|Accounting Software Series
on the Web
This entire series of articles on accounting software appears on the AICPA Web page at www.aicpa.org/pubs/jofa/index.htm
| || |
|J. CARLTON COLLINS, CPA, is a partner of K2 Enterprises, Atlanta, a professional and consulting organization that provides technology continuing professional education. His e-mail address is firstname.lastname@example.org. K2 Enterprises's Web page is www.k2e.com.|
ive years ago, few buyers of accounting software gave much thought to the underlying database—after all, they were pretty much stuck with the database engine the developer had designed into the product. Today, however, many accounting vendors have engineered their applications to accommodate a variety of database engines—a major advantage for the user. Select the database that matches the user's unique needs, and not only can it save lots of money but also the accounting package can run faster and better—even allowing it to be scaled up as the business grows.
As a result, it's now imperative that prospective buyers give careful consideration to database selection, which depends mostly on the volume of transactions processed across an organization's local or wide area network (LAN or WAN). In general, the more transactions, the more robust the database needs to be and the higher its cost in terms of purchase price, implementation and maintenance. Exhibit 1 summarizes this information for some of the more popular databases.
What gave users the flexibility to select a database application was the decision by most top accounting software vendors to separate the so-called business logic part of their programs from the underlying database. For example, Great Plains sells its standard version of Dynamics, which runs on either the Btrieve or C/tree database, for the average retail price of $5,000; the higher end version, Dy namics C/S+, operates with the Microsoft SQL Server database and sells for about $50,000. The two accounting programs share similar programming code; however, the underlying database used in Dynamics C/S+ is far more powerful—accounting for the higher cost of the system.
SEPARATE IS BETTER
A significant advantage of separating the program's business logic from the database is that it allows vendors to easily adopt new database technologies as they emerge. For example, databases of the future are expected to offer better Java support, better functionality across the Internet, improved security, better support for handheld devices and thinner applications—that is, those that rely more heavily on the power of the server. It's likely, too, that tomorrow's databases will exhibit better compression support for the storage of wave (sound), image, video, fingerprint and retina scan files. So it's important to be able to upgrade a database without a major rewrite of the accounting software.
It's sometimes to the user's advantage to select the most powerful database it can afford. For example, earlier this year, Open Systems introduced a version of Traverse that operates on the Microsoft SQL Server 7.0 database. Its own testing showed that running Traverse on the Microsoft Access database required more than six hours to process a transaction that consisted of 85,000 records. The same transaction was processed in just five seconds when running Traverse on the SQL Server 7.0 database.
Although these are impressive results, don't conclude that the more powerful databases are always the best option; such a conclusion can be very expensive. For example, historically the decision to implement an accounting software package on the SQL Server database added about an additional $100 per user to the cost of the accounting system. Most often this additional cost was buried in the overall price of the software. Assuming your company has 250 users, the additional cost of the database application would be $25,000.
But those aren't the only costs. The typical consulting fees for installing a SQL Server-based accounting package range from $70,000 to $190,000. Although SQL Server 7.0 is said to be much less expensive to implement and administer, when this article was written 7.0 was too new to gather evidence to support this claim. In addition, there are ongoing support costs, which are much higher for complex systems.
Determining which database is best suited for a company often is difficult. The goal is to select the lowest priced database that meets customers' current and anticipated needs. While SQL Server is a popular choice, less expensive alternatives are available from Pervasive Software, which offers two products that are widely embraced by the accounting software industry: Btrieve and Pervasive.SQL. Both products perform well in the middle market and get the job done for a fraction of the cost—for example, $49 per user for Pervasive.SQL compared with $160 per user for SQL Server 7.0. Further, based on historical figures, the consulting fees and ongoing support costs for the Pervasive products appear to be 50% to 75% less than those for SQL Server, Oracle8 or IBM DB/2. See exhibit 2, below, for a summary of the databases supported and programming languages used by the leading accounting software packages. Microsoft currently offers three editions of SQL Server: Back Office Small Business Server (SBS) is limited to 50 users and 10 gigabyte databases; both SQL Server 7.0 and SQL Server 7.0 Enterprise contain no such limitations. Microsoft says the installation costs of these editions will be lower due to reduced administration requirements.
