing resources (e.g., networks, servers, storage, applications, services) that can be rapidly provisioned and released with minimal management effort or service provider interaction.

Virtualisation is a technique used to abstract the physical characteristics of computing resources from the systems, applications or end users that interact with those resources. Virtualisation technologies typically let a single resource (such as a server, an operating system, an application, or storage device) appear as multiple logical resources; or makes multiple physical resources (such as storage devices or servers) appear as a single logical resource.

In analysing the definitions above, trying to compare cloud computing and virtualisation is similar to comparing a car to an engine respectively. A car is a complex system including parts, interfaces, inputs and an engine that function as one to provide the best and most efficient drive for the owner. Like the car, cloud computing (public or private) is essentially the coming together of technologies, operational processes and financial models to provide organisational flexibility with optimum cost-efficiency. Continuing with the automotive analogy, the engine of a car is a core component because without the engine, the car won’t move, regardless of whether there are any seats in it. With cloud computing, virtualisation is the core component enabling the majority of characteristics required to make any cloud computing model work.

Going one stage further, you can compare the cloud to a cost-effective metered taxi service, always at your disposal. You now have a range of highly efficient vehicles that can be requested whenever you need to travel. They are operated and maintained by someone else and you only pay for the length of the journey (paying-as-you-go) with the ability to get out whenever you like. Although cloud computing encompasses a large range of compute services, typically labelled as either infrastructure (IaaS) or applications (SaaS and PaaS), they all fit within the taxi model. End users only pay for the services or resources they use, the service and ability to provision is always available. Users have the flexibility to request different services however the underlying infrastructure is not their concern and cannot be modified. In summary, virtualisation improves IT efficiency - enabling traditional computing with fewer resources; whereas cloud computing improves IT effectiveness - empowering more people to build services with more flexibility and fewer experts. When implemented accurately, both technologies can provide attractive an ROI for the IT department.

What is “Cloud Computing”, What’s needed to get there and how does VMware fit in?

By Rob Zylowski Director, Services and Director, IP GlassHouse Technologies, Inc.

VMWorld 2009

Wiki - Cloud computing is a style of computing in which dynamically scalable and often virtualized resources are provided as a service over the Internet.  Users need not have knowledge of, expertise in, or control over the technology infrastructure in the “Cloud” that supports them.

I’m not sure everyone would agree with this definition.  The key term here that I find problematic is “resources”.  It seems that most people think of the “Cloud” more in terms of providing applications than infrastructure (though infrastructure can be supported) but “resources” indicates only rather than applications.

Consistent Themes

Doing a little research on the NET, it seems that several themes arise again and again with respect to “Cloud Computing.”

• “ I don’t care what’s up there as long as it works”
• Abstraction from the underlying technologies
• It’s provided as a service billed as a utility based on usage
• Closely aligned with Web 2.0 and Software as a Service (SaaS)
• May include HaaS — Hardware as a Service where the platform is what is being purchased as the “application.”
•  

What about VMware

Of course I am writing this at VMWorld 2009 and must consider VMware’s definition of the “Cloud’.  From a platform perspective most current “Cloud” offerings are built on open source infrastructure like Xen and KVM most likely due to price pressure.  Carl Eschenbach announced at the show this week that he believes VMware will compete strongly in the public space however it seems to me that VMware will be strongest within the Enterprise or what’s known as the “Private Cloud.”

What does it take to build a private cloud?  If we use the amalgamation of definitions found above the “Private Cloud” would need to appear to its end users as an abstracted service where they could use/purchase applications or business functionality paying only as they use these services.  Of course, that definition would apply only to end users in our IT world.  If “our” users were application developers or application owners then the abstraction might be to purchase application support services such as web, database, storage and compute; again as a utility.  As these are VMware’s primary clients the portion of the “Private Cloud” likely to be filled by VMware is only the platform services.

Service Provider Model

As stated above one critical aspect of the “Private Cloud” is that it must be provided as a service.  GlassHouse has been working with Enterprise IT groups for years working on ways to migrate their services to a “Service Provider Model” and I can attest to the fact that while extremely beneficial in many ways this is not an easy transition.  Most companies have built the financial policies and processes governing the IT function over many years.  These processes are ingrained into the corporate culture and very difficult to change.  The types of challenges I am thinking about here are:

• Project based funding
• Departmental or Business Unit ownership
• Yearly allocation of costs based on very high level budgetary numbers using a metric like number of desktops or number of employees
• No real understanding of the underlying costs associated with infrastructure or applications at a micro level

In order to successfully build a “Private Cloud” an organization will need to be able to function as service provider.  They must define their services and service levels in terms that their clients will understand so that the clients can make informed decisions on what service level they require.  In order to support the service definition the IT group must be able to predict and track their costs at a level which allows them to appropriately and fully cost each service.  With this level of information it will be possible to provide services on a utility or usage basis to the end users.

