Migrating Your Legacy Software

Eight Key Questions to Explore

Over the years, you’ve built a highly-successful business around a custom software application.  Early on, it was a wonderful asset.  But all the while it helped to drive your company’s growth engine, the software remained the same.

Fast-forward a decade or two, and the software is really showing its age.  While it used to run lightning-fast, today it bogs down under the increasing load of database bloat and more users than anyone could foresee.  Perhaps your IT department has begun e-mailing you dire reminders:  They can’t find compatible hardware any more.

There’s a second, related issue:  Typical software applications comprise thousands—if not hundreds of thousands—of lines of code.  Finding programmers who can (or are willing to) support your software may become a serious challenge.  (We know of several business owners facing this situation.)

What’s a pragmatic decision maker to do?  One option:  Nothing.  After all, the software still runs.  Never mind the occasional glitch now and then.  If, on occasion, you lose half a day’s worth of data because your IT staff has to restore from backup, well that’s just a cost of doing business, right?

Perhaps your exit strategy includes letting a future manager deal with the issue.  So, for now, you cross your fingers and pray that the momentum of the status quo will propel your business forward, until aging software is no longer your concern.

But what if you worry about your legacy software?  What if it keeps you awake at night?  How will you ever carve out enough hours to manage a software development project, without neglecting your core business?  How can you capitalize upon what you’ve already built, while taking advantage of everything that modern software languages offer?  And what other questions should you be asking—questions you may not even realize should be taken into consideration?

Migrating to newer software can be a daunting proposition.  Short of writing a textbook, it’s not practical to cover every possible decision that business owners must weigh.  But here are several key questions you may want to contemplate…

1)  Will the project require new hardware?

A handful of our customers are still using outdated operating systems.  Not because they want to, but because they must.  Their software simply won’t run on newer hardware.  Considering that Windows XP, Vista, and Windows 7 have all reached their end of mainstream support[i] dates, now might be a great opportunity to migrate to new hardware as well.  Do you plan to update your workstations in conjunction with your software upgrade?  Will a server upgrade be required?

As you develop a budget for new software, you’ll also want to include the cost of new hardware into your projections.

2)  On which platform(s) will the new software be developed?

The three most likely candidates are:

  1. Desktop (applications running on your PC or laptop—the ones you launch by double-clicking a shortcut)
  2. Browser-based (i.e., web sites you access via Internet Explorer, Chrome, Mozilla Firefox, etc)
  3. Native applications (i.e., smart phone apps written for IPhone and Android)
  4. A combination of all three

Each platform offers both advantages and unique programming challenges.  The following matrix spells out the relative strengths and weaknesses of each.

FactorDesktopBrowser-basedNative (Smart Phone)
User ExperienceDesktop applications utilize a wide variety of robust, user-friendly controls which make pointing and clicking a snap.

This environment is familiar territory for most end users, meaning that time spent on user training will often be minimal.

Web-based apps are designed to run on most browsers.  They usually can’t access an individual device’s capabilities.  Thus, they must be hardware-agnostic.  Beyond standard browser controls (check boxes, buttons and the like), web apps require extra coding to incorporate specialized controls and functionality.The main benefit that native apps offer over web apps is that they are fully compatible with the device’s native features, such as GPS, camera, phone, and accelerometer.  They give you access to a device’s hardware and software, allowing you to incorporate photos on the fly, access GPS information, and communicate via phone or text—all directly from your custom app.
PerformanceProperly configured, a well-written desktop application will run quickly.  Users won’t experience the slower connection speeds they might occasionally encounter on other platforms.

Smart developers will configure the application to download the majority of application content onto the local PC, meaning that the only network traffic will be data as it’s read from and saved back to the database.

Web applications require a reliable internet connection.  If your connection is slow, they may creep along much more slowly than their desktop counterparts.

This will be especially noticeable on mobile devices, because images and animations take time to download.  Graphic-intensive web applications may also consume considerable bandwidth and battery reserves.