THE ACCESS ADVANTAGE
At the lower end of the market are less robust databases, in terms of speed and performance, but they typically offer strong end-user tools and are easy to use. For example, Microsoft's Access is excellent when it comes to integrating accounting data with the other Microsoft Office products.
Because it's written in Access, Traverse inherits a host of powerful functions. Reports produced by Traverse can be sent directly into Word or Excel with the click of a button. Reports also may be e-mailed directly to individuals or groups, avoiding the costly process of printing and distributing the report. Traverse also can print any report to an HTML format, which makes the process of publishing financial or accounting data on the Internet or intranet much easier. Companies requiring a more powerful database can implement Traverse on SQL Server.
Aside from price and performance, there is another important difference between the middle-market and the low-end databases: Higher end databases offer users client/ server functionality. By definition, a client/server system involves the delivery of data from a server computer to a client workstation. However, client/server architecture makes the database operate faster even on a single computer system. In that case, the delivery of data from the computer's hard disk to the application running on the computer is more efficient and, hence, faster. Therefore, client/ server architecture offers two types of performance gains: efficient record delivery (it's often referred to as ERD) on the single computer or server computer and client/server delivery (CSD) from the server computer to the client workstation.
Because client/server benefits can be impressive, many customers conclude it's always the best option; this can be a costly assumption. With lower volumes of data, client/ server databases actually can perform more slowly than other nonclient/server databases because of the heavy data administration overhead. As transaction volume increases, client/server databases far outperform their older counterparts. Therefore, the actual performance gains you realize at the single computer or server computer will depend on the volume of your data.
Further, if your company currently has plenty of excess bandwidth on its LAN, it's unlikely a client/server system will improve performance across the LAN dramatically; this is because a client/server system helps solve LAN data bottleneck problems. However, if your company doesn't have such problems or doesn't expect to have them in the foreseeable future, then a client/server system may be unnecessary.
If your company's wiring system is reaching overload, consider upgrading the network cards, cables and hubs. For example, you may want to explore higher capacity cabling, such as fiber optics, or the latest one-gigabyte technology called Level VI wire. This strategy, which may be far less expensive, improves the performance of all applications, such as e-mail, printing and accounting.
Companies with WANs face an additional problem. While the cost of upgrading LAN wiring is fairly affordable, not so the costs of upgrading WAN wiring. WAN connections—the cables that link LANs together—usually employ slower wiring technologies such as phone lines, ISDN lines or T1 lines. These slower connections become major bottlenecks. The cost of upgrading a WAN connection to run at the speed of your LAN can total hundreds of thousands of dollars per year. So in a WAN environment, a client/ server system is most likely the best choice.
|The Power of
To understand how client/server technology works, consider this example: A customer calls to change his address. Without client/server technology, the file server sends a copy of the company's entire customer file of 5,000 names and addresses down the wire to the order entry clerk's workstation. There the workstation sorts through the 5,000 names and addresses until it finds the appropriate customer record. After the clerk makes the necessary changes, the updated file travels back up the wire where it updates the file server. If the company used a client/server system to accomplish the same task, the file server would sort through the 5,000 names and addresses, find the appropriate customer record and send a copy of that single record to the workstation. The order entry clerk would make the necessary changes and the updated record would then travel back up the wire to update the database in the file server. In the latter scenario, because the file server sorted the records, 4,999 names and addresses didn't have to make the two-way trip, saving precious bandwidth and time on the company's LAN. In this example, the actual time savings may seem insignificant—a second or two at best; however, if you consider a large environment, with 100 simultaneous users of the system, each processing thousands of transactions per hour, the benefits of client/server computing shine. Client/server computing offers another significant benefit: distributed processing, which allows a single process--such as the aging of accounts receivable or the printing of monthly invoices—to be distributed across multiple computers throughout the organization. The client/server computer can monitor the available processing time on all computers and distribute tasks among them in order to squeeze the maximum amount of processing from the available resources. For example, Platinum for Windows uses a three-tier architecture that allows a company to set up multiple application servers throughout the organization. Together these multiple servers work seamlessly as one database even though the data are actually distributed across multiple servers. The result is faster processing time. It's also easier for software applications to make SQL queries across a network.
