|
Waiting tables: Sometimes all the information of interest to a business operation can be stored in one table, which is at the top. Sometimes not; below there is a department number field in both the Employee and Department tables. In the Employee table, the department number represents the department in which the employee works. In the Department table, the department number represents a valid department within the business. In both tables, they are department numbers; in essence, the contents of the Dept field in the Employee table represent the same thing as the contents of the Dept_No field in the Department table. It's not necessary that the linking fields have the same field names. What's important is their value and what they represent. |
From back office to front page,
relational databases are a must
Newspapers. Information. Two peas in a pod
In some respects you could say that newspapers dispense a pseudo flat-file database to subscribers everyday. It even has links and cross-references.
But flat-file databases no longer serve the needs of a newspaper in the midst of pagination. Aside from that, the non-publishing side couldn't grow in today's financial climate without a functional, relational database.
Databases fuel sales, advertising, circulation and, above all, marketing. Without reams of collected data waiting patiently in rows and columns that are tucked into tables, no one would be able to mine those data and uncover that forgotten or neglected subscriber or advertiser.
Newspapers revolve around information in more ways than just dispersing it for public consumption. Information is data and these data need to be stored and retrieved on a continuous basis. Newspapers, like all organizations today, use relational databases to drive their business needs.
Relational databases store bits of information within fields that are housed in a table. A database can contain multiple tables that can be related to one another with identifying keys or shared data fields. A relational database management system has been called "a very big spreadsheet that many users can simultaneously edit and update."
IBM researcher Edgar (E.F.) Codd is credited with creating the concept of a relational database in the early 1960s. His idea relied on multiple tables that contained separate information but could be linked to each other. This design provided the database with relational functionality without changing the appearance to the user. In 1985, Codd published an article proffering 12 criteria for a full relational database. Among the criteria are "The Guaranteed Access Rule," "Systematic Treatment of Null Values," "High-Level Insert, Update, and Delete," "Integrity Independence" and "Distribution Independence."
Newspapers have a couple of distinct database needs: storing data to publish the paper and storing data related to the business of publishing the paper. It does not make sense to store all this in the same warehouse, so to speak. But it does make sense to rely on the same database for related information -- that which will be published and archived in one database, everything else in the other. In most cases, newspapers are running multiple database systems on the non-publishing side of the business. A few even run multiple databases on the publishing side.
Storing data has never been a problem. Retrieving and sharing information, on the other hand, has been a problem. Relational database management systems provide the mechanisms through which related data can be shared in a timely fashion.
However, robust database systems are relatively new.
Early database models include the hierarchical model where data were stored -- that's right -- in a hierarchy. Tables were linked via pointers and relationships were those of parent-child.
There also was the network model. The model structure resembled an inverted tree with branches, and relationships were those of owner-member.
Modern newspapers systems need more than a flat-file database to handle pagination.
Zoning meerkats
Today, just about anybody can set up a relational database system, given their easy-to-use interfaces. But setting up a bidirectional system that knows that the photo of a family of meerkats belongs on B3 City Edition vs. B3 South Edition are vastly different.
While the information to put in a newspaper is rather the same from paper to paper -- text and graphics -- the ways in which to store, retrieve and publish these data are myriad. But fear not, because the number of different databases offered by newspaper system vendors is but a handful.
The confusing part is how they tell you your data are stored and retrieved. In addition, maintaining and administering a large relational database management system is serious, full-time work.
Yet, does a database make or break a system?
Most of the newspaper industry suppliers rely on well-known databases to drive their systems, and the database itself would seem to be an important consideration in choosing a system that would have to be supported on the back-end. The vigorous databases out there now include Sybase, IBM's DB2, Informix (which is now owned by IBM), Microsoft SQL Server and of course, Oracle. Of these, Microsoft's SQL Server is the least robust, but many companies have been able to make good use of it just the same.
When looking at today's larger newspaper systems suppliers, customers are more likely to encounter Oracle, Sybase, Informix and DB2. Yet one supplier has chosen a different path -- it went with the database called Progress.
Progress Software Corp. of Bedford, Mass., offers Progress 9.1 as a viable high-end business relational database. It handles linear scalability, XML (eXtensible Markup Language) and asynchronous processing, in addition to a variety of tools. Publishing Business Systems (PBS) of Des Plaines, Ill., offers its advertising and circulation systems with Progress.
"Progress may not be the most well-known database," said PBS President Steve Smith. "But we have plenty of customers using it."
Smith said PBS went with Progress because engineers were looking to be able to scale their product from small to large installations. He also said Progress chose a path of selling to the developer as opposed to the end user.
Smith said he has been happy with Progress, and PBS has been able to sell plenty of systems in recent years. "Progress is a business processing bullet," Smith said. "They went after business processing."
Smith also pointed out that Progress is lower in maintenance than other database systems in terms of data space and administration responsibilities. He also likes the fact that Progress offers native-level interfaces into other databases.
Maximum functionality
At Neasi-Weber International of Van Nuys, Calif., the flavor of database is Oracle.
These days, Oracle, from Oracle Corp. of Redwood Shores, Calif., gets a lot of ooohs and aaahs. While it might be the biggest name on the block, it's also one of the most daunting to support.
"We only use Oracle," Neasi-Weber Chief Operating Officer Michael Brier said. "We use it primarily to provide maximum functionality to our system."
Neasi-Weber, whose market is medium to large newspapers, is also a value-added reseller of Oracle. "We take advantage of the tools within Oracle," he said.