Native apps tend to be faster and more responsive.  After a native app has been installed, most of the graphics and code that run the app are already stored on the device.  Thus, the time and bandwidth required to access content on the server will be lower than a similar browser-based application.  As a result, native apps are often much faster, and provide a better user experience.

They can also be configured to automatically sync data stored locally on the phone with a web-based server when the device isn’t in use.

SecurityMost often, your application data is stored within your Local Area Network (LAN) or secure cloud-based storage.

Provided that your IT staff has configured your firewall and VPN properly, hacking into the database will be considerably more difficult than other platforms.

Any time your company allows access to content over the internet, it becomes slightly easier to exploit the connection to your database, and thus easier to compromise your data.  At a minimum, you’ll want to insist on using Secure Sockets Layer (SSL) encryption.

For added security, if IP addresses of end users are static, you may want to have your IT department restrict access for all unknown IP address.

Think of how much sensitive data lives on your phone (contacts, e-mail addresses, phone numbers, etc.).  Mobile devices often connect to the internet through untrusted WiFi connections, which malicious eavesdroppers can readily exploit.  If one of your users loses his phone or sells it on eBay without resetting to factory defaults, it’s possible that your corporate resources could be compromised—especially if users select the option to “Remember My Password”.

These risks can be mitigated somewhat, by implementing a strict security policy and deploying logic on the server to prevent device spoofing.

ExtensibilityVirtually all device manufacturers provide drivers for their peripheral devices.  This makes it relatively easy to integrate desktop applications with printers, cameras, scanners, RFID readers and the like.In earlier days, browser extensions and plug-ins (such as ActiveX objects) unwittingly opened the door for security breaches.  As a result, most web browsers no longer grant access to computer resources, such as file systems or peripheral devices (at least, not by default).

If your company wants to develop a web application that interacts with users’ hardware—or automatically creates, prints, e-mails and/or saves files, please give serious consideration to either a desktop or native app instead.

Many smart phone manufacturers provide hardware ports to allow for interaction with external devices.  One would think it should be easy for programmers to write code to communicate with accessories.  But this is not always true.  The device manufacturer must provide a set of driver files conforming to the type of smart phone(s) you plan to use.

The bottom line:  if you need to link an external sensor or other device to your native app, make sure that the manufacturer provides compatible drivers.  Please don’t assume that your programmers will figure out how to make it all work.

Program- ming ComplexityOf the three platforms, desktop apps are usually the fastest to develop and deploy.

Desktop applications are programmer-friendly.  Many programmers learned how to program in this environment, and are very comfortable developing applications for the desktop.  Sometimes it’s simply a matter of dragging and dropping the desired controls on a form, eliminating the need for writing additional code.

As a result, the overall cost of a desktop application will likely be lower.

Today, web applications frequently utilize what’s known as the MVC (Model-View-Controller) design pattern.  This approach allows for efficient code reuse and simultaneous development by multiple programmers.  But it takes time to configure properly.

In fact, web applications almost always require more time to set up, configure and test than a desktop app.  Consequently, the cost for web applications is often higher than a similar application designed for the desktop.

One of the greatest challenges in developing smart phone apps is that we must develop a separate code base for each platform.  Each mobile vendor provides its own Software Development Kit and user controls, which must be used to develop apps for that device.

Thus, if your company wants to deploy an iPhone app, programmers will develop and upload an iOS app to Apple’s App Store.  Need an Android version of the same app?  Programmers will write a separate set of code for Android, and then deploy it to the Android app store.  As a result, native apps are often the most costly of the platforms.

However, native apps usually run faster on smart phones and are better equipped to harness the capabilities of the hardware on which they’re installed.

Installation and accessDesktop applications are usually installed on your company’s local area network (LAN).  If users must access the application from outside your office, they’ll need to use connectivity software (Remote Desktop, for example) to gain access.  This isn’t necessarily a bad thing—especially if security is a concern.

