Why we chose Amazon Web Services as our cloud infrastructure platform
December 11, 2018
At vintrace we are committed to continuous improvement of our platform and ensuring customers get the best experience whilst using our software. We knew there were improvements to be made in our cloud infrastructure, so in the third quarter of 2018 we invested the time and money to evaluate new platforms, plan our approach and start migrating our cloud customers.
We evaluated three major Cloud Providers including Microsoft Azure, GCP (Google Cloud Platform) and AWS (Amazon Web Services). We landed on AWS for it’s platform maturity, backup/recovery features, automation features, cost/value, global coverage, and best overall performance based on our platform requirements.
Performance was a major player for us in the evaluation process and EC2 (Elastic Compute Cloud) instances on AWS new Nitro platform outperformed the rest with over 6x improvement over our existing platform on intensive reporting tasks. The Nitro platform provided us with near bare-metal performance on highly optimised Intel processors whilst gaining all the benefits of EC2 features.
What follows are some tests we carried out and the results that contributed to our decision making process.
Our application benefits greatly from single threaded performance and the Geekbench benchmark which covers various compute categories showed that latest generation AWS instances topped GCP and Azure in the single-core performance.
Big differences between Azure alongside AWS and GCP instances when it came to testing database workloads on our preferred database configuration. AWS performed best in this space with GCP closely behind. We saw Azure severely IO bottlenecked due to much bigger capacity allocation required to gain the same level of performance.
The vintrace Bulk Stock Report is a popular reporting tool amongst our customers and so we included it as a benchmark in our tests. This particular customer we tested had over 10 years of data. On the old infrastructure the report would take 30+ seconds to run. We saw AWS instances beat out the competition by at least 4 seconds on average and taking a maximum of 6 seconds to generate. This was a huge improvement over the old infrastructure.
At the start of November 2018 we completed migration of our Asia Pacific customers to the new platform. We received positive feedback as migration progressed with reports of specific operations performing 4 times faster and pre-existing performance related issues completely disappearing after migration. Customers in United States, Canada and Europe have also been migrated without issue. Our plan will see production migration finalised by the start of December 2018.
The infrastructure improvement journey does not end in 2018. In early 2019 we aim to make further feature improvements including ability for customers to spin up demo/sandbox versions of their instances on-demand. We’re also making improvements to security, high availability and disaster recovery, as well as more automation features translating to faster support response times and better experience.
If you’re looking for winery software with a strong customer focus and true commitment to platform improvement, stability please reach out to us. Contact our sales team for a no obligation demo.