Creating the Production Environment

The term "migration" implies that we are moving from one environment to a different environment. The environment to which you are moving is referred to as the target environment. As illustrated by the E-stack in Chapter 3, the application is only one component of the target environment.

In this section, we explain how you prepare the facilities to support the new compute, network, and storage platforms. In addition, we explain how to install and configure these platforms and how to create the application infrastructure that will support the application.

All of these activities are distinct and separate from the migration of the application. They all require unique, disparate skill sets, and all have differing timelines . We strongly recommend that these activities proceed in a parallel fashion with the migration of the application, to reduce the amount of time required for the migration effort.

Building the Production Facilities Environment

Modifying a building or the data center is a significant task. Depending on the changes required, this can take a great deal of time. Care must be taken to minimize the disruption to the existing IT operation when bringing in new power, knocking down a wall, or increasing air conditioning efficiencies.

Inspections and building permits might have to be arranged for and acquired in order to make structural modifications to real estate. In addition, new facilities might have to be created or leased to support the new environment. Heating and ventilation additions might not be feasible in some older environments, and building design might limit the amount of heat that can be dissipated. You should also consider any limitations on the generating capacity of the local power utility.

Facilities modification should be executed according to the plan that was created as a result of the facilities assessment that was described earlier in this book.

Building the Production Platform

Once the facilities are in place, you can deploy the compute platform. This platform might be integrated with existing storage and networking components , or it might require its own storage and networking subsystems.

If existing storage is to be used, ensure that it integrates with the new environment at both the hardware level (fiber channel, SCSI, and the like) and at the software level (file system type, volume management, and so on.) When using existing storage, be extremely careful to ensure that no data on the existing storage is lost or corrupted.

We recommend that experts in the storage technology and the compute platform be hired to configure storage because these systems are typically complex. In the case of NAS and SAN implementations , firmware patches and OS upgrades are frequently required.

If the platform is to use a new storage platform and networking gear, the impact on the existing environment can be minimized, although specialized skills will still be required.

Once the storage platform has been configured, ensure that extensive testing is conducted to verify that the system is performing correctly. These tests should be conducted with utilities supplied by the compute platform or storage vendor, not the application. We provide more information about the testing process later, but it is critical that the newly installed storage be validated before you install the data.

Network infrastructure is usually easier to test and configure than are storage platforms. Ensure that all switches, routers, hubs, and load balancers have been correctly installed and configured. Ensure that you generate sufficient loads for these components to verify that they are performing as intended. When appropriate, use external consultants to generate representative loads for systems before placing them into production.

If a larger, enterprise-class machine is used as the compute platform, the compute platform itself might require extensive configuration. Frequently, large enterprise-class machines can be chosen for the new production platform because of their availability, scalability, dynamic reconfiguration, and hot swap capabilities. Optimal configuration and testing of these complex platforms takes planning and time. In addition, training might be required to understand, design, and administer this state-of-the-art technology. In certain cases, you can reduce TCO and improve ROI by partitioning a large symmetric multiprocessor machine into smaller virtual machines. This can simplify administration, reduce floor space, and improve availability, but might require the deployment of resource management software to ensure that service level agreements can be maintained across all virtual machines. Configuring and sizing this infrastructure will require some experimentation and testing.

These activities can and should take place while the application is being migrated . The production platform build is independent of the migration activity and requires a completely different skill set.

Building the Application Infrastructure

Once the platform is installed and the OS is up and running, you can start configuring the infrastructure for the application. All the software used to support the enterprise and the application should be available in the new environment, including the following items:

• Directory name services

• Identity services

• Accounting packages

• Management infrastructure

• Backup facilities

• Third-party runtime products required by the application

• Databases

As we discuss in the following sections, these products should be tested as standalone utilities whenever possible before the installation of the migrated application.