Post on 11-May-2018
The Great Debate: Persistent vs. Non-Persistent Virtual Desktops Persistent desktops that sustain all user and IT customizations? Or non-persistent desktops
that revert back to a pristine state after each use? This has been one of the great debates in
Virtual Desktop Infrastructure (VDI). Is it even relevant anymore? Find out in this white paper.
Introduction
The debate between persistent and non-persistent virtual desktops has been hotly contested ever since Virtual Desktop Infrastructure (VDI) became a viable alternative to physical PCs. This white paper takes a look at the history and evolution of the VDI market to understand why persistent and non-persistent desktop models exist.
First, we will define each model, explain the use cases, and discuss the technologies available to implement both desktop types.
Then, we will debunk some of the common myths and misconceptions, and outline the pros and cons of each model.
Throughout, you will benefit from actionable insights that can be used when architecting your VDI environment, and learn from the real-life customer examples that have been included for reference.
Evolution of Persistent and Non-Persistent Desktops
To understand why there are even two different types of virtual desktop models, it is important
to understand how the VDI market and the technology have evolved over time.
The Persistent Desktop Era
Early adopters of VDI chose to virtualize desktops primarily for the security, mobility, and
anytime/anywhere access benefits. The easiest and fastest way to realize these benefits was to
simply move PCs from the edge into the data center.
The first virtual desktops, then, were “full clones” – full-sized virtual machine (VM) copies of
physical desktops. These full clones were assigned at login to specific users. Every time users
would login, they would access the same VMs. Full clones were also persistent – every change
made to desktops by an end user or an IT administrator was saved.
Full clone desktops were managed exactly the same as physical PCs – using manual methods, or
agent-based PC lifecycle management software such as Microsoft System Center Configuration
Manager (SCCM), Norton Ghost, KACE, LANdesk, Big Fix, etc.
Full-sized, thick-provisioned VMs proved to be quick to implement. However, they had two
significant problems that limited broad VDI adoption:
1. Management. The management challenges associated with physical PCs still remained.
Like physical PCs, persistent desktops would deteriorate over time, as IT and end users
both made changes to the C: drive and Windows registry. As desktops deviated from
each other, Windows and application updates wouldn’t always “take,” resulting in the
same 5-15% patch failure rate and costly service desk escalations as PCs. Provisioning
new desktops and delivering new applications was only marginally better than PCs.
2. Storage. Persistent desktops made VDI unaffordable to the mass market due to the
high cost of shared storage. The 40 GB local disk drive in a PC is cheap. That same 40
GB desktop hosted on SAN storage in the datacenter is expensive – roughly 10X the cost
of PC storage.
Because of these issues, only organizations with large IT staffs and large budgets (e.g. Wall
Street financial firms) were able to deploy persistent VDI at any scale.
The Non-Persistent Desktop Era
In an attempt to solve the storage and management challenges, Citrix and VMware introduced
the concept of non-persistent, floating desktops. With this model, virtual desktops would
reside in pools. When a user needed to access a desktop, a VM would be pulled out of the pool
and assigned to the user. When the user was finished using the desktop, it would be returned
to the pool and all changes made by the user would be thrown away – hence the term “non-
persistent desktop.”
It was hoped that this model would address the two challenges of persistent desktops:
1. Management. By resetting desktops back to their pristine state after each use and
using shared image technology to provision desktops, management is greatly simplified.
Disk data is streamed dynamically and in real time from a single shared image, providing
machine image consistency and enabling large pools of desktops to completely change
their configuration, applications, and even OS in the time it takes them to reboot.
Because all desktops are bitwise compatible, the 5-15% patch failure rate vanishes, and
support costs are greatly reduced.
2. Storage. Since no desktop changes would be saved, block-based image sharing
technology could be applied to reduce storage.
VMware already owned an image
sharing technology called Linked
Clones, which had been used for many
years to reduce the hardware costs of
using VMware Workstation.
