Home' Defense Systems : January and February 2013 Contents is an IT system in place that could sup-
port, say, 80 percent of the current pro-
cesses used by our big user population.
And we have developed other systems
to handle the rest of the process.
With that kind of situation what
we want to look at is: does it make
economic sense, once we selected an
IT system that meets a high percentage
of the process and also meets it for a
high percentage of the population, to
have a discussion about changing the
processes of what I call the "outlying
users" so that they can be supported by
the enterprise IT system?
DS: So it's worth the money to
change the 20 percent of outliers?
Halvorsen: Sometimes it is. In gen-
eral, our philosophy has been that
you don t let IT drive the process. You
build the best process, and then you
have IT support it. But I think when
we re talking about big bureaucracies,
in general, what happens is that mul-
tiple IT systems do similar processes.
So if two or three of those IT systems
will support 80 percent or better of
the process and 80 percent or better
of the users, then [we need to] force
the 20 percent or less of the process
and that 20 percent or less of the user
base to change their process and IT
systems so that we end up eliminat-
ing some of the smaller IT systems. I
think that s going to be another way
forward for us.
DS: You commented recently that
data-center consolidation might
seem slow at the moment, but
soon you will be in a position to
close many centers all at once.
Halvorsen: Data center closure is
like a step-function activity. Obvi-
ously, we have systems that might be
located right now in multiple data
centers. So when you cleanse those
by system, you don t end up closing
down a single data center. You have
got to wait until you have mass at a
data center, and then clean up what I
call "all the little pieces" of that data
center, and then close it. It s going to
be a step function.
We have closed about 100 systems
and 3 data centers. I think what you ll
see by the end of this scal year [that
number will be] around 9 or 10 data
centers and maybe 200 systems. We
are right on track with data center
closures, but one of the things we
are looking at is what can we do to
accelerate that process. Ideally, what
we would like to do is get ahead of
schedule because every year that we
can pull the money out earlier gives
us a higher level of savings, and that s
certainly what we are a er right now.
DS: Most existing military cloud
initiatives are related to public-fac-
ing data. Is the public-facing data
most suited to the cloud because
of security considerations, as op-
posed to operational data?
Halvorsen: e "cloud" is an interest-
ing word. So we will use cloud, but
I will also de ne cloud as meaning
distributed data. It is basically ac-
cessible via the Internet or a web. It s
not data that s stored on your laptop;
that s really what the cloud comes
down to. So when you ask whether
public-facing data is more suited to
the cloud, maybe a better way to ask
it is: can I host public-facing data and
make it available via, say, a commer-
cial cloud and leverage commercial
investments? We think the answer is
yes, and we are doing some aggressive
pilots on that.
Now I want to be clear that we are
not lowering the security standards
on the data. ey still have to meet
all the same security standards, but
I think people feel that any time we
don t own the data or it s not one of
our sites, then there is risk.
DS:There is also risk when there
are outages in the commercial
clouds, but maybe that's as not as
much of a concern when you're
talking about public-facing data.
Halvorsen: Right, but I will tell you
that most times the reliability factor---
and this is something I think we truly
have to understand---in many of the
commercial sites meet or exceed our
own standards. ink about it, when
Amazon goes down, Amazon knows
down to the penny just how much mon-
ey it lost. ey have a direct bottom-line
incentive in not letting that happen.
So I think you will see more of our
public-facing and less-sensitive data in
some commercial sites as we look to
save cost. We will be very careful and
probably move on that pretty slowly.
And then, as we put other types of
data into the cloud, I think you will see
us use what I call a "private commercial
cloud," where a commercial provider
has---because of the size of our data
requirement and because we want to
have more control and not have it inter-
mingle with other data---a Navy-only
or a Marines-only commercial cloud.
at still lets us leverage some of the
productions that are available in the
commercial world, yet still maintain
at doesn t always necessarily mean
it will be a Department of Navy entity.
Certainly we are looking hard at where
does it make sense to, say, use an Air
Force location, or an Army location. We
already, in some places, use a Defense
Information Systems Agency location
where that makes sense. ■
DefenseSystems.com | JANUARY/FEBRUARY 2013 9
To read more questions and answers
with Terry Halvorsen, including his
thoughts on enterprise e-mail/mail.mil,
see the Leadership Views box at www.
Additional Online Resources
Links Archive December 2012 March 2013 Navigation Previous Page Next Page