For me, the best feature of ECS is Fargate because I don't have to manage anything. Instead, everything is managed by AWS and all I have to do, in essence, is configure my containers and deploy them.
Learn what your peers think about Amazon Elastic Container Service. Get advice and tips from experienced pros sharing their opinions. Updated: April 2024.
For Amazon EC2 Container Service, providing the ability for users to select specific processor, memory, disk, and interface types might be an ideal feature. But, the practicality of offering all possible physical combinations is nearly impossible due to the underlying physical machines. AWS and Azure organize options into groups based on essential components like powerful processors or critical interfaces, considering physical restrictions. While expanding these choices is conceivable, it may not be feasible from a financial and practical perspective. Customers generally comprehend this limitation, as even in their own data centers, exact physical machine requirements are often a result of a combination of factors such as price, availability, and new machine generations.
DevOps Engineer at a financial services firm with 10,001+ employees
Aug 23, 2020
It has an Auto Scaling group feature. We can use this feature to have an Auto Scaling group to specify a minimum and maximum count for all types of configurations. Based on the specified values, Amazon Elastic Container Service scales the required CPU environmental metrics.
Visualization is an important factor for me, and I don't think that the visuals within ECS are good enough because it doesn't show you all the details you might need to see at a glance.
Learn what your peers think about Amazon Elastic Container Service. Get advice and tips from experienced pros sharing their opinions. Updated: April 2024.
For Amazon EC2 Container Service, providing the ability for users to select specific processor, memory, disk, and interface types might be an ideal feature. But, the practicality of offering all possible physical combinations is nearly impossible due to the underlying physical machines. AWS and Azure organize options into groups based on essential components like powerful processors or critical interfaces, considering physical restrictions. While expanding these choices is conceivable, it may not be feasible from a financial and practical perspective. Customers generally comprehend this limitation, as even in their own data centers, exact physical machine requirements are often a result of a combination of factors such as price, availability, and new machine generations.
DevOps Engineer at a financial services firm with 10,001+ employees
Aug 23, 2020
Sometimes, the instances fail under the ECS container cluster, and we have to manually go and find out the black sheep in the ECS container instance.
We had an issue earlier where one of the instances under the ECS container cluster went down, and we were not able to identify that instance. The instance got terminated, but a new instance did not come up. Therefore, I had to manually get that instance up. It could be optimized better. In production, we normally cannot sustain such things. It can be optimized in terms of instances, durability, and serving the requests of customers.