VMware integrated Linked Clones with
VMware View Composer and
introduced it to customers as the
preferred method of provisioning
desktops for VMware Horizon View.
With Linked Clones, the cloned desktop
uses the virtual disk of the parent
virtual machine from which it was
cloned. This dramatically reduces the
time needed to set up a virtual
machine, and the amount of disk space
the clone uses.
When the desktop is created as a
Linked Clone, a Delta Disk is "linked" to
the replica disk. The Delta Disk is where
all changes or differences between the parent VM and the cloned desktop are stored.
Citrix also owned shared image
technology. Citrix Provisioning Services
(PVS) has been used for years to manage
terminal server farms. It was a natural
step for Citrix to make PVS a provisioning
option for Citrix XenDesktop.
Like View Composer and Linked Clones,
PVS shares a single shared disk image
(vDisk) rather than copying full images to
individual machines. This enables
organizations to reduce the number of
disk images that they manage.
PVS uses a cache file to capture the
changes to the underlying OS when
running in "Standard" mode. There is
also a "Private" mode that enables you
to save changes to a vDisk in the disk
itself.
Generally, administrators use two vDisk
images for a build. One vDisk image is in
standard mode streaming to the desktops. The other vDisk image is updated for new revisions.
The two vDisk images either have to be kept in sync, or clones need to be made each time an
update is performed. For VDI, this process can be tedious compared to other solutions, but the
ability to use either local storage or SAN storage is a huge benefit.
Citrix also offers a similar image sharing technology called Machine Creation Services (MCS) for
customers who want to provision only VDI desktops.
Early VDI adopters were eager to implement the non-persistent virtual desktop model because
of its potential to reduce storage costs and simplify image management. It proved adequate for
users who required the same applications and who didn’t mind that their customizations were
lost after each use. Call centers, kiosks, and static labs that rarely changed were ideal use
cases.
However, these use cases accounted for only 5-7% of all desktops. For end users who expected
their virtual desktops to act like PCs and preserve settings and user-installed applications, and
for IT administrators who needed to frequently reconfigure desktops to meet the needs of
different departments, use cases, projects, and business units, non-persistent desktops proved
inadequate.
The “Non-Persistent Desktop Plus Point Tool” Era
To address the need for desktop customization and personalization while retaining the efficient
resource utilization and single image patching benefits of non-persistent desktops, Citrix and
VMware embarked on strategies to integrate and/or acquire third-party solutions. By adding
profile management, user virtualization, and application virtualization technologies on top of
their existing non-persistent desktop models, customers would – in theory – get everything
they needed.
User virtualization tools would capture and
restore the user-installed applications, IT-
installed one-off applications, and application
plug-ins and add-ins that live outside of a user
profile, and therefore cannot be captured by
profile management technology. Citrix acquired
RingCube, whose technology became Citrix
Personal vDisk.
Profile management tools would capture and
restore the user settings and customizations that
were lost when non-persistent desktops were
reset. VMware acquired RTO Software, whose
technology became VMware View Persona.
Citrix acquired Sepago, whose technology
became Citrix User Profile Management.
Application virtualization tools would enable
different applications to be delivered to desktops
based on the needs of each user by separating
the applications from the shared
base/parent/master image. VMware acquired
Thinstall, whose technology became VMware
ThinApp. Citrix recommends Microsoft App-V
through its strategic relationship with Microsoft.
Image management and storage optimization
tools would enable desktops to be provisioned
from a common base image to reduce storage
costs and simplify patch management. VMware
combined Linked Clones with VMware View
Composer. Citrix integrated PVS with Citrix XenDesktop, and later launched Machine
Creation Services (MCS) to offer a simpler provisioning solution designed specifically for
VDI.
This model became the most common desktop virtualization model, despite being fraught with
complexity and limitations that will be discussed later.
The Layering Era
More recently, a new, unified solution for
provisioning and managing virtual desktops and
applications has taken hold, replacing the need
for non-persistent desktops managed by point
tools. Hundreds of organizations are now using
desktop layering technology from vendors like
Unidesk to provision persistent and non-
persistent desktops from a single shared
Windows OS layer, any number of shared
Application layers, and unique Personalization
(User) layers.
With desktop layering technology, the old debate of persistent vs. non-persistent desktops
becomes obsolete. Layered desktops can be persistent or non-persistent. Yet, both types of
desktops are as storage-efficient
as non-persistent desktops
provisioned with block-based
image sharing technology.
Layered desktops are assigned to
specific users or roles at first login.
Desktops share a single instance
of the Windows OS layer and all
Application layers, so patches and
updates are applied only once and
storage utilization is greatly
reduced.
Because layering is based on file
system and registry (C:)
virtualization – not block-based
image cloning – all desktop
customizations, including applications and plug-ins that are installed by IT administrators and
end users, are sustained. Base layers can be patched as often as IT wants without affecting the
upper layers.
With layering, the only difference between persistent and non-persistent desktops is whether
the Personalization layer is reset after each use. If a desktop is configured as persistent, the
Personalization layer is left intact. If a desktop is configured as non-persistent, the
Personalization layer is wiped clean. The decision of whether to deploy persistent and non-
persistent desktops can now be based solely on use case, since both require minimal storage,
and both are built using common layers that only need to be patched once.
Layering software like Unidesk integrates with the leading brokers and the leading hypervisor
so that the brokers are still used for virtual desktop access, but provisioning, application
virtualization, personalization, and storage optimization are handled by the layering console.
Debunking Myths and Misconceptions
Vendors, bloggers, and analysts have made the persistent/non-persistent VDI discussion more
confusing by using similar terms to describe different approaches. Before we dive deeper on
technology, use cases, and pros and cons of the two most common models, let’s clear up some
common misconceptions.
VMware’s Persistent Disk is the Same as a Persistent Desktop
Myth. VMware’s decision to rename the
“User Data Disk” to the “Persistent Disk”
has made some IT administrators think
they are deploying persistent virtual
desktops capable of capturing all
customizations. This is not the case.
When VMware Linked Clones and View
Composer are used to provision desktops
for VMware Horizon View, there are
actually four different types of disks:
Delta Disk: The Delta Disk is where
all changes to a desktop are stored. The Delta Disk grows over time until the desktop is
recomposed (usually to apply a Windows patch or application update). At this point, the
Delta Disk becomes invalid, and all changes are lost.
Disposable Disk: The Disposable Disk contains all page and temporary files. It gets
cleared every time the desktop is rebooted.
Internal Disk: The Internal Disk holds the Windows Active Directory machine password,
which ensures that the desktop does not lose its domain trust when the desktop is
recomposed.
Persistent Disk: The Persistent Disk is optional and is assigned at the pool level. It is
typically used to redirect file saves to a D: drive so that data can be preserved and
shared. Files redirected to the Persistent Disk are not lost during a desktop recompose
or reboot. It is also common for administrators to redirect Windows profiles to the
Persistent Disk. User-installed applications, however, don’t typically work with the
Persistent Disk, since too many Windows apps write to the registry and hard-code C:
directory names. The Persistent Disk also doesn’t capture other information that does
not take kindly to redirection – computer name, MAC address, volume serial number,
and disk signature. Without these, strange application behavior is likely to occur.
It is important to note that data and profiles will only survive a desktop recompose operation if
the user or administrator explicitly saves them on the Persistent Disk. This is why complaints
from end users about applications, settings, and files being lost when IT recomposes desktops
are so common. To avoid this end user backlash, IT organizations often delay desktop
recomposes. This results in three problems:
1. The Delta Disk grows in size, using up more storage.
2. Windows patches aren’t applied as often as they should.
3. Updating applications that are bundled as part of the Windows image takes much
longer than end users would like.
Persistent desktops – especially persistent desktops managed with layering technology – don’t
have these issues.
Full-Sized, Thick Clones Are the Only Way to Deploy Persistent Desktops
Myth. Before the advent of layering technology, the only way to create persistent desktops
was to allocate full-sized VMs, each with their own copies of Windows and applications. This is
why many people use “persistent desktops,” “1:1 desktops,” “full clones,” and “thick-
provisioned desktops” as interchangeable terms.
Technology has advanced, however. Persistent desktops can now be created using “layers” –
containers of files and registry keys that can be merged into a virtual C: file system. In the
world of layering, the Windows OS layer and Application layers are shared across all desktops –
persistent and non-persistent alike. Persistent desktops are distinguished from non-persistent
desktops by a Personalization layer (aka User layer) that captures all customizations. Unlike the
VMware Persistent Disk, the Personalization layer remains unchanged when underlying
Windows and Application layers are patched and the desktops are rebuilt.
State of Ohio Department of Developmental Disabilities is proof that 1,400 persistent
desktops can be created on a small storage footprint, and be managed from a shared set of
single instance, “patch-once” layers.
Persistent Desktops Require a Lot of Storage
Myth. Before layering technology was introduced, this was correct. Full clones were the only
way to provision true persistent desktops without the fear of losing user customizations every
time the desktops were patched. Having to allocate 40 GB or more of disk space for every
clone to store its own copy of Windows and applications often made VDI cost-prohibitive.
Fortunately, this is no longer true. Desktop layering technology enables persistent desktops to
be created on 70-80% less storage capacity than full clone virtual desktops. If you combine
layering with next-generation flash-enabled disk arrays that offer in-line de-duplication and
compression capabilities, the storage reduction can exceed 90%.
Construction company Egan Co. in the Twin Cities of Minnesota is deploying persistent virtual
desktops to all of its knowledge workers using layering technology and flash arrays with de-
duplication and compression. Egan’s CIO reports that their first 200 desktops are only using
500 GBs of storage – for an average of 2.5 GB per desktop.
Persistent or Non-Persistent Desktops is an All or Nothing Decision
Myth. Again, this used to be the case. When the built-in management tools that come with
Citrix XenDesktop and VMware Horizon View were the only options for provisioning desktops,
you had to choose different technologies and management techniques to deploy persistent and
non-persistent desktops and manage images. Often, this setting had to be applied to an entire
pool of desktops. If a non-persistent VDI architecture was deployed and later deemed
inappropriate, the VDI project became a complete rip-and-replace.
With VDI managed by layering software, the underlying technology for non-persistent and
persistent desktops is the same. With Unidesk, for example, you simply select the desktop type
when you create a desktop. If you want to change the type, just select the other option.
Sunrise Health is deploying VDI to 3,000 users. Its desktops are almost equally split between
non-persistent (nursing stations, kiosks) and persistent (clinicians, staff) using the same layering
foundation.
Here is how four educational institutions are implementing non-persistent and persistent
virtual desktops using the same layering technology:
Colby Sawyer College. Colby-Sawyer virtualized its faculty and staff desktops first. They
knew that persistent desktops would be required to satisfy the heavy customization
requirements of its professors.
Mercer University. Mercer has almost 1000 virtual lab desktops in production to deliver
on its “Borderless Classroom” vision. Unidesk is used to provision the non-persistent
desktops that are accessed through VMware Horizon View.
Tennessee Tech University. TTU has a mix of persistent and non-persistent desktops for
labs, engineering classrooms, the health center, and staff, all provisioned from 1 gold OS
layer.
William Woods University. William Woods’ deployment is completely non-persistent,
since its goal was to make its lab desktops accessible to students from any location.
Analyzing the Two Most Common Desktop Models
Let’s finish by taking a more detailed look at the use cases, technologies, and pros and cons of
the two most common virtual desktop provisioning and management models now in use:
Non-Persistent Desktops with Point Tool Management
Persistent and Non-Persistent Desktops with Desktop Layering
Non-Persistent Desktops with Point Tool Management
The most successful implementations of the non-persistent model with management by point
tools are environments that require high availability and have simple application needs.
The classic use case is the large call center - hundreds or thousands of desktops that all have
the same set of applications and must be available 18 to 24 hours a day. The key to being
successful with non-persistent desktops is that the applications are either the same for all
desktops, or they can be delivered using application streaming or traditional application
virtualization.
Use Cases
Call Centers
Static Classrooms, Student Labs, and Training Rooms
Libraries
Task Workers
Kiosks
Technologies
VMware View Composer, Citrix Provisioning Server, Citrix Machine Creation Services
Desktop Provisioning
VMware Linked Clones, Citrix vDisk Image Sharing and Storage Optimization
VMware Thin App, Microsoft App-V Application Virtualization and Streaming
VMware View Persona, Citrix User Profile Management
Profile Management
Citrix Personal vDisk User Virtualization
Application Delivery For any desktop model, application delivery is the big challenge. In this model, applications can
be delivered and updated in two ways:
Included in the base image.
Streamed or virtualized with Microsoft App-V or VMware ThinApp.
Including applications in the base image is not advisable except for the simplest use cases.
Otherwise, you’ll face several challenges:
Building every possible app into a single Windows image would force you to license
every app for every user.
Having to update the master image every time an application needs to be updated will
impact all desktops.
Creating different Windows images with different combinations of apps will create
patching inefficiencies and drive up the operational costs of VDI.
For these reasons, most organizations opt for application virtualization.
With Microsoft App-V, streaming servers serve virtualized applications to desktops. The
desktop has a client with a large cache. You can either load the whole application at run time
or pre-cache the application on the desktops.
App-V has a concept of "sequencing" an application where it figures out the files that are
needed when you first run the application. Those files are streamed to the desktop first when
the application is launched to speed up launch times.
ThinApp works slightly differently in that it encapsulates the entire streaming file into an
executable that can be run from a file server or copied locally to the desktop.
One benefit of these traditional application virtualization products is that they isolate
applications from each other, enabling you to run conflicting applications side-by-side. Another
benefit is that applications can be dynamically assigned to users as they logon.
Traditional application virtualization also has several drawbacks:
It is difficult and time-consuming. By the time you’ve finished the desktop setup, pre-scans,
post-scans, scripting workarounds, Windows registry changes, and deployment to 50
desktops, you’ll find a full day has passed. Or more. It’s not unusual to spend a week
virtualizing a single app. And that’s if you’re an expert.
Not all apps can be virtualized. Even if you are an expert, there’s a long list of apps that
cannot be virtualized with traditional app virtualization tools. Apps with system services
and boot time drivers (e.g. antivirus, printers, scanners, etc.), homegrown apps, and apps
with complex Setup procedures often won’t work.
Isolated apps can’t cross-communicate. Application isolation puts apps into their own
protective “bubbles,” effectively hiding them from Windows and other apps. This is perfect
for running multiple versions of the same software (e.g. Java or Microsoft Access) on the
same desktop. But it’s a showstopper for the other 95% of apps that need to share data,
link to each other, and cross-communicate.
Pros and Cons The Pros of non-persistent desktops with point tool management:
High Availability – If a host or storage array fails, users can simply logon to another
desktop. There are still parts of the environment that can cause an outage, but with all
the desktops being the same, it is easier to provide DR with a mirrored site.
Good Manageability – Management of the Windows OS is efficient, as long as you don’t
create multiple images. Any patches/updates to the central image are deployed to all
cloned desktops. If applications are the same for all desktops, application manageability
is also excellent. If apps must be different, manageability depends on the ability to
virtualize the applications and the skill and size of your IT staff.
Minimal Storage Footprint – Both VMware View Composer with Linked Clones and Citrix
Provisioning Services (PVS) with vDisk greatly reduce storage requirements. PVS can
even be used with local storage on the PVS servers.
Fast Deployment of Simple Apps – Using application streaming/virtualization,
applications can be deployed very quickly to all desktops in the environment. For
applications that are not easily virtualized, this becomes harder and may result in OS
image sprawl.
