A Niche Market, EAI and ETL, The Anatomy of Legacy

Readers comment on past issues.

Niche Market
I read Joe McKendrick's September 2002 Host Integration column ("Newest Integration Strategy: Wireless") and am interested in his advice on promoting a consortium of installation companies known as the National Communications Contractors (NCC). With the plethora of information IT professionals see regularly, the need for a professional, malleable and mobile team of outsourced installation contractors is either last-minute or overlooked. The NCC can be visited at www.ncc-usa.com.

—Dan Gibbons
Multicom Inc.
NCC Owner/Member
Phoenix, Ariz.

I want to compliment Stephen Swoyer on his even-handed treatment of the EAI versus ETL issue in the August 2002 issue ("Application Integration: Mix and Match.")

I agree with his basic message that the most important questions for an organization to answer are: What are my integration needs? What is available to meet them? And what tool or tools should I use to meet these needs, given the particulars of my situation?

All too often, technical publications and vendors focus on the trees by playing up the differences between classes of tools in this area. David S. Linthicum's book, Enterprise Application Integration, offers an excellent view of the integration forest. I highly recommend it as an antidote to the hype that surrounds many articles on EAI "versus" ETL, Web services "versus" EAI, etc.

—James J. Humpf
IT Architect
Columbus, Ohio

The Anatomy of Legacy
What is a legacy? Merriam-Webster defines it as "something transmitted by or received from an ancestor or predecessor or from the past." Is a legacy a bad thing?

We name airports, roads, bridges, and even entire towns after those we respect and honor so that they might be remembered by future generations. Most of us would like to leave a legacy of some kind to those who follow. In most contexts, a legacy is something positive; something handed down to be cherished by those who follow.

The use of the term "legacy" with respect to computer systems seemed, at one time, to be limited to describing applications that were not Web-enabled. However, lately I've observed that this term has been used more frequently to refer to mainframe computers in general and the applications that run on them. In far too many cases, the term is intended to impart a negative connotation and has—for many—become synonymous with obsolescence.

Negative perception, not technological inferiority, is the most important and powerful force threatening mainframes today. We should simply call a mainframe a mainframe, and mainframe data mainframe data. Saying we want to give our Unix Web application access to our legacy mainframe data sends too many of the wrong signals. The truth is, the Web application could just as easily be hosted on the mainframe, possibly with better performance and RAS characteristics.

Though it may seem ridiculous, I suggest the term legacy be removed from our collective vocabularies forthwith.

—Edward E. Jaffe
Manager, Research & Development
Phoenix Software International
Los Angeles, Calif.