The key, Brier said, is to learn the database and then support it. That's exactly what he says Neasi-Weber does. "We are the first line of support for our clients for everything that is in the system we sell them," he said.
Another business systems supplier using Oracle is Data Sciences Inc. (DSI) of Laurel, Md.
"We made the decision to go with Oracle in 1995," said DSI President Mark Ganslaw. "Oracle seemed to have their act together."
Ganslaw said DSI looked at Sybase and Informix. He said Oracle's support seemed superior to Sybase. He also said pricing was an issue, but not what would be expected.
"Sybase's pricing from a reseller standpoint was excessive," he said. "Oracle was more reasonable."
He said Oracle's pricing policy has run the gamut.
"Today it hasn't been that expensive to buy run-time licensing for medium papers," he said.
He added that "from a financial standpoint, Oracle is very stable, with no debt."
The stability of Informix (now owned by White Plains, N.Y.-based IBM) doesn't seem to bother Harris Publishing Systems Corp.
"We don't expect any changes from IBM," said Grady Cooper, vice president of product marketing for Melbourne, Fla.-based Harris, which was sold in August to MediaSpan Group Inc. of Durham, N.C. "We put a huge investment into Informix and don't plan to move away from Informix, but do plan to possibly offer other relational databases."
Cooper said speed was one of the prime reasons Informix was chosen.
"It was done for speed; related to that was the ability to store graphics in Blob [binary large objects] space," he said. "Informix was anywhere between four and 10 times faster at the time we chose it."
The term "Blob" generally applies to images in the database. One of the advantages of using Blobs is that they maintain the integrity of the image or document when it is in the database. With Blob, the images and other elements in the database are no longer stored with pointers or paths. Being able to handle Blob makes a big difference in an editorial and pagination system.
Blob handling is one of the main reasons Digital Technology International chose Sybase Inc. of Emeryville, Calif..
"We started with Oracle," said Don Oldham, chief executive of the supplier based in Springville, Utah. "Oracle didn't handle Blob data and graphics well. None of the relational databases did that well."
Oldham pointed out that there are major differences between a business processing database, like Oracle purports to be, and a relational database that newspapers can use for publishing.
"We were driven by the requirements to manage all the data a newspaper needs," he said. "You open a story and the photo appears with it because it's a relational database."
Oldham said that Oracle is good at handling small transactions very quickly, but it's not good at handling large objects. He also said it's popular for reasons that don't apply to newspapers.
"Newspapers just don't have thousands of transactions," he said. "Oracle has tried to solve its Blob handling with a universal database. It handles Blob data by using another object-oriented database."
However, he said IBM's DB2 handles large objects pretty well.
"I think DB2 is very good," he said. "IBM went to the effort to create a data type to handle the Blob data. DB2 may have even leap-frogged Sybase in this area."
Newspaper policies
On the other side of the fence are newspapers that have to support new systems, including the databases that drive them. Do newspapers have policies regarding the types of databases they will or won't support?
Most of the suppliers said they hadn't run into specific policies, but some said they had lost a few potential customers because of the database they did not offer.
"We've never lost a client because we had Oracle," DSI's Ganslaw said. "In fact, it's usually a big plus."
At Harris, "we have lost a few customers because of not wanting to support Oracle," Cooper said.
Oldham said DTI has run into corporations that have standardized with a database. "We've lost a couple of sales because of that stance," he said.
Milt Goldwasser, chief information officer of Copley Press Inc. of La Jolla, Calif., tackles the dilemma of the database from a different angle.
"We consider best of breed regardless of the database," he said. "I want to put more weight on the user functions than on the system management."
Keeping in line with that policy, Copley -- which publishes the San Diego Union-Tribune and other papers -- recently purchased a PBS system running Progress.
Goldwasser points to two reasons why newspapers would come up with a database standardization policy.
"One is, if you have only one database, you only need one database administrator," he said. "The second is that you want systems that will integrate easier to help the end user."
But he also says that newspapers are not likely to do enterprise resource planning. "That negates the database standardization," Goldwasser said.
He also said he sees that only a minority of newspapers actually have policies regarding databases.
Jim Hitchman, manager of newspaper technology at Tribune Co. of Chicago, said Tribune doesn't have a database policy, but "directionally we are more likely to be headed toward Oracle.
"Most of our large papers run out of platform gas," he said.
While Hitchman said his company would favor a specific database in the interest of the paper, "vendors still should know a couple of database systems for economies of scale," he said.
Suppliers and publishers tend to look at the application rather than the database. True, if the system runs on Oracle and you don't have the talent in-house, you either train someone or purchase another system. But in today's climate of multiple systems running at newspapers, finding the right talent in-house isn't as daunting as it used to be.
-- Jason Zappe, jz@colepapers.net
Data Sciences Inc.,
(301) 957-0100,
e-mail: martini187@aol.com;
Digital Technology International,
(801) 853-5000,
e-mail: dtinfor@dtint.com;
Harris Publishing Systems Corp.,
(321) 242-5330,
e-mail: hpscmktg@harris.com;
IBM Corp.,
(800) 426-4968;
Neasi-Weber International,
(818) 895-6900,
e-mail: mbrier@nwintl.com;
Oracle Corp.,
(800) 672-2531;
Progress Software Corp.,
(800) 477-6473;
Publishing Business Systems,
(847) 381-9476,
e-mail: marketing@pbs.com;
Sybase Inc.,
(510) 922-3500.
| 