The Bottom Line

This is a VMworld 2009 brief so there is not enough space for me to fully expand on what work must really be accomplished to create a “Private Cloud” within your organization.  While “Cloud Computing” is currently much to do with marketing and hype there is certainly a high degree of optimism that this concept will enable us to simplify corporate computing in the future. 

One thing that is certain is that VMware and others are creating a significant buzz around this topic.  The buzz is driving development on a massive scale so that with the tools VMware and its partner community are bring in to the market Enterprises will eventually be able to deliver their services as true service providers.

However, what I wanted to get across here is that before this hype reaches any real fruition there is a significant amount of intense work to be performed by ISV’s and corporate IT groups to attain the ideal where users can go to a web site, then purchase and provision the services they need for only the time they need them.  But as they say Rome was not built in a day!

VKernel Capacity Analyzer
http://www.vkernel.com/products/CapacityAnalyzer
Product Data Sheet available here.
Pricing is $199/socket. VKernel is a certified virtual appliance, so no Windows licensing is necessary.
Evaluation licensing is a 14-day license, and will manage one vCenter or 3 hosts. In my eval, it is pointed to a vCenter, and sees the 8-host cluster with no problem.
This tool also interacts with vCenter to gather its information. However, what it does with the information seems to be a bit unique. As you might expect from a product called Capacity Analyzer, performance is monitored from a capacity standpoint, and not a resource standpoint. You might think those seem the same, but they are different enough to mention it. Where as other tools have a lot of views centered around performance, and you can get to capacity, VKernel is centered around capacity, with the performance statistics as a secondary piece of the puzzle. The main overview screen lists highlights of: current and future bottlenecks, capacity availability, and top resource consumers. Then there is the expanded “Capacity Availability” view that just shows you what you have available. You can drill down at any level to find out more of the raw performance and utilization data used to create the capacity categories. You can even create your own groups of VMs, and view capacity data based on your custom grouping. At every level from Virtual Center to Cluster to Host, the same capacity views are available. This makes it incredibly easy to see how many VMs will fit in your environment.
Summary
So - there are five tools to consider when monitoring your VDI, or really any virtual environment. These are not specific recommendations, just some high level information in my opinion. As I’ve said previously, these are not the only tools, and you may find others that you like better. If you have other preferred monitoring / capacity management tools, please leave a comment and share your experience.

-Bruce Heavner, GlassHouse Senior Consultant

Tek-Tools Virtual Profiler
http://www.tek-tools.com/virtual/overview.php
Product technical brief available here. (Registration required)
Pricing is per host, with a required base framework as well. A starter pack with the framework and 3 host licenses retails for $4495, with additional hosts at $995. This tool runs on top of windows, so count that in your cost estimates.
The evaluation license will monitor two hosts. You can point it to a vCenter, and it will only monitor two hosts - but you can choose which ones are included.
This tool works by pulling events and data from virtual center (and the hosts) and wrapping some meaningful statistics around it. Virtual Profiler is an enterprise-class monitoring tool, with enterprise-level features. See the technical brief (above) for the whole feature set. The performance monitoring and other features are great, with very detailed information available for an array of components. Capacity planning is where this product gets interesting. Rather than run a report that shows you utilization graphs, this tool continually monitors the resource usage, and has a running counter of how many additional VMs can fit, based on average usage. This is easily viewed from a cluster or individual host level. There is also a nifty load simulator, where you can put in an expected number of VMs and disk space, and it will tell you how full the cluster will be. Keep in touch for my findings on VKernel Capacity Analyzer.

