Some time towards the end of Q4 2015, I was tasked with re-designing an Azure environment which had initially be commissioned to host a custom web solution by an outsourced team. The original design lacked any real resilience or high availability, with the solution itself being, for the most part, a collection of single instance servers. For example, SQL offered no HA via clustering or AAGs; no load-balancing had been implemented on the frontends, etc.
After a number of website outages due to server failures and/or high traffic volumes resulting in application crawl, the organisation approached my team with a request of redesigning the infrastructure.
My design was simple; I would continue to utilise Azure’s Infrastructure as a Service (IaaS) for the medium-term, deploy multiple, load-balanced IIS front-ends (which would eventually host the Sitecore Content Delivery nodes), and a simple, two-node SQL Always-On Availability Group (AAG) back-end with witness server. Continue reading