APA
VDI
Solution


Research
Report


Tony
Habash,
CIO
ITS


American
Psychological
Association
 | 
 | American
Psychological
Association
 | | | | | 
 | February
4,
2013
 | | | 

 |

Prepared
by:
 John
Doe,
Manager
ITS
Network
Systems
Engineering


Table
of
Contents

Executive
Summary
........................................................................................................................
1

Introduction
....................................................................................................................................
1

Results
of
Research
........................................................................................................................
3

Completed
Work
........................................................................................................................
3
 1.
 Gain
clear
understanding
of
Virtual
Desktop
Infrastructure
(VDI)
...............................
3
 2.
 Find
out
which
VDI
solution(s)
is
cost
effective
for
APA
..............................................
5
 3.
 Identify
applications
that
will
not
function
with
VDI
....................................................
5
 4.
 Investigate
required
VDI
training
.................................................................................
6

Conclusion
......................................................................................................................................
7

References
......................................................................................................................................
7


Executive
Summary
 In
2012,
an
Information
Technology
Services
(ITS)
project
was
opened
to
complete
a
Virtual
 desktop infrastructure
(VDI)
proof
of
concept
(POC).

The
VDI
POC
project
was
opened
to
 investigate how
VDI
will
be
implemented
in
our
current
enterprise.

At
the
same
time,
we
must
 find out
which
solution
works
best
for
the
American
Psychological
Association
(APA).

The
 combination of
limited
staff
resources
and
outdated
patch,
desktop
and
application
 deployment results
in
ineffective
desktop
management
practices.

We
currently
struggle
with
 desktop management
at
APA.
As
it
stands
today,
we
do
not
have
an
effective
way
of
upgrading
 desktop applications;
we
are
still
visiting
the
desktop
to
install
most
applications.

The
 organization will
be
embracing
office
hoteling
which
will
result
in
more
teleworking
employees.


ITS
Operations
will
need
to
provide
teleworkers
with
a
consistent
desktop
(Thomas,
2013).

 The
VDI
POC
project
will
help
us
determine
if
VDI
is
our
solution
to
our
desktop
management
 shortfalls. The
scope
of
the
VDI
POC
project
includes:
 • Selection
of
a
VDI
solution
 • Engineer
training
 • Install
and
configuration
of
a
VDI
POC
environment
 • Application/desktop
discovery
 • Measure
the
level
of
effort
needed
to
implement
VDI
 The
selection
of
a
VDI
solution
includes
the
selection
of
a
vendor
partner
such
as
Citrix,

Microsoft
or
VMWare
as
well
as
the
selection
of
a
VDI
solution
type
which
will
best
fit
APA.

The
 latter being
the
most
challenging
effort.

APA
is
a
midsize
enterprise
with
a
variant
of
use
cases,
 applications and
desktop
needs.
Before
selecting
a
solution
type
we
had
to
research
the
types
 of VDI
solutions
currently
available.


This
required
an
understanding
of
the
VDI
solution
types,
 how they
are
implemented
and
there
capabilities.

We
have
completed
preliminary
and
 secondary research
of
VDI
solution
types
the
results
of
our
research
will
be
essential
 information we’ll
use
to
choose
a
vendor
partner
and
solution
for
the
VDI
POC
project.


Introduction
 There
are
two
types
of
virtual
desktop
infrastructure
for
APA
to
consider
given
our
current
 experience with
application
virtualization
and
Remote
Desktop
Session
Host
(RDSH
previous
 known as
Terminal
Services).

Both
application
virtualization
and
RDSH
are
currently
used
in
our
 enterprise. However,
to
address
our
desktop
management
issues
and
to
achieve
the


teleworker
functionality,
we
need
implement
either
a
Server‐hosted
or
Type
1
or
Type
2
client‐ side VDI
solution.

All
of
which
are
full
VDI
solutions
unlike
application
virtualization
or
RDSH.


These
full
VDI
solutions
present
users
with
their
own
desktop
and
operating
system.

Type
1
VDI
 is the
“server‐hosted”
solution.

Type
2
VDI
is
a
“client‐side”
solution.

All
of
these
VDI
solutions:
 • Provide
the
flexibility
we’ll
need
to
fulfill
requirements
of
our
user
community
 • Address
our
current
desktop
management
woes
 • Provide
the
services
needed
for
future
hoteling
concepts.


 • Support
a

variety
of
devices
 Users
with
Macs,
Windows
PCs
and
the
various
tablet
operating
systems
can
access
the
VDI
 environment. This
is
important
as
APA
gravitate
towards
more
teleworkers
who
will
bring
their
 own devices.

Bring
your
own
device
(BYOD)
users
will
grow
as
the
hoteling
concept
catches
on.


Last
but
not
least
Type
1
“server‐hosted”
and
type
2
“client‐side”
VDI
solutions
provides
better
 security and
more
control
over
user
desktops.

Organizations
have
fewer
security
outbreaks
 when they
implement
VDI
environments
(Browning
&
Margevicius,
2012)
 Initially,
we
considered
Remote
Desktop
Session
Host
(RDSH)
as
a
VDI
solution
type.

After
 preliminary research
we
found
the
RDSH
solution
will
essentially
provide
the
same
service
as
 the CitrixZenApp
solution
we
currently
use
for
remote
access.

This
solution
will
not
address
our
 current desktop
management
problems.

With
RDSH,
users
are
not
presented
with
their
own
 virtual desktop,
instead
they
are
provided
with
a
shared
virtual
server
running
with
the
server

OS.

With
a
full
VDI
solution,
the
user
is
presented
with
their
own
virtual
desktop
running
a
 client operating
system
such
as
Windows
7
or
Windows
8.

This
provides
users
an
experience
 closer to
their
experience
with
a
traditional
desktop.

Although
RDSH
has
been
a
good
solution
 for remote
access,
it
lacks
the
flexibility
needed
for
our
user
community.

RDSH
was
removed
 from the
scope
of
this
research
for
following
reasons:
 • APA
has
users
who
require
admin
access
to
their
desktops
 • Some
APA
applications
do
not
work
in
an
RDSH
environment
 • Some
peripherals
do
not
work
in
a
RDSH
environment
 • Includes
the
same
amount
of
administrative
overhead
as
traditional
desktops


Results
of
Research

Completed
Work

1. Gain
clear
understanding
of
Virtual
Desktop
Infrastructure
(VDI)

Types
of
VDI


Server‐hosted
Server‐hosted
VDI
is
where
users
are
presented
with
remote
virtual
desktops
that
run
on
a
server.

Each
user
has
their
own,
unique/personal
and
completely
isolated
workstation
(Spruijt,

2012).

Some
Server‐hosted
vendor
solutions
are
Microsoft
VDI,
Citrix
XenDesktop
and
VMware
View.

Remote
virtual
desktops
can
be
non‐persistent,
persistent
or
layered.

Non‐persistent
virtual
desktops
are
pooled
desktops
which
are
shared
by
users.

Any
changes
or
profiles
are
lost
once
the
virtual
desktop
is
rebooted.

Persistent
remote
virtual
desktops
are
the
exact
 opposite of
non‐persistent
desktops;
user
changes
and
profiles
are
retained
after
the
virtual
desktop
reboots.

Layered
virtual
desktops
are
hybrid
virtual
desktops
where
persistent
desktops
are
layered
on
top
of
non‐persistent
desktop.

Layered
desktops
allow
some
user
customization
which
persists
across
reboots
and
allows
IT
staff
to
revert
back
to
a
clean
desktop
image.
 With
non‐persistent
virtual
desktops
the
user
is
presented
with
a
new
virtual
desktop
with
each
 connection. Non‐persistent
virtual
desktop
images
are
identical
allowing
fast
rollouts
of
 software and
patch
deployments.

They
require
less
administrative
overhead
because
of
the
 standardized images.

All
software
updates
and
installs
must
be
done
by
IT
staff.

The
lost
of
 changes and
personal
settings
once
the
virtual
desktop
is
rebooted
is
not
ideal
for
some
users.


This
is
the
biggest
disadvantage
to
using
non‐persistent
virtual
desktops.
However,
profile
 management tools
can
be
used
to
reinstate
profile
changes
for
users
upon
login.

 Persistent
virtual
desktops
allow
users
to
install
and
configure
their
own
applications/software.


To
retain
the
user
changes
and
profiles
after
reboots,
the
user
changes
and
settings
are
stored
 within their
own
virtual
desktop
image.

With
that
said,
persistent
virtual
desktops
allow
more
 flexibility. The
virtual
desktop
images
are
usually
stored
on
expensive
SAN/NAS
disk
storage
 within the
datacenter.

This
storage
requirement
makes
persistent
virtual
desktops
more
 expensive to
store
and
maintain.


 Layered
virtual
desktops
include
the
best
of
both
worlds.

The
foundation
of
a
layered
virtual
 desktop is
a
non‐persistent
virtual
desktop.

The
non‐persistent
desktop
is
layered
with
any
 changes that
need
to
persist
after
the
virtual
desktop
is
rebooted.

IT
staff
install
and
configure


applications
but
still
there’s
less
administrative
overhead
than
with
persistent
virtual
desktop.


The
SAN/NAS
storage
requirement
still
exist
with
layered
virtual
desktops,
however
the
amount
 of storage
needed
is
less
than
that
of
totally
persistent
virtual
desktops.


Client
Side
Client
side
VDI
solutions
are
ran
locally
on
a
client
running
a
hypervisor.

Like
Server‐hosted
VDI,
client‐side
VDI
allows
central
management
of
virtual
desktops
and
offer
flexibility.

However,
client‐side
VDI
also
offer
high
performance
and
an
experience
closer
to
that
of
a
traditional
desktop.

A
hypervisor
is
hardware
or
software
used
to
run
virtual
machines.

For
this
reason,
client
side
VDI
solutions
are
categorized
as
Type
1
“hardware”
and
Type
2
“software”.

The
Hypervisor
ensures
that
every
virtual
machine
operates
independently
of
hardware
and
renders
it
possible
to
utilize
several
Virtual
Machines
on
the
workstation
at
the
same
time
(Spruijt,

2012).



Type
1
hardware
hypervisors
provides
performance
very
close
to
that
of
a
traditional
desktop.


The
user
must
have
a
supported
hardware
hypervisor
to
access
their
virtual
desktops.



Type
1
 hypervisors will
not
run
on
smart
phones
or
tablets.

This
disadvantage
of
Type
1
hardware
 hypervisors limits
the
devices
users
can
use
to
connect
to
virtual
desktops.

Client‐side
Type
1

VDI
solutions
does
not
allow
users
to
access
their
virtual
desktops
from
anywhere
at
any
time.


 Type
2
software
hypervisors
are
installed
on
top
of
the
client’s
operating
system.

Performance
 is not
optimal
for
some
power
users.

However,
since
software
hypervisors
are
supported
on

Macs,
Windows,
smart
devices
and
tablets,
they
allow
anywhere
access
to
virtual
desktops.




2. Find
out
which
VDI
solution(s)
is
cost
effective
for
APA
 There
is
no
return
of
investment
(ROI)
expectations
with
the
implementation
of
VDI
solution
at

APA.

The
reality
is
there
are
no
immediate
cost
savings
for
APA
with
the
implementation
of
a

VDI
solution.

As
Browning
and
Margevicius
stated,
“cost
reduction
is
over
long
term”.


There
is
 just hope
the
implementation
will
allow
us
manage
desktops
in
a
more
efficient
and
timely
 manner (Thomas,
2013).

Even
with
this
reassurance,
we
still
must
consider
which
solution
is
 the most
cost
effective
for
APA.

The
up‐front
capital
expense
is
where
the
bulk
of
the
cost
lies.


The
up‐front
cost
of
any
of
the
VDI
solutions
depends
on
the
complexity
of
the
implementation
 and the
licensing
needed
for
the
VDI
environment.

The
selected
VDI
architecture
will
dictate
 the complexity
and
licenses
needed
for
APA’s
VDI
implementation.

