With modern businesses built around their IT infrastructure, ensuring data is protected and always available should be a priority to any business that relies on any type of centralized IT infrastructure, regardless of it being a traditional legacy, virtualized, or public cloud based infrastructure. There are a multitude of vendors out there that can sell any number of products for servers, storage, and backup that promise high availability, but in my career I’ve found far too often that both the vendors and their clients don’t put enough thought into the total picture when its time to implement.
The first step in ensuring data protection and high availability is to define the metrics of success and failure. These will often change over time, and should be agreed and documented with the business. Use these metrics to define your internal SLA’s and those with external vendors and service providers.
Once you have your SLA’s defined, you can begin designing solutions to meet or exceed those metrics. Eliminate single points of failure wherever possible. Most mid to enterprise level equipment includes N+1 components, but also remember that a single cabinet, cable bundle, or power feed for the utility company can all be points of failure. Costs should be weighed against the benefit (or lack thereof) against downtime or data loss.
Don’t wait until you have put a new application or system into production before you implement a backup solution. Regardless of the backup solution(s) chosen, it should be setup and tested before the environment is released. This way, you’re sure that it not only works but is able to be completed within the SLA period. And remember to have your vendor or service provider include this testing as part of your Statement Of Work for any implementations you don’t do yourself.
Over the next several posts in this series I will dive deeper into the subject of data protection and high availability to share the successes, failures, and musings from my 17 years of experience in IT.