-Bruce Heavner, GlassHouse Senior Consultant

Veeam Monitor
http://www.veeam.com/vmware-esx-monitoring.html
Product datasheet available here.
Veeam Monitor also has two versions - a free (yes FREE!) version with full monitoring capabilities, but it does not include capacity planning. The non-free version is $250/socket. This tool also runs on top of Windows Server, so include a Windows server cost in your estimating.
The evaluation displays as an “Evaluation License” but there is no mention of host or time restrictions. I am unclear what limitations are involved with this license.
This tool is a client-server application, and requires a windows OS to run on. The pieces can be installed separately so the client can run on your own desktop while the server is safe in the data center. The tool works by interfacing with vCenter - your first task once setting up the application - and making sense of the many logs and events therein. Multiple vCenters can be added, with limited federation of data available. The monitoring pieces are pretty standard, but seem to be a lot cleaner and simply laid out then some others. This is probably a result of the client-server model, since they can make the front-end look however they like and is not dependent on sometimes clunky web browsers. Capacity planning is report based, and requires Excel 2007 on the client station (or server, if that’s where the client is running). The reports seem to be on-demand only, with no scheduled report capability. Running a Trend Report creates a Veeam Report Viewer file. That file gets processed through the Report Viewer common to most Veeam applications, and this is where Excel 2007 is used to generate a spreadsheet full of graphs and raw data. Like vFoglight, the resulting graphs show you capacity used and VMs using the capacity, but there is no mention of how much VM capacity is left. Regardless, this is a very well-done tool, and provides ample amount of data for you to plan for virtual desktop expansion. Are you ready for Tek-Tools Virtual Profiler?

-Bruce Heavner, GlassHouse Senior Consultant

Quest vFoglight
http://www.vizioncore.com/products/vFoglight
You can get their product brief PDF here.
vFoglight is priced per CPU, with two versions. SE is $499/socket, Pro is $699/socket. There is also a vEssentials package where you can pick your three favorite vizioncore products (hello vRanger Pro!), for $899/socket. These prices include 1 year of maintenance - 3 year packages are also available. (Also - vFoglight runs on top of a Windows Server installation, so include that in your TCO calculations)
The evaluation license is a Pro version, with a 30-day expiration.
I won’t regurgitate the marketing info or the differences between the versions - you can read just as well as I can. Instead, I’ll get to my observations. vFoglight is truly an enterprise-class monitoring tool, with some serious requirements, but also the beef to make them worthwhile. In a single-tier 32-bit installation four cores (1 quad core, 2 dual cores, or 4 single cores) and 4GB of memory are listed as minimum requirements. (Keep in mind that Windows Server 2003, Standard Edition has a 4GB memory maximum) A component called the “Cartridge for VMware” interfaces with vCenter to pull data into the vFoglight database. Once it’s installed and running, all management is browser-based. The monitoring is as robust and thorough as you’d expect an enterprise level monitoring tool to be. I’d recommend watching their demo to get the official feature set. Capacity planning is report-based, and can be run on-demand or scheduled periods. The ESX Server Capacity by Virtual Center report template seems to be the best match, and generates a PDF with average CPU and Memory usage for each host in the VC, along with how many running VMs used that capacity. You’ll have to do your own math to determine how many more VMs can be placed before host capacity runs out. There is no report template for storage capacity, although there is an extensive reporting engine that should allow you to make one of your own. Stay tuned for Veeam Monitor.

-Bruce Heavner, GlassHouse Senior Consultant

The past few days I have been setting up some evaluations of various tools used to monitor a VMware View implementation. I thought I’d share what I’ve seen so far. These are not full technical reviews of the products, in fact I’ve linked to quite a bit of external content that I’ve found, but rather these are some high-level observations of the products. Please note - this is not an exhaustive list of available monitoring tools, this is just the few that my current client chose to look at. These happen to be in order of IP address of my evaluation servers - I don’t have a preference (yet). As such, this is not a recommendation of any product above another - these are just some options for you to consider.Note - the following series of my blogs are list prices, and do not factor in volume discounts of any kind. Socket is defined as physical processor chip, not cores.

vmSight Application Profiler

http://www.vmsight.com

A detailed product review with some screen shots is available here. (An online product brief is not currently available, as the website is being restructured due to a recent acquisition by Liquidware Labs)

vmSight is priced at $50/user, list price.

Evaluation licensing is for a single host.

You have to register with their website in order to gain access to the download, which will come in the form of a link in an email. What you end up downloading is a certified virtual appliance, with minimal configuration needed. You should also be able to download directly into vCenter, but my hosts are behind restrictive firewalls so I was unable to do so. The appliance expects two network connections - one to the normal network, and one that sits on a vswitch configured in promiscuous mode to sniff the traffic. Because of this, each ESX host in the environment must have one of these appliances, which come in two flavors. There is a primary ‘vmSight Center’ that is the management server, where all licensing is, and then all other hosts get a ‘Monitoring Station’ that reports back to the primary. There is an additional component, their patented “Connector ID” technology that goes as far towards the end-users as you like. It can go into the VDI desktops, and down to the thin-client or physical desktop level. This gives you end-to-end monitoring from a user perspective. All management, aside from initial installation, is browser-based. If performance monitoring was all that you needed, I think this is a slick end-to-end product, and could be very powerful in maintaining the user experience. However, with the exception of some highly-customized reporting, it does not have any specific trending or capacity planning functions. Next up, Quest vFoglight.

