Searching for the Holy Grail

Ever since I started deploying enterprise system management (ESM) solutions the Holy Grail has been finding a methodology for Rapid Deployment. Twenty to twenty-five years ago a typical project would not be counted in days or hours but months and years. I spent a pleasant 9 months in Frankfurt deploying Tivoli and when I left the project continued without me for many months. When I came back from Germany I spend another 9 months writing a rapid deployment methodology based on the work we had done in Frankfurt for IBM called RapTER. It was a decent tool (notable for its fantastic logo) but the UK IBM team found out that there was a similar project underway in Holland and impetus was lost.

Recently we have had to start the search again. Orb Data has been working on a cloud version of Netcool Operations Insight (NOI) called OpsAlerts and it was inconceivable that for each customer we would ask them to wait weeks while we deploy their cloud solution. Therefore we looked at a number of technologies to bring the deployment time down and we have now succeeded in deploying a complete NOI solution from start to finish in 3 hours with 1 button push,

The basics behind this are shown in the diagram and described below:

  1. IBM Cloud (formerly Softlayer) allows the assignment of a script to be run during the build of a server. This script is pulled from an Orb Data HTTP server and bootstraps the process so that after the Operating System is deployed, the initialisation script is run.
  2. All of the software and configuration is mounted on an NFS share. This is linked to the software repository software tool GIT so it can be automatically updated. The script connects to the NFS share and runs all of the configuration steps without intervention.
  3. A configuration matching the company/servername needs to be defined (in the git repository).  The configuration allows all variables during the build process to be overridden and also provides a mechanism for defining the software to create (object server definitions, db2 databases etc).

I kept things reasonably simple, but having now made all of these scripts, extracted the variables etc, I think it would be reasonably simple to put these into a more fully-fledged system like puppet or chef

We don’t plan to stop there. We have now created even quicker deployment mechanisms for the UI web app and OpenLdap using docker, which speed up the process considerably.  As the automated build process is already defined, we are now working at migrating all the scripts into docker build files.  We anticipate that when this is complete the deployment time would come down to a few minutes.

If you would like to try OpsAlerts then please send an email to or visit

Visits: 9