Host scalable and globally distrubuted compute systems to get best value for money in cloud
May 24, 2019
Host scalable and globally distrubuted compute systems to get best value for money in cloud
Score 8 out of 10
Vetted Review
Verified User
Overall Satisfaction with Google Compute Engine
We use Compute Engine for one of our main consumer and business products. We have a globally distributed infrastructure for this. I believe we are using almost all the features/functionalities offered by GCE. e.g Instance, groups/templates with full rolling updates capability coupled with CI/CD. We also host part of our monitoring systems in GCE. Other than that, some other teams are on the POC phase for some critical backend products as well. Moreover, we experiment with some of our machine learning on Compute Engine. The benefit we get by using GCE is the combination of the autoscaling, rolling updates, preemptability, and global network.
- Advanced autoscaling logic to cater scenarios that involve high load at the global level.
- Seamless and reliable rolling updates with support for releases.
- Backup data via very fast snapshots helps to quickly back up systems
- Good support for things like metadata (pre-defined and custom).
- Ability use Windows client OS VMs (or support import capability)
- Increase the offered default monitoring metrics set.
- Adjustable shutdown cooldown period (instead of fixed 30 seconds window).
- Positive: We had good cost savings
- Positive: Very good stability and reliability compared to hosting in on-prem.
- Negative: Had to re-architect some part of the application to fit the limitations.
- Amazon Elastic Compute Cloud (EC2) and VMware ESXi
Neither of them has managed a user-friendly approach to setting up the global distribution of computing systems. Modern day business must be able to cater to the entire world instead of being regional to be competitive. Only Google offers this capability, hence why we have no second thoughts on using GCP for such product hosting.