Liftr Look Ahead: Zones and Regions - Their Importance to Cloud Customers
Why do regions and zones matter to cloud customers?
There are a few answers to that question:
Fast end-customer response times is one of the primary reasons. It’s called “latency” in the engineering community. Low latency is important for many classes of applications, from fast search and website response times to fluid map and game interactions. Enabling low latencies often requires running applications in data centers that are geographically close to end customers.
Another primary reason is regional data privacy laws. Many countries specify that local consumer and company data must be kept in-region. So, applications serving those countries must use in-region storage, which means using in-region compute resources to keep data transfer costs to a minimum is also a good idea. Plus, that also helps reduce service latencies, so it has a compounding effect for latency sensitive applications.
A third primary reason is service continuity or redundancy. Entire data centers can experience power interruptions and sometimes a large-scale catastrophe like a typhoon can take down many data centers in a zone. Planning an application architecture to span zones or even to span regions enables the application to be fault-tolerant. That way the application can continue to operate even if some of the zones serving the application can’t meet their service-level agreements or SLAs.
Operational Costs Determine Cloud Components Pricing
However, the opposite force for localizing compute resources is cost. Electrical power costs and regional business tax environments are a couple of the major operational cost components that determine cloud instance hourly prices. The cost of electrical power and taxes can vary dramatically by region, creating substantial price differentials between identically configured instances in the same cloud but in different regions.
So, what are regions and zones?
Well, its complicated.
The base unit of cloud computing at a geographic scale is a data center. A zone can be one data center or it can be many data centers located geographically close to each other and connected by a single very high speed network.
A region contains one or more zones. Some regions have only one zone. Some cloud service providers choose to provide fault tolerant architectures within a single-zone region. Cloud customers therefore must carefully consider the SLA terms for single-zone regions.
Regions typically have less than five zones, three zones being very common, but some regions have more than five zones.
A region can be a whole country, depending on the size of the country, the business demand for cloud computing in that country and local data privacy laws, such as Germany and France. But regions can also contain multiple smaller countries, such as South America, and a few larger countries are subdivided into many regions, such as the US and China.
Typically, as a region matures economically or grows rapidly it may start out with one zone and/or region and then add zones to its region, and then split into separate regions, each region containing multiple zones.
Our Liftr Cloud Components Tracker records detailed cloud service provider regional changes every month. For regions and instance types where zone information is available, we also record that. We use region and zone metadata to compare the size of cloud service provider IaaS deployments worldwide.
Liftr Cloud Components Tracker: http://bit.ly/2QceXlT
Liftr Cloud Regions Map: http://bit.ly/2LGB5PV
Follow Us On: