Rose Business Solutions Dynamics GP

Recently we have discussed best practices about upgrades in our company. One of the big challenges in an upgrade is installing a large number of Dynamics workstations at the appropriate time with minimal user interruption as possible. Installing Dynamics has always been a pain. MBS has approached fixing this issue with their mass deployment tools but they never work and it usually ends up easier to do the installation manually. Depending on the installation you could spend 15 mins – 40 mins per workstation. Launching the installs simultaniously (3 at a time in the same area for example) may cut time a bit but still a huge pain.

This led to a series of responses by our technical team. Here is a summary from one of our team member Jason Young. Summary of response below:

One of the tricks I have used many times to minimize downtime for large rollouts is to leverage imaging tools like Acronis True Image. This tool lets you take a picture of a server or servers and redeploy them in a virtualized environment or dissimilar physical hardware. In essence, you can take the customers environment back to the office for the test upgrade without risk of downtime or impacting day to day production of the customer.

Imaging gives you some big advantages:

1. You don’t change the state of the customers environment until testing is successful and you are ready for the production rollout. This gives you huge kudos in the eyes of local IT staff.
2. You always have a clean rollback image.
3. You don’t have to bother local IT staff because you are working with an image offsite.
4. You can couple this type of test upgrade with a Terminal Server (which we have and is easy to build) and now you can do your user acceptance testing via web. Again, no customer impact.

To do the production rollout it is always optimal to do a parallel installation, but the obvious drawback is the cost of hardware. For customer without additional server resources, there is another trick that can minimize downtime and customer impact. For customers with a single backend server (single SQL server), you install a name instance of SQL and restore production data to the new named instance.

Here are the advantages;

1. A named instance of SQL means you have to use a different DSN name than the default instance. This means you can pre-install workstations with the new version and point them at the new instance without worry of an accidental upgrade.
2. The current state is not affected so you have a rollback in the event of an upgrade failure. You don’t have to do any type of restore because you never changed the data in the default instance of SQL.
3. You are basically taking a single server and doing a parallel application upgrade. So, it’s an in-place upgrade in terms of hardware but parallel in terms of the application and data.

End of summary.

Discussion: I like the seperate instance of SQL approach. I do this with different versions of Dynamics. I have 9.0 on one instance and 10.0 on another. You’d have to be careful that users do not launch and continue working in the older version instead of the newer version but that can be overcome with training or simply stopping the older instance of SQL when the time is right (probably need to set the older instance to manually start in case of a server reboot).

Anything you do to make this task easier?

One Response to “How do you install a large number of workstation after an upgrade efficiently?”