Looking for case studies

[klm]

"klm" wrote:

"klm" wrote:

"klm" wrote:
One of my clients is the development manager of a large insurance company. He runs a predominantly Solaris/Websphere/Oracle application platform. As the size of his databases increase, so too do his software licensing and hardware costs increase.

He's fairly "old school", likes running on commercial "supported" platforms and firmly believes in the "If it ain't broke, don't fix it" approach to application maintenance.

Now, here's the problem. In the face of increasing hw and sw costs, his CIO has ordered him to begin investigating open-source solutions. He is naturally resistant to this ordered change and is being dragged kicking and screaming into the Linux/Apache/Tomcat/Jboss arena.

I need some information and research data which I can use to present to him a case study of why this will work and why it makes a lot of sense.

As a long time user and advocate of open-source technologies, it makes perfect sense to me, but my common-sense intuition doesn't translate well into the sort of hard analysis that he needs to acheive an appropriate comfort level.

If you have done this, or are about to do this, please consider sharing your experiences with me so that I may create a detailed case study or two showing why Jboss makes sense from both a technical and a TCO point of view.

All replies will be held in the strictest confidence.

Thanks.

[paulogilvie]

"Paul Ogilvie" wrote:

"Paul Ogilvie" wrote:

"Paul Ogilvie" wrote:
I am a business consultant/development manager. We are developing a single application using a server, some LAN clients and lots of PDAs on a mobile packet switched network (GPRS). The application must support field engineers and central planners in managing Work Orders taht come from an external ERP-system.

Since we are developing a fairly closed application, using Java and an Oracle database, what would be the win points of using JBoss?

As a classic developer (data driven, C,...) I may not be able to grasp why my developers are so interested in using an App Server. This week (starting Jan 18), we must make our final design decissions.

Lastly, we must deliver mid March already. A range of components are ready already (Java Planboard, PDA business logic, database) and we must decide now to tie them together using classic process designs or using JBoss.

Some key reasons as to WHY we should use JBoss would be of enormous importance to me and would be much appreciated.

[nmartel]

"nmartel@ll" wrote:
Paul,

JBoss is an application server and from a functional stand-point is essentially equivalent to BEA Weblogic Server, Sun One Server (SOS), IBM Websphere Application Server (WAS), and others. If you need an application server, you need an application server.

The choice of JBoss over the others can be quite complex. Most folks seem to choose based solely on acquisition cost... JBoss is free to acquire, you just download it. You can choose to optionally pay for JBoss support but that is your choice. Sun One and Websphere on the other hand, cost $$$ to acquire.

JBoss has high-availability clustering/load balancing abilities built-in while the others presently do not. This can make JBoss attractive to use over the other servers if you need these features.

The cost of development and complexity of development is about the same no matter which server you run on. The most complex part of the choice is whether you are prepared to make your own fixes to JBoss should it have issues or whether you want to put pressure on a vendor you pay to support your server. You need to remember that open-source is mostly a help-yourself mindset (I see that as good). The commercial folks provide you with a supposedly more stable platform, but you are at the mercy of what they want to charge you for support and their time table for bug fixes, maintenance releases, and support calls.

Good luck with your choice.

[nmartel]

"nmartel@ll" wrote:
KLM,

You need to approach your client from the point of view of what it costs him to maintain his application and what he is paying for total fees in a given period of time (say that number is something like $2-$3M in fees and $1M in software maintenance).

You then need to show what it will cost in money and time to migrate to a new platform. Show the client what the new costs will be including acquiring the platform, training staff, licensing various components, actually migrating and hardening the software, cost of ongoing maintenance, etc.

Let the numbers speak for themselves. You do not need an instinctive feel to show what needs to be done. The toughest part of this job is firguring out how long and difficult the software port will be and what risk you will introduce as a result of porting to a new platform.

You might say you will save your client money, but if a migration takes the vital system down for any length of time or the cost of re-training and re-certifying staff exceeds the benefits, then the project won't happen. You need to show a plan that addresses these concerns and risks.

Good luck.

P.S. I almost forgot, I almost always see this problem. The answer is to break the migration down to incremental steps. Take baby steps and move forward slowly. This lets your client feel comfortable while you make deliberate and forward-moving progress. Start with changing our the database from Oracle to MySQL. If you can get that to fly, take another step.