There
are
three
trains
of
 thought with
desktop
virtualization
architecture
(Berger,
2012):
 • Build
it
yourself
 • Converged
architecture
 • Appliances
 Of
these
three
architectures,
appliances
are
the
least
expensive
to
implement.

VDI
appliances
 such as
Citrix
VDI‐In‐A‐Box
or
VMWare
Rapid
Desktop
are
server‐hosted
VDI
solutions.

VDI‐in‐ a‐Box eliminates
complexity
by
bundling
all
the
VDI
functions
including
load
balancing,
 connection brokering,
dynamic
desktop
provisioning,
and
high‐availability
into
a
single
turnkey
 virtual appliance
(Spruijt,
2012).


Also
with
an
appliance,
you
can
minimize
the
complexity
of
 the VDI
environment.

Management
servers
and
SAN
storage
are
not
needed
when
 implementing VDI
using
an
all‐in‐one
appliance.

These
all‐in‐one
appliances
make
the
 calculation of
the
necessary
storage,
CPU
and
RAM
needed
to
service
a
set
amount
of
users
a
 easy to
project.


 The
overall
cost
savings
lie
in
the
time
it
takes
to
implement
the
solution.

With
appliances,
 there’s a
more
less
of
a
learning
curve
for
engineers
and
the
appliance
acquisition
includes
all
of
 the components
necessary
to
start
deploying
virtual
desktops.
As
a
non‐profit
organization,
 licenses are
sold
to
APA
at
charity
prices.

Licensing
costs
will
be
minimal
compared
to
the
cost
 indicated by
“for
profit”
companies
and
organization.


3. Identify
applications
that
will
not
function
with
VDI


Research
showed
most
applications
worked
with
VDI.

The
compatibility
application
problems
 noted with
VDI
were
noted
for
Remote
Session
Server
Host
(RDSH).

As
stated
earlier,
this
 solution was
removed
as
a
feasible
VDI
solution
for
APA.

There
are
problems
with
some

CAD/CAM
applications
in
VDI
environments.

This
is
not
due
to
the
virtual
desktop
images
 themselves since
they
run
the
traditional
desktop
operating
systems.

There
are
performance
 issues when
users
run
CAD/CAM
applications
in
a
shared
VDI
environment.

The
same
is
so
for
 graphic intensive
2D/3D
applications.

Changes
to
the
VDI
environments
can
be
made
to
 optimize performance.


 For
these
applications
to
work
in
a
VDI
environment
additional
graphical
processor
units
(GPU)
 are required.

Additional
physical
GPUs
can
be
installed
to
support
users
who
need
to
use

CAD/CAM
or
graphic
intensive
applications.

As
with
a
traditional
desktop,
desktop
publishing
 software users
and
video
editors
are
provided
higher‐end
desktops
which
lend
themselves
to
 high CPU/RAM
usage.


 According
to
Cynthia
Swann,
“We
have
less
than
20
users
who
need
to
run
CAD/CAM
or
graphic
 intensive software”.

These
advanced
users
can
also
continue
to
use
traditional
desktops.

We’ll
 have to
do
complete
use
case
analysis
after
the
VDI
POC
project.



4. Investigate
required
VDI
training
 With
server‐hosted
VDI,
the
end
users
will
not
require
any
additional
training.

Server‐hosted
VDI
presents
users
with
their
own
desktop
running
the
same
operating
system
and
applications
as
their
traditional
desktop.

The
training
needed
with
a
server‐hosted
VDI
solution
will
be
for
systems
engineers
who
will
be
building
and
implementing
the
VDI
solution.

For
implementation
success,
systems
engineers
will
need
to
complete
the
necessary
VDI
technical
training.

Per
(Madden,
Knuth,
&
Madden,
2012),
“VDI
requires
VDI‐specific
skills”.
 Client‐side
VDI
solutions
will
require
end
user,
help
desk
staff
and
system
engineer
training.


End
users
will
need
to
know
how
to
connect
to
their
virtual
desktops.

