If
we can do more with less
then we weren't doing our job
right in the first place.
An Interview
with Dan
Paolini
Dan
Paolini
Director, Data Management Services
Office of Information Technology
State of New Jersey
The 2nd Annual Enterprise
Data Forum (EDF) will be conducted at the Hilton Philadelphia/Cherry
Hill, in the first week of November. And for those who may not be
familiar with the geography, Cherry Hill is actually in southern
New Jersey, just across the Delaware River from the city of Philadelphia.
Given the location of the EDF, it seems appropriate to give some
local flavor to our latest edition of Data Discussions, by chatting
with Dan Paolini.
Dan is the State
of New Jersey's first Director of Data Management Services. In this
role, he leads the database administration, data administration,
data warehousing and data integration efforts for both agency-specific
and enterprise-wide information technology projects. Prior to this
position, he was the State's first data architect. He is a former
contributing editor for a monthly database magazine and the technical
editor for three database books. He has presented more than forty
technical papers at over twenty conferences in North America and
Europe, and will be speaking at the EDF program on “Enterprise Common
Data Architecture - Roadmap to Integration.”
Tony
Shaw, Wilshire Conferences (Wilshire): Dan, State governments
across the country are going through one of the most stringent periods
of fiscal constraint they’ve ever experienced. How can someone in
your position help the number-crunchers and policy-makers? How is
your DM function being called upon to help solve some of the challenges
they are facing?
Dan
Paolini (Paolini): We hear this battle cry of “Do More
With Less” from a number of directions, which is meaningless. If
we really could do more with less, then we weren’t doing our job
properly in the first place and should get less (or get fired).
What we recognize is that in this environment, you have to “Do Different
With Less”. It isn’t enough to cut spending; we must find ways to
create additional value. The only way to do this is by re-examining
traditional roles and assumptions. The development of a top-down
architectural vision is one step in this direction. I tell our people
that the most serious form of mental illness is to keep doing the
same things the same way with the same people and expect different
results.
The biggest
change we made was stopping development of independent data marts.
They had proliferated throughout the State. By creating an enterprise
data warehouse environment, we are able to enforce the creation
of dependent data marts with conforming dimensions. This has already
saved us several million dollars in development. We have estimated
that ongoing support at this point is less than half of what it
would be in an independent data mart environment, resulting in savings
of more than a quarter million dollars per year. This savings will
increase, as additional data marts are deployed from within this
environment.
Our database
administration reorganization also saved the State money. Prior
to the reorganization, we had forty-two employees in seven separate
DBA groups under five different managers. Each group supported a
separate platform and separate client agencies. We have consolidated
into twenty-nine positions in one group under three managers. We
have been able to move DBA's from sunset mainframe platforms to
distributed RDBMS systems for better workload balancing. We have
also moved DBA staff into data warehousing and data architecture
roles, increasing our capabilities in those areas. We were able
to do all of this while improving our levels of service.
Wilshire:
It sounds like you’re in the hot seat there in Trenton (the New
Jersey state capital) - the state’s first data architect then first
Director of Data Management – you’re obviously having to break down
some barriers and pioneer some new trails. Please tell us what you
do exactly and how you arrived in your current role.
Paolini:
The Office of Information Technology is the central IT agency for
the executive branch of New Jersey state government. My unit was
formed two years ago initially to focus on data warehousing. Last
year, we took on the complete data management function, including
architecture and database administration. My role is to evangelize
our data architecture and develop our staff. I’m not a career civil
servant. While in the private sector, I consulted on various state
data management projects over a ten-year period. I often complained
that the State was too project-focused with too many data silos,
and did not employ sound data management practices. While this provided
me more consulting revenue, it resulted in higher project costs
and higher taxes. Four years ago, I received an offer to put up
or shut up. Today, I am fortunate to be here at a time when our
Governor, James E. McGreevey, and our CTO, Charles S. “Steve” Dawson,
are working to make government more effective and efficient by leveraging
strategic IT architecture.
Wilshire:
So draw some comparisons and contrasts if you would between what
you do in state government and what a similar manager would do in
the private sector?
Paolini:
There is a perception that the public sector is so different from
the private sector that one cannot draw meaningful comparisons.
While there are some significant differences, the truth is that
we face many similar challenges. I believe it is a mistake for public
sector managers to overlook what we can learn from the private sector.
We have universal concerns: Developing a collaborative organization;
Incorporating diverse business units and systems; Dealing with legacy
systems; Developing a data architecture to support data integration;
providing better customer relationship management; and, the ubiquitous
need to achieve project targets with limited resources.
On the other
hand, we do have some major differences: Changes to laws and regulations
drive our transactional system designs. A deadline for a fee increase
might be mandated without regard to the impact on the existing transactional
system. Our executive sponsorship can evaporate in the next election.
And probably our biggest difference: we’re the government - we can
lock up our customers, whereas most private sector organizations
cannot (although they may at times want to).
Wilshire:
Let’s talk about your specific approaches to key data management
issues. Perhaps we could start that by having you explain your “Enterprise
Common Data Architecture.”
Paolini:
We developed the New Jersey Common Data Architecture (NJCDA) to
address data reusability issues. The architecture is a collection
of related tools and technologies, along with standards and policies
and the methodology and the expertise to employ them. While not
described in pure Zachman terms, we have a conceptual, logical and
physical view of the architecture that enables us to discuss it
with different communities. I have been asked if this architecture
is the same as the concept of a government information factory.