Installing a desktop application on a new PC can be a very simple process, provided that the developer configures the necessary tools for doing so.

 

Since most of the code associated with a browser-based app runs on the web server, no installation is required.  Just navigate to the site and you automatically have access to the latest content.

Web applications rely on consistent internet connectivity and speed. The lack of either can cause significant issues.  This is especially noticeable when running web apps on smart phones.  Users may have to wait for the browser to download all the content.  If your application incorporates numerous photos or graphic images, you may want to consider developing a native app instead.

In addition, web applications require extra effort on the part of end users.  They must remember your specific web address (or find it on a search engine).  If users fail to bookmark your site, they’ll need to repeat the entire process any time they want to connect with your business.

Mobile apps are usually distributed via App Stores.  When users install a native app, they will only have to remember the loading instructions long enough to download it.  From that point on, your products and services will be just one click away whenever they use their smart phone.

Native apps usually run faster than web-based apps.  Once installed, they don’t have to continue downloading the same resource-intensive files over and over. Thus, they often use significantly less bandwidth.

UpdatesDesktop applications can be configured to automatically update to the latest available version with little or no user intervention.

But not all desktop applications function this way.   If you opt for a desktop application, be sure to ask your programming team how they plan to deploy future updates.

Since most of the code runs on the web server, updates to web applications download automatically whenever you navigate to a new page.

End-users do not have to bother with installing updates because they’re managed by the web server administrator.

Native apps user may need to download and install updates periodically.  In some cases, users may be prompted to completely uninstall the previous version prior to installing a later version—but nowadays, this is rare.

In the world of custom software applications, one platform definitely doesn’t fit all needs.  In fact, your ideal solution may involve a customized mix of platforms and programming languages.

3)  Which programming language(s)?

Beware the programmer who insists that his language of choice is the absolute best—and there’s no room for debate.  As technology advances, programming languages continue to evolve.  Some fully embrace newer hardware architecture and software development patterns.  Others are headed for retirement.

For example, Visual FoxPro (VFP) is a wonderful, robust programming language.  We still support over a hundred custom applications written in it.  But its vendor discontinued support a few years ago.  Today, we wouldn’t suggest migrating even an ancient 16-bit application to VFP.  Instead, we would likely recommend C#, PHP or ASP.NET.  Why?  The chief reason:  programmer availability.  At some point in the future, you may decide to entrust your software development to a different programming company.  That will be a serious challenge if you can’t find anyone who speaks the language.

So, rather than letting an obscure language bind your company to a programmer for decade(s), you’ll want to opt for a programming language that’s widely supported.  Spending a little time on this question up front will help ensure you aren’t saddled with a decision you’ll regret for years to come.

Here’s a useful site to visit when considering software developers and the language(s) they’re proposing:  https://www.tiobe.com/.  The TIOBE Index compiles a monthly list of programming languages based on popularity and the number of lines of code written.  If you can’t find the programmer’s proposed programming language in the top 50, you might want to consider finding a different programmer.  Speaking of programmers…

4)  Which programmer(s)?

One thing I’ve learned over decades of hiring (and firing) programmers:  just because someone enjoys tinkering with code doesn’t mean s/he’s qualified to tackle your project.

You’ll want to dig deeper—looking beyond flashy marketing handouts and web sites—to carefully examine the competency level of the programmer/team you hire.
How?  Here are a few suggestions:

  • Ask for references. Then, follow up with each reference, verifying that software users are truly happy with both the quantity and quality of code written.
  • Schedule a screen-sharing session to review examples of applications the programmers have developed which are similar to yours.
  • Ask whether the coding will be outsourced, or performed in-house.
  • Find out whether programmer(s) will be available to communicate with you in real time, or whether time zone differences will necessitate more staggered communications.
  • Ask what steps the programmers take to avoid SQL injection (they should surely understand your question—even if you don’t).
  • Finally, be sure to verify who will own the source code[ii] once the project is complete.

5) Which database?

According to a recent DB-Engines[iii] survey, the five most popular database engines are:

  • Oracle
  • MySQL
  • Microsoft SQL Server
  • PostgreSQL
  • MongoDB

Each has inherent advantages and disadvantages in terms of cost, speed and support.  Your company may already have settled on a standard.  If so, you have one less thing to worry about.  If not, ask your developer for recommendations, and reasons for those recommendations.  Be sure to consider the total cost of ownership.  Database costs fall into two expense categories: capital and operational.  The cost of database licenses, and installing and configuring the database can vary widely.  Opting for an open-source database may save money initially.  But lack of support (and therefore, higher maintenance costs) might offset any benefits you gain up front.

6)  Should we migrate our data?

Migrating existing data to your new application will ensure that users can find historical data for orders, invoices and inventory without requiring them to launch the old application.  It’s usually the best approach.

Data migration can also burn up a lot of programming hours—and dollars.  Why?  Because frequently, older tables aren’t structured very well—the former application may have deliberately duplicated data across multiple tables.  But subsequent updates to some of the tables—without updating the duplicate records in their counterparts—can leave us scratching our heads and asking, “which data should we migrate?”

Secondly, earlier developers often used a combination of fields to ensure uniqueness.  They just assumed that every record will be unique because of its content.  After all, how many John Smiths could possibly have the same birthdate?  Too often, these assumptions were wrong.  When a database is designed poorly in the beginning, migrating to a more logical structure can be a real challenge.

Because of these issues, and the extra programming time required, some customers prefer to ‘Drain the Swamp’.  In other words, keep using the old application until outstanding invoices are paid, purchase orders have been received, and so on.   But once a cut over date has arrived, all transactions will flow through your new application.

7)  Will the new application talk to other software?

Today more than ever:  no software is an island.  To prevent re-keying the same data into multiple applications, software developers build electronic bridges to seamlessly pass data back and forth.

Most software applications we write communicate with multiple systems, interfacing with one or more of the following:

  • Accounting software
  • Scheduling systems
  • Time clocks
  • Payroll services
  • Real-time inventory management
  • CRM systems
  • Payment processing services
  • Phone systems / billing software
  • Web services (API’s)
  • Web site scrapes
  • File imports, including CSV, Excel, XML, JSON, SDF
  • E-mail
  • SMS
  • Document scanners
  • RFID / Barcode Scanners
  • Document generation, archiving and transmission structures

Software interfaces can be quick and easy to write.  They can also be extremely complex.  The more you identify and document your integration needs up front, the more complete road map you’ll have.  And the less likelihood of future sticker shock.

8)  Should we migrate everything all at once?

Whenever possible, we recommend migrating your application in phases.  If your legacy application spans several modules that run independently of each other, it would be wise to divide and conquer—migrating modules one-by-one.

This will give you the chance to assess your progress, and course-correct along the way.  It also gives you more budgeting flexibility—possibly allowing your company to stretch expenses over a longer time frame, as each new phase comes online.

The Upshot …

Migrating a legacy software application is not for the faint of heart!  We’ve explored several of the weightier questions you’ll want to consider.  No doubt, you’ll identify many more as you move forward.

How you arrive at the answers to your questions will depend on your company’s specific needs, your budget, and time frame.

We wish you all the best.

[i] Microsoft Corp. Windows lifecycle fact sheet Accessed July, 2017.   https://support.microsoft.com/en-us/help/13853/windows-lifecycle-fact-sheet

[ii]I/O Technologies, Inc.  Who Owns Your Source Code?  Accessed July 2017. http://www.iotechno.com/iot/who_owns_your_sourcecode.pdf

[iii] “DB-Engines.” The Most Popular Database Management Systems, July 2017. Accessed July 2017. https://db-engines.com/en/

Leave a Reply

Your email address will not be published. Required fields are marked *