|

|
|
First
American Real Estate Solutions (FARES) is the nations largest
provider of real estate information, serving more than 50,000 customers
with the most current data available by any competitor. Our databases
of information provide a broad range of information, including
property profiles, maps, valuation models, legal and vesting information,
and others for 85 percent of U.S. counties. FARES manages these
very large volumes of information on the latest technology platforms,
ensuring timeliness and accuracy of all data.
Such large systems, storing multiple Terabytes of data, with daily updates require a data migration tool that is capable and efficient. FARESs
search for such a tool, after careful analysis, led to the selection
of Treehouse Software as the solution provider of choice. Using
their Data Propagation System (DPS), we were able to solve the
problem of moving large amounts of data from our Data Warehouse
Platforms to our Delivery Systems.
DPS
allows for multiple platform, multiple database data migration
and mapping of our data. Treehouse has worked very closely with
FARES to improve the efficiency of these processes. Most recently
by addressing our specific requirement, they were able to reduce
processing time by 70% along with systems resource efficiencies
to these tools. Treehouse tools have enabled FARES to improve the
time to market and refresh rate of its data and are directly contributing
to the competitive advantage of FARES.
As you can see, FARES and Treehouse have a partnership on
this effort, not merely a customer/vendor relationship. We help
each other, and Treehouse is continually checking on our processes
to ensure we are happy.
Treehouse
has been a good, responsible partner for us and we would recommend
them to others trying to solve the ADABAS to Oracle problem. During
our initial implementation, we ran into difficulty getting all
data loaded to our system. I contacted President George Szakach,
and his team worked with our technical team to come up with a solution
that helped us complete the initial data load."
Dennis Gilmore
Group President of Database Information & Services
First American Corporation
|
The following is a recent discussion between Fran Stevens, Vice President of Information Technology at FARES, and TSI Project Managers. This interview was coordinated by Tina White, Production Manager at FARES.
Fran,
can you provide a business description for the First American Real Estate
Solutions (FARES)?
We are the nations largest collector and provider of real estate data, serving more than 400,000 users who depend on us to make decisions every day. We collect data on one hundred million properties annually and on two million property and mortgage transactions each month. In addition, we also provide access to three billion document images using the industries most innovative technologies.
Please
describe your core ADABAS Application, Advantage?
Advantage is a system of over 5000 database files containing Real Estate information on many counties across the nation. The information includes ownership, values, features, and a host of other information used by realtors, appraisers, direct marketers, and many other types of businesses.
How
many counties does Advantage currently support?
We currently support over 600 counties nationwide in Advantage.
Can
you elaborate on the Database and File environment for Advantage?
For each county in our system, we maintain separate files for sales, building, and land (parcel) information. This information comes to us from a variety of sources, the most important of which are the counties themselves. We typically receive and process information for every property in a county once a year. This may consist of from one to thirty files that we match and combine into a usable set of database files. In addition, we regularly receive new sales that are recorded in the counties. This information is keyed off-shore and then loaded to our system.
Please
describe the business objective for the Keystone Project?
Our objective with the Keystone Project was to move away from our Legacy Systems to a relational database that would enable us to handle more concurrent on-line users, add additional county coverage, and give us the ability to provide more complete and robust on-line products.
|
|
"Prior
to the DPS V4.00 upgrade,
we were starting to plan
a mainframe upgrade, which
is a very costly proposition
for our business. Once the
V4.00 upgrade was completed,
we recovered enough processing
cycles to cancel our upgrade
plans.”
- George
Livermore
President
FARES |
|
|
| |
|
|
|
|
Please
describe the ORACLE application, Keystone?
Keystone is a set of ORACLE databases and server applications, currently running on SUN Solaris hardware. This system provides the back-end for an array of on-line products, both dial-up and Internet, and it is also the source for our CD-based products.
How
many Keystone customers do you support?
We currently support more than 50,000 customers.
Please
describe the Keystone Project and how you arrived at the selection of
the TSI product solution?
The Keystone Project was set up as a new data delivery system that takes advantage of the data manipulation done in Advantage. This new data delivery system was designed to handle our growth over the next 10 years.
What
were the most significant factors on selection of tRelational and DPS?
The
most significant factor was the movement of data
from ADABAS to Keystone, which became critical
to the success of the project. We looked at writing the necessary
programs and also researched other companies that provided a solution.
TSI was willing to come out for a day and discuss how their product
could assist us. It seemed like a good fit and we were pleased
with TSIs
responsiveness to our questions. In addition,
TSI had the ability to deal with repeating groups from ADABAS,
read ADABAS Plogs, and create updates for daily processing.
Can
you please describe the ADABAS Annual County Refresh processing and the
ramifications to Keystone?