ACCOUNT NUMBER STRUCTURE
One feature can knock a product out of the running instantly: the account number structure. If that structure is inadequate—no matter how superior the product is in every other way—the software should be rejected. It's important to understand what's behind that statement.
Originally, DOS-based accounting systems were developed with PCs in mind. In those days, hard disks had little data-storage capacity, performance was slow and the critical RAM was significantly limited. At the time, those limitations were hardly considered a serious problem: After all, in the 1980s most accounting software developers never envisioned their products being used by large companies—that is, with sales in the tens of millions of dollars and with multiple, even global, operations. As a result, software developers didn't seriously consider the need to add additional account numbers for large organizations.
Of course, the capacity of the low-cost PC has grown exponentially, meeting the needs of even the largest companies. But when developers later tried to modify the account number structure to accommodate more segments and larger numbers, they found it was difficult, if not impossible, without completely rewriting the underlying program codes.
There are many reasons why it's important to have a larger account number structure. Today's larger companies often have subsidiaries and divisions that must be identified in the accounting system by a fourth, fifth or sixth segment in the account number. In fact, most large not-for-profits need at least four segments in their account number structures—for the account, subaccount, department and program. Many government agencies need at least four segments in their account number structures to identify the account, subaccount, department and fund. Exhibit 3 summarizes the maximum account numbers offered by some of the well-known accounting software packages.
|Flexing Account Numbers
To illustrate the importance of a flexible account number, consider how a typical high school tracks the expenditure for textbooks. The expense is debited to book account--the primary account number that describes the primary reason for the expenditure. The school also may use a subaccount to show that the textbooks were purchased for Ms. Johnson's class, thereby requiring a second segment to record the transaction. Further, the school may wish to indicate the English department purchased the textbooks—thus requiring a third account number segment. Finally, the school may want to record which funds (federal, state, local, trust, lottery) were used for the purchase. In many cases, the funds provided schools have stipulations (restrictions) attached. Because of that and because the agencies providing the funds require a report detailing how those funds were expended, it's important to track which funds were used in each expenditure.
An example of how this feature can influence the accounting software industry is illustrated by how Platinum solved the problem. Originally, Platinum's standard general ledger module provided for an 8-character, 2-segment account number—a structure that often was too small to meet the needs of large companies. In the late 1980s, Platinum created a second general ledger (called the Premier Ledger) accommodating a 32-character account number with up to 30 segments. Today Platinum sells three versions of its general ledger—priced from $2,000 to $10,000. For many years, Platinum's Premier Ledger represented the most powerful general ledger available in its price range. Today, several products offer similar functionality.
RealWorld Expertise includes as a default a 12-character, 12-segment account number with options to upgrade to a 25-character or 50-character account number. This feature, along with several other built-in fund accounting features, makes Expertise particularly well suited for government and not-for-profit organizations.
Next month, the last article in this series will focus on the Web features of accounting software; various packages' ability to handle foreign currency; how to determine whether the "look and feel" of the software meet the needs of your organization; the year 2000 bug; prices; and where to get more help.
As you can see, the job of assessing accounting software is complex. There are no easy answers, and you can't make a selection based entirely on someone's recommendation. It's rare that another company's needs precisely match yours. It's vital that each of the issues mentioned in this article be addressed carefully and comprehensively. If you skip one, it's at your own risk.