There’s
slight
differences
 how a
user
interacts
with
both
the
Type
1
and
Type
2
client‐side
VDI
solutions.

This
makes
end
 user training
a
requirement
with
the
implementation
of
a
client‐side
VDI
solution.

As
with
the
 server‐hosted solutions,
technical
training
will
be
essential
for
systems
engineers
and
help
desk
 staff supporting
our
end
users.


Conclusion

The
implementation
of
a
VDI
solution
will
be
a
challenging
undertaking.

I
think
we’re
on
the
 right path
with
completing
a
VDI
POC
prior
to
deciding
to
proceed
with
implementing
VDI.


We’ll
need
to
factor
in
deep
analysis
of
our
use
cases,
user
applications
and
understand
the
 potential issues
we’ll
come
across.

We’ll
need
to
truly
understand
the
overall
cost
of
 implementing a
VDI
solution.

My
suggestion
after
the
completion
of
this
research
is
to
limit
our
 risk and
cost
by
implementing
a
solution
the
least
amount
of
complexity.

We
should
also
 implement VDI
for
a
small
set
of
users
in
departments
where
is
makes
the
most
since.

Starting
 the VDI
implementation
off
with
users
requiring
the
least
amount
of
control
will
be
a
beneficial
 approach. We
should
be
realistic
about
the
amount
of
users
who
will
be
able
to
work
off
virtual
 desktops. This
research
in
conjunction
with
the
information
gained
during
the
VDI
POC
project
 will dictate
which
solution
will
work
for
APA
given
our
use
cases,
users,
and
system
 requirements. According
to
APA’s
requirements
for
flexibility,
a
server‐host
VDI
solution
appears
to
be
the
 best fit
for
APA.

The
least
amount
of
end‐user
training
is
required
with
a
server‐hosted
VDI
 solution. Most
of
the
applications
used
in
our
enterprise
will
run
with
a
server‐hosted
VDI
 solution. When
you
take
a
look
at
1)
the
time
needed
to
learn
the
respective
VDI
solutions,
2)
 the implementation
complexity
and
3)
the
cost
per
user;
a
server‐hosted
VDI
appliance
is
the
 most cost
effective
VDI
solution
for
APA.




References
 Berger,
G.
(2012,
September).
Desktop
virtualization
architecture
–
solutions
&
benefits?

Retrieved
from
Gartner:
http://community.gartner.com/t5/Data‐Center/Desktop‐ virtualization‐architecture‐solutions‐amp‐benefits/m‐p/7298/thread‐id/255 Botelho,
B.
(2010,
July).
Why
move
from
Terminal
Services
to
VDI?
Retrieved
from

SearchVirtualDesktop:
http://searchvirtualdesktop.techtarget.com/Why‐move‐from‐
Terminal‐Services‐to‐VDI

Browning,
J.
A.,
&
Margevicius,
M.
A.
(2012,
September).
Desktop
virtualization
best
practices
 for the
midmarket.
Retrieved
from
Gartner:
 http://my.gartner.com/portal/server.pt?open=512&objID=235&mode=2&PageID=8163 64&resId=2179219&ref=QuickSearch&sthkw=vdi+cost

Madden,
B.,
Knuth,
G.,
&
Madden,
J.
(2012).
The
VDI
Delusion:
Why
Desktop
Virtualization

Failed
to
Live
Up
to
the
Hype,
and
What
the
Future
Enterprise
Desktop
Will
Really
Look

Like.
Burning
Troll
Productions.

QIPeople.
(n.d.).
Retrieved
from
http://www.qipeople.co.uk/virtual‐desktop‐infrastructure.html


Spruijt,
R.
(2012,
August).
VDI
Smackdown!
Retrieved
from
PQR:
 http://www.pqr.com/images/stories/Downloads/whitepapers/vdi%20smackdown.pdf Swann,
C.
(2013,
February).
Manager,
ITS
Support
Services.
(R.
Toles‐Abdullah,
Interviewer)

Thomas,
D.
(2013,
2).
Director,
APA
ITS
Operations.
(R.
Toles‐Abdullah,
Interviewer)