Fast Deployment of New Desktops – Desktops can be deployed very quickly given that
the OS and the parent/master desktop are already built.
Simple Desktop Refresh – Desktops can be reset to a pristine state with a simple reboot.
This is well-suited for student labs, some classrooms, kiosks, and training facilities.
The Cons of non-persistent desktops with point tool management:
Application Delivery – There will be a large percentage of applications that take too
much time to package with traditional app streaming/virtualization tools, and a smaller
percentage that will not work at all due to the use of drivers, the way they implement
licensing checks, and their designs not being compatible with isolation.
Image Sprawl – Because of the challenges with app virtualization, you’ll have to fall back
to delivering apps as part of the Windows image. If your users have diverse application
requirements, this will result in many images. The more images that are required, the
more time and effort it takes to patch and update the images.
User-Installed Applications – Enabling users to install their own applications or IT to
install one-off apps on behalf of users is difficult with this model. VMware offers no
native capability. Citrix offers Personal vDisk, which creates a persistent layer on top of
the PVS image for each desktop. When users and IT admins install applications into the
Personal vDisk, the apps survive base image updates. However, most organizations
using Personal vDisk encounter the same inefficiencies as including apps in the base
image, since delivering and updating an app used by 20 people would require installing
and patching the app 20 times in 20 Personal vDisks.
Complexity and Cost – Many lean IT organizations struggle to implement and manage
non-persistent desktops with the mix of point tools. The complexity of this model often
results in VDI being managed by Level 2 and 3 server administrators, rather than the
Level 1 admins who managed PCs. This adds unforeseen costs in the form of hiring
outside consultants, or diverts senior IT staff from forward-facing, strategic projects.
End User Confidence – Users will often lose settings, plug-ins, or data every time
Windows is updated and desktops are recomposed because data redirection and profile
management cannot capture everything. End user productivity – and confidence – will
suffer when they have to manually reconfigure their desktops. The service desk will also
spend more time taking support calls.
Persistent and Non-Persistent Desktops with Desktop Layering
Desktop layering offers the flexibility of creating persistent or non-persistent desktops. It also
combines desktop provisioning, application virtualization, image management, personalization,
and storage optimization into one, easy-to-use technology platform. For these two reasons,
layering is being rapidly adopted by Citrix XenDesktop and VMware Horizon View customers as
an alternative to non-persistent desktops managed by point tools.
Use Cases
Office staff / knowledge workers
Professional workers such as lawyers, engineers, and architects
Healthcare clinicians and staff
Students whose desktops stay with them all 4 years
Classrooms, student labs, and training rooms where apps change often
Developers
Technology
Desktop Layering (e.g. Unidesk)
Storage Optimization
Desktop Provisioning
Image Sharing
Application Virtualization
Profile Management
User Virtualization
Application Delivery In the layering model, applications can be delivered and updated in two ways:
Included in the base OS layer.
Virtualized as an Application layer.
Virtualizing an application as a layer is fast and simple – an administrator selects an Installation
virtual machine, logs onto the machine, installs the application (doing whatever they would
normally do to install the app on a regular desktop), and clicks Finalize. The application layer is
now available to be assigned to any desktop on top of the base OS layer.
Because the layering process is so fast and easy and almost any application can be layered,
most customers keep the base OS layer clean, and deliver applications as independent layers.
Virtualizing applications with layering technology has many advantages:
It’s fast and easy. Layering often takes less than 15 minutes. The process is so simple that
many customers have interns layering applications.
It works with 99.5% of applications. Apps with system services and boot time drivers (e.g.
antivirus, printers, scanners, etc.), homegrown apps, and apps with complex Setup
procedures can all be layered.
Apps can cross-communicate. Layered apps are not isolated. They appear to Windows, and
to other apps, as if they are natively installed. As a result, customers who rely on add-ins
and plug-ins for Microsoft Office and other core applications can virtualize the plug-ins as
separate layers to make patching and updating fast and easy. Yet they don’t have to worry
that the plug-ins won’t work with their base application.
