CHOOSING AN AWS REGION
The AWS EC2 is hosted in multiple locations worldwide. The locations comprise regions, availability zones, local zones, AWS outposts and wavelength zones. Each region is a separate geographic area and AWS has 26 different geographic regions around the world, with plans announced for a further eight in Australia, New Zealand, Canada, India, Israel, Spain, Switzerland and UAE.
An AWS account provides users with the ability to launch Amazon EC2 instances in locations that meet their requirements, but the specifics of that account will determine which regions are accessible. Global or multinational organisations may be physically located in one region, but choose to launch instances in Europe, for example, to be closer to their customers based on that continent or to comply with certain legal requirements.
“The recommendation is to place the workload where it is closest to the majority of its users,” says Greg Cockburn, Head of Hyperscale Cloud at AC3. “Most of our clients stick to the ap-southeast-2, which is Asia Pacific (Sydney).
“Each region has an entry point in the closest location. So, if users are located in Sydney and have access to the Sydney end point, they will have a better response time.”
The seven other Asia Pacific locations are ap-east-1 (Hong Kong), ap-southeast-3 (Jakarta), ap-south-1 (Mumbai), ap-northeast-3 (Osaka), ap-northeast-2 (Seoul), ap-southeast-1 (Singapore) and ap-northeast-1 (Tokyo).
The potential downside of selecting just one of the locations to host your infrastructure relates to availability. “If the whole region goes down, then nothing works and all of the organisation’s workloads also go down,” says Cockburn.
MULTI-REGION SOLUTION
It is the potential for such an incident that is the main reason an organisation may choose to select more than one region, primarily for disaster recovery. If something happens on the AWS data centre network program or storage program and the host region becomes unresponsive, users will be unable to access the servers.
An alternative infrastructure that creates the same workflows, load balancers and database in a different region means there is a script available to switch over to the next region if necessary. Simply put, it’s a safety device and the great advantages of multi-region solutions are reliability and high availability.
ON THE OTHER HAND...
The trade-off is cost. If you double your infrastructure, you will double the cost. However, just as a car with four wheels will only have one spare, a potential way forward is to limit the disaster recovery footprint.
“One of our clients has five servers in its main region, but only one server in its disaster recovery region,” says Cockburn. “If a switchover happens, they can spin up more servers in the region to solve the workload issue.”
In a normal, everyday situation, the customer will save costs by running a minimally sized instance, and scaling up if and when it becomes necessary.
“You do have to keep everything in sync all the time, however, because you never know when you may need to switch over,” warns Cockburn. “This means continual backups, database synching or application version syncing. You need to ensure you are running the same version of an application in each different region.”