Client Overview:
The client provides next generation business solutions that streamline the corporate payment process. Headquartered in Atlanta, GA, our client serves numerous Fortune 1000 companies, including 7 of the Fortune 100 companies.
Business Problem:
Following the client acquisition by American Express, they were asked to move over 200 servers from a downtown Atlanta location to a more modern data center at the clients headquarter location. The servers were a broad mix of aging IBM, Dell, and Compaq/HP hardware. Some systems were more than seven years old and ran Windows NT 4.0, but still performed mission critical tasks.
Strict system uptime requirements, along with the aging critical hardware dictated several layers of recovery methods in the event of a failure during the move. Although the IT department had spent several months on planning the move, there was no clear answer on how to protect the company assets to prevent catastrophic system and/or data loss during the physical move process. Traditional backup/restore methods were not viable due to the downtime window being far less than the time it would take to build and restore data to the systems. There were just a few weeks left before the move date and there were only a total of 13 overnight hours available each weekend in order to capture systems images for backup.
The Facts:
- There were a large number of antiquated hosts running critical applications on Windows NT 4.0 platforms
- Systems had to be captured during relatively short maintenance windows and there was not a lot of room for error based on the published move date.
- An entire swing environment was rented for a VMware farm and stand-alone physical servers that were to be used to deploy images of the production environment to.
- Some systems required that the backup instance be updated to the point of the start of the maintenance window in order to ensure data consistency and transaction accuracy.
- We were able to choose the tools for this project – PlateSpin PowerConvert and VMware Server.
The Solution:
We utilized standardized IBM xSeries swing gear that was provided on a rental basis by a local partner. Additionally, we employed PlateSpin PowerConvert and VMware Server to create an agile recovery environment within the customer’s existing network.
As the first level of protection, we virtualized each critical system and housed them all in a VMware Server environment that was built on-site.
PlateSpin PowerConvert was used to reprovision the virtual server instances on to the bare metal IBM physical swing gear to provide a physical replacement option for the more resource intensive and demanding systems. PowerConvert was also used to maintain data consistency between the running production servers and the standby virtual and physical systems after initial capture.
Due to an extremely limited timetable, a strategy was developed to leverage the remaining after-hours maintenance windows and extended daytime hours to perform all virtualization and restoration activities. The client provided space in the new data center and verified the functionality of the standby instances of the production environment in an isolated environment.
The Results:
When the move window opened at midnight, the critical systems were in a standby state as virtual machines and stand-alone physical systems.
In accordance with the customers desired options for recovery, we had created three distinct options for recovery in the event of a failure – hardware repair on the original system, a physical standby server, and a virtual instance that could be immediately turned on and put into production.
The client expressed gratitude for the level of insurance that was created for their systems. Unplanned downtime after the window would cost them millions of dollars per minute, and we were prepared to ensure and guarantee that such outages would be remedied immediately and with confidence.