The AWS Cloud is an ever-growing network of Regions and points of presence (PoP), with a global network infrastructure that connects them together. With such a vast selection of Regions, costs, and services available, it can be challenging for startups to select the optimal Region for a workload. This decision must be made carefully, as it has a major impact on compliance, cost, performance, and services available for your workloads.

Evaluating Regions for deployment

There are four main factors that play into evaluating each AWS Region for a workload deployment:

  1. Compliance. If your workload contains data that is bound by local regulations, then selecting the Region that complies with the regulation overrides other evaluation factors. This applies to workloads that are bound by data residency laws where choosing an AWS Region located in that country is mandatory.
  2. Latency. A major factor to consider for user experience is latency. Reduced network latency can make substantial impact on enhancing the user experience. Choosing an AWS Region with close proximity to your user base location can achieve lower network latency. It can also increase communication quality, given that network packets have fewer exchange points to travel through.
  3. Cost. AWS services are priced differently from one Region to another. Some Regions have lower cost than others, which can result in a cost reduction for the same deployment.
  4. Services and features. Newer services and features are deployed to Regions gradually. Although all AWS Regions have the same service level agreement (SLA), some larger Regions are usually first to offer newer services, features, and software releases. Smaller Regions may not get these services or features in time for you to use them to support your workload.

Evaluating all these factors can make coming to a decision complicated. This is where your priorities as a business should influence the decision.

Assess potential Regions for the right option

Evaluate by shortlisting potential Regions.

  • Check if these Regions are compliant and have the services and features you need to run your workload using the AWS Regional Services website.
  • Check feature availability of each service and versions available, if your workload has specific requirements.
  • Calculate the cost of the workload on each Region using the AWS Pricing Calculator.
  • Test the network latency between your user base location and each AWS Region.

At this point, you should have a list of AWS Regions with varying cost and network latency that looks something Table 1:

RegionComplianceLatencyCostServices / Features
Region A

15 ms$$
Region B

20 ms

$$$

X

Region C

80 ms$

Table 1. Region evaluation matrix

Many workloads such as high performance computing (HPC), analytics, and machine learning (ML), are not directly linked to a customer-facing application. These would not be sensitive to network latency, so you may want to select the Region with the lowest cost.

Alternatively, you may have a backend service for a game or mobile application in which network latency has a direct impact on user experience. Measure the difference in network latency between each Region, and determine if it is worth the increased cost. You can leverage the Amazon CloudFront edge network, which helps reduce latency and increases communication quality. This is because it uses a fully managed AWS network infrastructure, which connects your application to the edge location nearest to your users.

Multi-Region deployment

You can also split the workload across multiple Regions. The same workload may have some components that are sensitive to network latency and some that are not. You may determine you can benefit from both lower network latency and reduced cost at the same time. Here’s an example:

Figure 1. Multi-Region deployment optimized for feature availability

Figure 1. Multi-Region deployment optimized for feature availability

Figure 1 shows a serverless application deployed at the Bahrain Region (me-south-1) which has a close proximity to the customer base in Riyadh, Saudi Arabia. Application users enjoy a lower latency network connecting to the AWS Cloud. Analytics workloads are deployed in the Ireland Region (eu-west-1), which has a lower cost for Amazon Redshift and other features.

Note that data transfer between Regions is not free and, in this example, costs $0.115 per GB. However, even with this additional cost factored in, running the analytical workload in Ireland (eu-west-1) is still more cost-effective. You can also benefit from additional capabilities and features that may have not yet been released in the Bahrain (me-south-1) Region.

This multi-Region setup could also be beneficial for applications with a global user base. The application can be deployed in multiple secondary AWS Regions closer to the user base locations. It uses a primary AWS Region with a lower cost for consolidated services and latency-insensitive workloads.

Figure 2. Multi-Region deployment optimized for network latency

Figure 2. Multi-Region deployment optimized for network latency

Figure 2 allows for an application to span multiple Regions to serve read requests with the lowest network latency possible. Each client will be routed to the nearest AWS Region. For read requests, an Amazon Route 53 latency routing policy will be used. For write requests, an endpoint routed to the primary Region will be used. This primary endpoint can also have periodic health checks to failover to a secondary Region for disaster recovery (DR).

Other factors may also apply for certain applications such as ones that require Amazon EC2 Spot Instances. Regions differ in size, with some having three, and others up to six Availability Zones (AZ). This results in varying Spot Instance capacity available for Amazon EC2. Choosing larger Regions offers larger Spot capacity. A multi-Region deployment offers the most Spot capacity.

Conclusion

Selecting the optimal AWS Region is an important first step when deploying new workloads. There are many other scenarios in which splitting the workload across multiple AWS Regions can result in a better user experience and cost reduction. The four factors mentioned in this blog post can be evaluated together to find the most appropriate Region to deploy your workloads.

If the workload is bound by any regulations, shortlist the Regions that are compliant. Measure the network latency between each Region and the location of the user base. Estimate the workload cost for each Region. Check that the shortlisted Regions have the services and features your workload requires. And finally, determine if your workload can benefit from running in multiple Regions.

Dive deeper into the AWS Global Infrastructure Website for more information.

Categories: Architecture