Whitepaper - 2019
Microsoft Azure provides Availability Zones (AZ) to build and run resilient applications with High Availability (HA).
AZ protect your applications and data from datacenter failures through replication across Zones within a specific Region, located in a given country. Availability Zones are unique physical locations within an Azure region. Each zone is made up of one or more data centers and equipped with independent power, networking and cooling. This paper benchmarks AZ in order to assert performance impact, benefits and potential drawbacks.
The comprehensive tests at scale (up to 320 cores, taking advantage of Virtual Machine Scale Sets) reveal that activating AZ has negligible impact on performance, while providing resilience through zone balancing.
Secondly, we observed the benefits of a shorter average VM boot time with AZ which led to shorter completion times. Taking advantage of ActiveEon Elastic Dynamic deployment policy, the experiments demonstrate that such elastic policy can even reduce the wasted VM time using AZ, together with reducing the overall Azure cost. Moreover, it can also reduce completion time. Finally, taking advantage of AZ and ActiveEon Scheduler resilience, HA benchmarks with recovery are featured, simulating a zone failure.
All benchmarks have been conducted on a real-world application, a typical financial use case, featuring both Compute & IO Intensive tasks, and infrastructure currently in production on Azure.