Compiere from the Source

Agile Development - the cure all?

Kathy Pink: I just completed a 2 day class in Agile Development aimed at Product Managers.  While I have a good understanding of the concepts and practices of Agile Development, it is always a good idea to validate what you know (or think you know).  Also it helps to understand how others are using the processe and what issues and concerns they have.

What surprised me was that some people look to the Agile Development methodology as the 'cure all' or panacea for their devleopment woes.  No development methodology can make a disfunctional organization functional.

Cloud Hype and Reality

Cloud computing is "in" - and actually with the latest price change from Amazon it is becoming a bit more real.

I signed up with Amazon, installed the Java tools, created my own image, changed it via Windows Remote Desktop, saved that image and re-started that image - in half a day.  Far easier than I thought - just another half day to install database, application, etc.  As they have nice Java interfaces for the operations, I think that the entire process can be automated (or you could use the Management Console).

Realistically, the Amazon Elastic Compute Cloud (EC2) is just an alternative to dedicated hosting - but rather than paying for 24h/day, you pay just for the time you have your instance up.  You still need to maintain the operating system, application, configure firewalls, implement a back-up strategy - the tasks which are included in a Software-As-A-Service (SAAS) model.  Main benefit is that you can select from quite a few starting points, Amazon Machine Images (AMIs) - from several free Linux O/S as well as Windows or commercial vendors (Oracle, RedHat, IBM, JBoss, Compiere, ...) combining convenience and license payment.

We will be using it for our automated build with one instance and one instance per configuration for regression and load testing and are "burning" currently max 2 hours per day for the 3 configurations (4 planned).  As these tasks are not essential, delays and non-availability is not an issue.  As these are test instances, we just load the image, run it and remove it - no need to save as we are starting the tests always from a clean install. 
As you know, persistency is "a bit" of an issue as you need to create a new AMI for any change to your configuration.  If the instance crashes, you are back to the last saved image - only S3 storage is persistent.  For 24/7 production use, you must solve external database backup and recovery.  You might want also to save certain log files externally.