This document discusses an attempt by the Head of IT Services at the University of York to understand and quantify the true costs of central IT services. It outlines their methodology which involved identifying IT systems and services, attributing physical and virtual servers, determining hardware, software, staff and other costs, and allocating these costs across services. While still a work in progress, it provides some initial cost figures per service and highlights challenges around accurately attributing shared infrastructure costs. The process was iterative and took time to fully understand their detailed service landscape.
Salient Features of India constitution especially power and functions
What does central IT really cost? An attempt to find out! - Heidi Fraser-Krauss, Head of IT Services University of York
1. What
does
central
IT
really
cost?
An
a3empt
to
find
out!
Heidi
Fraser-‐Krauss
Head
of
IT
Services
University
of
York
2. Why?
• To
understand
where
money
is
spent.
• To
be
able
to
compare
our
costs
with
those
of
other
providers-‐
cloud…..
• To
be
able
to
price
services
that
we
offer
to
others.
• Curiosity!
• To
have
evidence
to
argue
for
more
resource.
• To
understand
central
vs.
local
provision
costs.
4. How-‐
understanding
services?
1. Started
with
geOng
an
understanding
our
systems
and
services
– DNS
– Finance
system
– AcKve
Directory
– Storage
2. Listed
physical
and
virtual
servers
and
a3ributed
each
server
to
a
service
There
were
lots
of
gaps!
5. How-‐
Costs?
3.
Got
a
handle
on
license
and
maintenance
costs
– Hardware
maintenance
– O/S
Licenses
4.
Added
up
hardware
capital
costs
5.
Worked
out
a
cost
for
a
virtual
machine
– Physical
infrastructure
– VM
license
6. How-‐
Costs?
7.
Worked
out
a
cost
per
Gb
of
storage
– DifferenKated
Kers
8.
A3ributed
these
costs
across
the
different
services
9. Group
heads
allocated
staff-‐Kme
to
the
previously
idenKfied
services
10.
We
a3ributed
staff
costs
to
services
– Difficult
to
do
as
no
Kme
sheets
– Accounted
for
Kme
spent
supporKng
non-‐central
systems
7. Work
in
progress…..
• Soaware
licence
costs
– To
date
a
broad
brush
for
major
systems
• A3ribuKon
of
data
centre
costs
– Capital
– Running
costs
• More
accurate
a3ribuKon
of
power
and
cooling
– Can
do
it
per
server
• More
sophisKcated
use
of
TraC
costs
• Add
in
depreciaKon
of
capital
costs
• Cover
whole
University
8. Keeping
up
to
date
• A3ribute
purchase
orders
to
service
• Once
model
in
place
and
costs
known
shouldn’t
be
difficult
to
keep
things
up
to
date
9. Lessons
learned
• Need
a
firm
grasp
of
services
at
a
detailed
level
– Takes
Kme
– Things
get
missed
out
• Staff
cost
allocaKon
without
Kme
sheets
is
a
best
guess.
• We
used
an
iteraKve
approach
for
obtaining
costs
– This
annoyed
people
– It
could
have
been
done
in
one
iteraKon
10. Challenges
• How
to
allocate
network
costs
or
other
costs
that
are
truly
infrastructural?
– How
do
you
cost
the
network
for
student
record
system?