-Bruce Heavner, GlassHouse Senior Consultant

Since this is my first blog entry relating to VDI, it makes sense to me that I should go over the basics and talk about what VDI is. This is especially important if you’ve never heard of VDI before, and for those of you who are not new to it, this can serve as a review. Let’s start with a definition that I found at http://en.wikipedia.org/wiki/Desktop_Virtualization:

“Virtual desktop infrastructure (VDI) is a server-centric computing model that borrows from the traditional thin-client model but is designed to give system administrators and end-users the best of both worlds: the ability to host and centrally manage desktop virtual machines in the data center while giving end users a full PC desktop experience.”
Ok, that’s great – but what does all that mean to you? Before we get too far into the definition, let me say that VDI, at least in this blog, is a generic acronym. Some other vendors (cough – Microsoft – cough) use the term HVD, for Hosted Virtual Desktops. However, this blog as a whole will attempt to stay vendor-independent and use the generic terms, unless we’re talking specifically about a product. Now that we’ve got that out of the way, let’s clarify the definition by talking about some of the benefits of VDI and what you will get out of it. In today’s computing environment, many companies are concerned with the sometimes delicate balance of cost vs. security. On the cost side of things, the current state of the economy dictates that costs stay low and spending stays down, or the “dollar” cost. Another less-common cost consideration is what I’m calling the “experience” value – the relative ease-of-use of a given solution for a typical end-user. We want this value to be high, indicating a highly usable system. Then of course there’s the security side of the equation. We all want to keep our data private and easily accessible, but only to those that need it. So how can you have a solution that:
a) Keeps the “dollar” costs down,
b) Maximizes the “experience” value,
c) And maintains data security?
If you put enough money into security (high “dollar” cost), you’ll get a generally secure system, but high security “dollars” do not always equal high security. Sometimes that high security system also comes with a hit on the “experience” value, since secure systems can be difficult to use. Or, if you design a system that’s highly usable, it may not be very secure or cost effective. And of course if you don’t spend any money at all, nothing will change. Data won’t be any more secure, and Users’ experience won’t change.

I believe that VDI can address these three areas of concern. VDI can help reduce costs by reducing management, administrative, and resource overhead, while at the same time increasing the user’s experience and securing data inside the datacenter. How does this work? Allow me to elaborate. With traditional computing, a user has a desktop or laptop computer, which uses local resources (CPU & Memory) to run applications. These applications are installed either by a person or a process, requiring time or software distribution management. Data is pulled in across the network to be manipulated locally, and then saved back on a file server. Everybody (end-users) knows how this works because it’s been this way for awhile. To keep the experience value high with any new solution, we need to be able to keep this same basic application interface concept.
Enter VDI…
Basically speaking, VDI takes the user’s desktop and moves it into the data center. When done right, resources become better utilized, applications are managed efficiently, and data never leaves the friendly confines of the datacenter. Properly utilized resources means better efficiency and reduced TCO, properly managed applications means happy IT administrators, and a properly configured virtual desktop environment means users can continue to use applications the way they know.
So, with these benefits and end results in mind, our modified VDI definition will read like this:

“Virtual Desktop Infrastructure (VDI) is a new way of managing the end-user experience. With a mind toward maintaining (or increasing) the usability of the environment, centralized computing resources become better utilized, and data is more securely protected inside the datacenter. Applications are easily managed because they also exist only in the datacenter, and not spread across a corporate campus. TCO goes down, IT and end-user satisfaction goes up.”

Keep in mind we’re still in the basics here, and just defining this blog topic. There is a lot more to VDI than these three basic benefits. I might even expand on that definition some more, as we get deeper into some other VDI topics. VDI in general is a young technology, with some major vendors getting into this space in the past two years. This definition, and how we use it, will very likely change as the solution matures. We haven’t talked at all about any specific products (and there are a few). We haven’t discussed any particular pros and cons (like thin-client and offline access, which will come with the specific products). We haven’t even thought about system requirements (like solid datacenter infrastructure to support these virtual desktops, which will also come with the specific products).

We have a lot to cover, so stay tuned as the world of VDI becomes clearer to all of us.

-Bruce Heavner, GlassHouse Senior Consultant