Summary

There are perhaps thousands of considerations that must be dealt with when building a viable production implementation, let alone one that needs to be highly available. You would be well advised to make the extra effort of first properly determining which highly available solution matches your application's needs, then to switch focus to what is the most effective way to implement that chosen high availability solution. There are critical design decisions that need to be addressed for each high availability solution. If, for example, you choose data replication to support your HA needs, you must determine the right type of replication model to use (like a central publisher and remote distributor), what the schema limitations might be, the fail-over process that needs to be devised, and so on. Understanding the characteristics of your highly available data will also drive the RAID level configurations that need to be used, number of controllers, controller types, network topologies, and the proper database backup/recovery techniques to use.

Disaster recovery planning and execution must also be made part of your high availability implementation from start to finish. Considering this upfront before you have selected your high availability solution is critical to your success.

This chapter also described some other high availability perspective on processing capabilities such as Microsoft Analysis Services (MSAS/OLAP). There is a huge proliferation of new MSAS/OLAP applications that have high availability requirements.

In addition, understanding that there other vendors (other than Microsoft) with very sound high availability alternatives should allow you to mix and match your capabilities depending on any existing (or intended) vendor relationship.

You must get the design basics down first for all of these high availability solutions. Then fine-tune your designs to achieve the optimal availability results.