While there is some usefulness to the factory metaphor in certain
situations, I have found that it turns off the very executive sponsors
we are trying to reach. The NJCDA is both broader and deeper in
concept and practice, and encompasses traditional batch data integration
as well as real-time application integration.
Wilshire:
Could you elaborate on your data “reusability” strategy, and how
you differentiate it from data “sharing” or data “integration?”
Paolini:
Data reusability is the process of collecting, managing and storing
electronic data in all forms and formats to promote the efficient
use of that data between our agencies and applications while minimizing
data redundancy. It is more specific than data sharing, in that
data sharing can involve data duplication, while the goal of data
reusability is to eliminate unnecessary duplication. It is more
specific than data integration, in that data integration is the
melding of information from disparate systems, while the goal of
data reusability is not just to integrate but also to optimize and
standardize the common data.
The purpose
of this reusability is to meet the State’s strategic and operational
needs. Our architecture has six drivers. Four of them were identified
by the National Association of State Chief Information Officers
in its report, “NATIONAL INFORMATION ARCHITECTURE: Toward National
Sharing of Governmental Information”: Shared information is more
accurate, timelier, more complete and less expensive. To these we
added that shared information is more accessible and more useful.
Wilshire:
You mention the real-time aspect of your architecture, which I find
interesting because I guess I think of state governments as large
bureaucracies without the same sort of competitive motivation for
an investment in real-time analytics that many private organizations
would have. Explain to me how the State is developing its real-time
analytical capability and why.
Paolini:
One of the values of a common data architecture is its use as a
roadmap to places you haven’t yet reached. We are just getting into
real-time data integration, but the architecture already describes
how, what, when, and why. There are some obvious areas involving
domestic security and fraud detection that benefit from low-latency
integration. We also have more technically savvy citizens and officials
that have higher expectations. While we sometimes have to manage
those expectations, we find that the technology is making it possible
to integrate information sooner and more often. A vendor has coined
the term “right-time”, as opposed to “real-time”, and that is the
way we try to approach integration projects.
Wilshire:
How are you handling meta data management? I’m particularly interested
in the regulatory and public policy aspects of this question please.
Paolini:
We created a data architecture office into which we placed our meta
data management efforts. To some extent, we were charmed by the
vendors and standards organizations that promised one global village
of meta data. Unfortunately, we aren’t there yet and may never be
there. What we are trying to do is converge toolsets that generate
meta data, to minimize the number of discrete repositories. We are
building, incrementally, a repository of these meta models to function
as a central catalog of these individual sources. Our challenge
is that we have to do it without full sponsorship, completing components
as pieces of other projects.
Two years ago,
I had to explain to management what meta data was. Last year, they
would say, “Dan thinks everything is about meta data.” This year,
we are starting to hear, “This meta data stuff is important.” I
fully expect next year to hear, “Shouldn’t we have a meta data strategy?
When are you going to do something about it?” So we are making progress!
The biggest
challenges we face from a regulatory or public policy perspective
are related. On the one hand, traditional policy, as well as statutes
and regulations, do not support the concept of enterprise infrastructure
and “good of the many” planning. There is resistance to centralization,
including of meta data. On the other hand, we have the Open Public
Records Act (OPRA) in New Jersey, which makes most information a
public record, unless specifically protected. We classified meta
data repositories and other derived facilitative data as protected.
People can still get access to information, but they must work through
the stewardship agency that is responsible for the original data.
Wilshire:
While we’re on the subject of regulation and public policy, I must
ask you about the “Privacy” issue. I have to assume that your organization
has been deeply involved in developing a strategy for protecting
private data? How are you dealing with these challenges?
Paolini:
Part of our meta data strategy is to categorize data in different
ways. For our security dimension, we classify data as Public, Secure,
Confidential or Personal. We anticipated HIPAA, and the “Personal”
category reflects both health-related information as well as other
personally identifiable information. The real challenge is to overlay
existing legacy data stores with such a framework. We’re not there
yet. We face the same diametric challenges as other organizations:
the public wants access to as much information as possible, as long
as it isn’t about them.
Wilshire:
Dan, there’s a bullet point in your EDF talk that reads: “How to
get IT staff, executive management and business users to see the
big picture.” Isn’t that the perennial challenge for data management?
What’s your answer to the problem?
Paolini:
I wish it were something as simple as “Speak slowly and use lots
of pictures.” The salient aspect, from my perspective, is that you
can’t get anyone else to see the big picture if you can’t. In other
words, as soon as possible, form a vision for your architecture
that you are comfortable with and that you can talk about for thirty
seconds or three hours. Then get out there and sell it. We tell
our staff to “Reach – Preach – Breach – Teach”.
I find that
executive management and business users will buy into a rational
vision if it clearly defines the benefits to the organization. The
difficulty may be in getting legacy system IT staff on the same
page. A well-defined common data architecture will threaten many
areas within your IT staff, for different reasons. Be prepared for
pushback, and have the stamina to overcome it. Most importantly,
do the right things for the right reasons.
I should point
out that while we have gotten pushback from some IT areas within
our organization, my staff is one of the more pleasant surprises
I received. In the midst of the expected turmoil of reorganization,
they displayed great professionalism and a commitment to our goals
that made my job much easier. We have more than a dozen people working
in different roles today than a year ago, and they continue to impress
me with their dedication and effort. I think it was Peter Drucker
that said that the business of any business is to develop its staff.
That is what we try to do.
Wilshire:
Well Dan, I certainly look forward to your presentation in November.
Thanks a lot for talking with me today.