The main drawback of application layering is that layers cannot yet be dynamically assigned to
users as they logon. Desktops must be rebooted before the layered apps appear. In addition,
layered applications are not isolated. However, layering works seamlessly with traditional
application virtualization technology if isolation is needed – many customers deliver ThinApp
packages in layers to get the benefits of central management, assignment, versioning, and layer
rollback.
Pros and Cons The Pros of persistent and non-persistent desktops with desktop layering:
Good Manageability – 1 gold OS layer can be used for all non-persistent and persistent
desktops for “patch-once” simplicity.
High Availability – Desktop layering uses the same hypervisor APIs as the brokering
management tools, and adds the ability to snapshot and version OS, Application, and
Personalization layers for easy rollback and recovery.
Application Delivery – Almost all apps can be layered in minutes, without deep
packaging expertise. This is probably the single biggest benefit of layering – the ability
to quickly and easily deliver different sets of apps to different desktops, without the
limitations of first-generation app virtualization tools.
Minimal Storage Footprint – Desktop layering greatly reduces storage requirements for
both persistent and non-persistent desktops. It also supports the use of local storage.
Fast Desktop Deployment – Desktops can be deployed very quickly just by selecting
which layers are needed. Templates, which pre-select OS and App layers based on
department or job function, can also be used to further speed up provisioning.
Simple Desktop Refresh – If the desktop is set to be non-persistent, it can be reset to a
pristine state with a simple reboot. Because the user layer is simply thrown out and no
recompose of the desktop is required, this is often faster than non-persistent desktops
implemented with VMware View Composer and Linked Clones and Citrix PVS/MCS.
User-Installed Applications – End users (if they are given Admin rights) and IT admins
can install one-off applications into the Personalization layer and the apps will survive
base layer updates. If it turns out that 20 people need the same app, the app can be
virtualized by IT as an App layer in minutes and assigned to all 20 desktops. From that
point on, it only needs to be patched and updated once.
Less Cost and Complexity – Many lean IT organizations find layering simpler than trying
to implement and manage non-persistent desktops with the mix of point tools. Level 1
administrators can manage daily desktop operations, freeing more senior IT staff to
focus on forward-facing, strategic projects.
End User Confidence – End users won’t be able to tell when Windows or applications
have been patched or updated. All of their icons, shortcuts, screen layouts, personal
apps, and plug-ins will be the same.
The Cons of persistent and non-persistent desktops with desktop layering:
No “Half-Persistence” – With layering, desktops are either persistent (all customizations
are captured and preserved in the Personalization layer) or non-persistent (the
Personalization layer is cleared after each use so that nothing is preserved). If you need
to provide desktops that retain a subset of personal settings, you will need to create
non-persistent layered desktops and add roaming profiles or a profile management tool.
Licensing Cost - Adding a third party layering solution increases VDI costs. However, 600
layering customers attest that the additional cost is usually recouped within 3 months
based on the persistent desktop storage savings and the operational cost savings from
patching Windows only once, virtualizing all apps in minutes, and enabling Level 1 staff
to manage and support VDI.
Summary
You should now have a clearer understanding of what persistent and non-persistent virtual
desktops are, what they are not, what technologies can be used to implement them, and how
customers are succeeding with both desktop types.
You should also realize that VDI innovations such as desktop layering and flash-optimized
storage have rendered the old debate obsolete. Now, the decision to implement non-
persistent or persistent desktops should be based solely on use case, not technology limitations
or cost impact.
Unidesk Corporation, 313 Boston Post Road West, Marlborough, MA 01752 USA Tel 508-573-7800 Fax 508-573-7801
Copyright © 2014 Unidesk Corp. All rights reserved. This product is protected by U.S. and international copyright and intellectual property
laws. Unidesk is a registered trademark of Unidesk Corp. in the United States and/or other jurisdictions. All other marks and names mentioned
herein may be trademarks of their respective companies. Item No: UNI-EB-PERSISTENT-NON-VDI