Like my company, your company uses an iSeries or even an older AS/400 to run many of your standard business applications. While you may not be writing many new RPG applications and buying new iSeries software applications, there's probably a lot of money spent supporting and enhancing the existing applications for users.
That's not all. Like me, you are probably spending significant time and effort trying to make the right decisions on how to integrate everything: users, applications, data, documents, different devices, voice, messaging and more. How are you looking to do that? With middleware, of course. Today, middleware is everywhere. It's even in my cell phone.
Before you purchase anything, make sure you've made the right decision.
I know HOW to make good decisions
The best CFO I ever worked for ALWAYS made me ask three key questions for anything I wanted to invest in:
- What is the solution worth to the business?
- What does the solution cost?
- What does it take to get the solution right?
He was very sharp. And right. Answering those three questions in that order often produces great results.
Question 1. What's it worth is logically the first one. If it isn't worth much, there's no point in going on. Discovering what the value is for a solution CAN be difficult. My favorite strategy is to get the users to "own" the problem and the solution. At that point, they'll normally produce reasonable estimates of the solution's value and how they'll generate it.
Question 2. What does it cost is not always simple. Total costs for a solution include the operating costs (maintenance fees, support costs and more) as well as the implementation costs. Hidden costs include time spent by in-house staff that wasn't budgeted or tracked. It's not always easy to get the complete costs for a solution from a vendor up front either.
Question 3. That was ALWAYS the show stopper. He'd look me straight in the eye, pause and then ask what it takes to get the solution right and WAIT for me to respond. We both knew that the solution offered a net value to the business at that point. I knew it was my LAST real chance to set his expectations on what we needed to make the solution work. So I always had to think VERY carefully on what we needed to do to eliminate all risks from the project.
How well did this process work? As an IT department we were able to "bat" better than .800 on hitting our projects on time and budget.
Real challenges in middleware decisions
The key to all three questions is getting ALL THE RIGHT INFORMATION specific to your situation. That's not easy to do in many cases when I'm looking at middleware.
Making smart middleware decisions ISN'T the easiest task I've been faced with. Depending on the problem I'm trying to address, my challenges may include the following:
- Lack of understanding on what the user's are trying to accomplish.
- Knowing what solution options exist NOW in our CURRENT environment. (Yes, it's actually POSSIBLE to not know that I already own the software I need for a solution!)
- Finding the other options are available in the market.
- Getting accurate information on the TOTAL costs for a given solution. (In some cases, I can't get ANY prices from IBM without calling a business partner.)
- Understanding the specific keys to successful implementation in our company.
- Identifying the key risks in the project or technology we need to manage.
- Knowing the real net value for the solution in our business.
- Recognizing who has the skills to deliver the solution correctly.
A simple example: Programmer Web access
Our goal was to find the best solution for programmer Web access to our iSeries server. We wanted to help our staff (four of us) work productively from home and avoid spending long hours at the office. It's not a big middleware decision compared to selecting a Web application server solution, but it is representative of the decision process many of us go through.
Some challenges and constraints we faced:
- Our current environment had a single T1 link to the Web from our network through a firewall.
- We don't really have Web networking experts on staff.
- I hadn't allocated any budget for this in the current year.
- The only standards we set were for TCP Web access and a 5250 emulator.
- Getting good information on all our options.
- Web security and authentication.
Information sources we looked at include IBM iSeries documentation, IBM redbooks, IBM magazines such as eServer Magazine and iSeries s400 Experts Journal, iSeries conferences (IBM, Devcon, COMMON), iSeries Web sites such as Search400.com, two business partners, several iSeries vendors and a friend of mine.
Problems with collecting information: Where do you go for it? Which information sources do you trust? What if they have conflicting info? Where do you go for help implementing? Where do you go for support?
Like many middleware decisions, we found four types of options:
- Use OS/400 and the accompanying no-charge iSeries software
- Buy more IBM software:
WebSphere, MQ, Portal server, business components, Rational, Tivoli, Lotus Workplace
- Use open-source software
Apache, SourceForge.net, Gnu, other sources
- Buy third-party software
Of course, IBM has more than one solution to sell for many situations. For user Web access to 5250 applications, IBM offers WebFacing, HATS-LE, Host Client Access, iSeries Access for the Web and secure Telnet. For programmer Web access to 5250 sessions, IBM offers HATS-LE, Host Client Access, iSeries Access for the Web and secure Telnet. WebFacing doesn't support system screens, so it's not an option here.
Choice 1: OS/400 no-charge software
- Run the Telnet server.
Security can be done with SSL support with an encryption program on the iSeries, an iSeries cryptographic hardware card or a simple, front-end Apache secure, reverse proxy server on Windows or Linux.
Choice 2: Buy more IBM software
- Buy iSeries Access, which includes iSeries Access for the Web ($7,500 and up to start, processor-based).
- Buy Host Access Client package ($300 per user).
- Buy HATS and Host Publisher (approximately $10,000).
Choice 3: Use open-source software
- We didn't find any open-source 5250 server software.
- Any standard Telnet client can work with the iSeries Telnet server, but they don't work well with 5250 keyboard maps. We didn't find any open source 5250 clients, but there were two shareware versions: Telnet400 by Albert York and Mochasoft 5250.
Choice 4: Use third-party 5250 software or hardware
- PC Anywhere
- Bosanova controllers
- Jacada, Seagull
The results of our decision process:
- IBM pointed us to buying IBM software. (No surprise.)
- They didn't bother mentioning the Telnet option that was free.
- We chose to use the Telnet server and purchased the Mochasoft clients (only $100 total).
- It took less than a day to setup the Telnet server with SSL security and test it over the Web to our remote Mochasoft clients.
What I've learned
Yes, IBM produces tons of marketing and technical information. Yes, business partners are always ready to help sell me a solution. But those aren't necessarily enough.
I can get good answers from IBM, but only if I ask the RIGHT IBMer the RIGHT question. However, I haven't found IBMers who can give me all my options. I've also found that sometimes it's smarter to PAY for help in making a decision than rely on the FREE help I get from vendors.
Also, my friend was right. Starting with the Telnet server and the Mochasoft client was affordable. If we need to revisit this decision later, we haven't spent much time or money proving the Telnet solution works. And it's the same solution that can work with Windows, Linux or Unix servers if we need to do that later.
What are your middleware decision challenges and solutions? How do you define your criteria for selection, your options, your implementation strategies for middleware. Write me at firstname.lastname@example.org.
About the author: The Value Manager is an IBM iSeries IT manager trying to make the right decisions to deliver better value for his company. He welcomes your comments and feedback. E-mail him at email@example.com.
- iSeries Access Family -- A good product you may not need
iSeries Access is a far cry from what it was not too long ago. Past quality problems have been eliminated, and the package is filled with new functions and software. But is it the best option for you? Not necessarily. The Value Manager looks at the features included in iSeries Access and the alternatives available.
- WebSphere: Don't believe everything IBM tells you
Each time a new version of WebSphere application server rolls out, IBM says that server is state of the art and no other server compares. Is it really the best option? Not necessarily. The Value Manager found other servers may be better worth your while.