We
face the continual challenge of having to reload
each county once a year, while also providing
the most recent sales updates to our customers. We do this, in
part, by loading the new files from each county into a staging database, while updating the sales information to our production databases. Once we have a county loaded to Advantage staging, we process the files through DPS to load them to Keystone staging, where we can perform detailed QC and analysis to ensure that we provide only the highest quality data to our customers. When we have verified that the new files are correct, we bring the sales up to current, move them to Advantage "production, and process the files through DPS to load them to Keystone "production.
What
were/are the most challenging aspects of the Keystone Project?
We couldn't move data fast enough from Advantage to Keystone. With the implementation of DPS V4.00, we have seen an improvement of 70% in our processing times.
How
did/does tRelational and DPS factor in the success of the Keystone Project?
DPS and tRelational allow us to move data from our flat-file data repository (Advantage) to our relational on-line system (Keystone) with confidence and in near real-time. Many of the transformations that occur in this process are complex and would not be easy to develop or maintain with standard programming tools. DPS provides the capabilities and the interface to make, or change, these transformations quickly and easily.
Are
you satisfied with the products, support, and services TSI provides?
Yes, we have been extremely satisfied with the level of support and services TSI provides. Your support team is always willing to go the extra mile to help resolve any issues we may be having. Additionally, your support doesn't stop once a resolution has been identified. Your support staff always follows through to ensure the solution provided has resolved our issues.
TSI Overview of the FARES Project
FARES contracted MCI Systemhouse (now EDS) to design Keystone and implement an efficient ADABAS data transfer methodology. After formal review of alternatives such as ADABAS Utilities (e.g., ADACMP and ADASEL), Triggers, and NATURAL extracts, MCI Systemhouse contacted TSI to discuss tRelational and DPS. Following a technical review and a three-day pilot, FARES purchased tRelational and DPS.
A primary factor for selection was not limited to the value of the DPS Propagation (Change Data Capture) but the need to reload each county annually. DPS Materialization provides an efficient Extract ,Transform, and Load (ETL) for any number of counties each day, often resulting in the refresh of millions of ORACLE rows.
At
that time, FARES maintained information
for over 300 counties. Currently,
FARES maintains over 600 counties, with anticipated continued
growth. Each countys data is partitioned to separate physical ADABAS files over 20 production databases. To satisfy the need to maintain a single logical model
per file definition, TSI enhanced
tRelational to provide the translation
and substitution processing of DBID and File Numbers during generation
of the DPS parameters.
Positionality (occurrence value) within Periodic Groups (PEs) had significance in many cases for the Advantage files. This required explicit positional mappings of PEs to ORACLE tables and columns.
During the development, tRelational enabled quick and easy revisions to the ADABAS-to-ORACLE mappings. These changes resulted from modeling revisions to Keystone and system re-engineering of Advantage. A programmed solution would not have permitted such a timely response to the changes.
In summary, the Advantage database is large in volume and complex in nature. Without the functionality and efficiency of tRelational and DPS the data delivery requirements could not be satisfied.
by Daniel Sycalik
I
accompanied Dan Sycalik on this initial trip
to FARES, as the TSI management/sales representative,
with Dan being the technical representative.
I
was happy to see about 16 experienced
FARES and MCI Systemhouse personnel
in our technical discussion, all of them engrossed in what Dan
had to say. Dan did an excellent job of telling them what tRelational
and DPS could, and at that time could not, do. The anxious looks
on the customers' faces told me, if Treehouse can't bail us out, we have a severe problem, and SAG is unlikely to be a part of the future at FARES we might as well move things to ORACLE now.
We promised to do what we could to ensure FARES could keep their huge investment in their application and in the SAG base products. After all, we can't possibly sell them TRIM, RACE, PEEK, N2O, etc. if they don't want to continue with ADABAS.
Dan and others got the products up and running quickly, fine tuned them, and did some external transformation routines to get us 99% of the way. What the customer deemed to be impossible to do with any other data extraction/movement process, due to the sheer volume and complexity, we were able to accomplish practically overnight.
And, I could see that we could accomplish all of the extras that they wanted quickly (with a flurry of development activity) using our good product framework.
Still,
FARES wanted better performance.
Some situations unique to FARES made
us rethink the products. FARES would have 600 sets of three identically
defined files. Rather than 1800 sets of identical-appearing parameters,
could we do a shortcut? Dan Sycalik and Dan Vimont, product designer/developer, and other technicians got together and decided, sure we could.
As
we found out at FARES, flexibility alone
can sometimes be a liability, causing
performance problems. Software companies that have products with
great flexibility stand to benefit from analyzing the customer's
use with its own live data. In our case at FARES, we did just
this, developing the right mix of "practical flexibility" and
tremendously improved performance.
And in DPS Version 4.0.0, we made another
quantum leap in performance.
The things we did for FARES, we did gladly, knowing these features would add to the value of the overall product set and benefit future customers.
This blending of ADABAS data into an ORACLE environment is something we know other large SAG customers need. TSI is ensuring that sites stick with ADABAS well into the future by efficiently disseminating ADABAS data to ORACLE and other RDBMSs.
by George Szakach
|
|