Share your experience using Oracle Compute Cloud Service

In my current organization, we deployed all our applications using AWS Fargate.

What I find best about AWS Fargate is that compared to deploying containers on EC2, where we need to check everything manually such as uptime, error logs, and other issues, AWS Fargate manages all these aspects automatically. If a container goes down, it automatically restarts it, and according to our requirements, it handles scaling up and down of all containers. This feature is really amazing.

They need to improve some UI-based interaction.

I have been using AWS Fargate for nearly six to seven months.

I do not face any issues while using AWS Fargate.

I do not face any issues while using AWS Fargate.

I have not connected with the customer support team from Amazon. We manage all errors from our side or find help to resolve the issues. For pro support, AWS charges additional fees, and in our plan, we did not take the AWS support plan, so we manage all errors independently.

Negative

The initial setup for AWS Fargate is busy.

Using AWS Fargate is becoming easier as the platform improves. On a scale of 1-10, I rate AWS Fargate a 7.

I am using multiple AWS services, including EC2 for cloud compute, S3 for storage, and VPC, Route 53, and load balancers for networking. We are using databases and Lambda functions as well. There are many components in EC2, including three types of load balancers, and IAM, which is the key aspect of AWS.

Previously, I have used EKS for microservices based applications deployed on our EKS clusters on AWS, and we managed the cluster ourselves according to our requirements, upgrading, deploying, and changing configurations as needed.

We are using EC2 and application load balancers as well as network load balancers. We have Amazon EC2 Auto Scaling group in place with two types of auto-scaling: time-based and CPU-based. For example, when the request exceeds 200 users, a new server automatically spins up in the Auto Scaling group to manage the load. In non-peak hours, we maintain a minimum of 20 servers, and our developer team continuously resolves bug fixes, deploying new releases to our web servers using Harness as our CI/CD tool.

Currently, I'm working on the FinOps side for AWS, managing about 15 to 16 AWS accounts and reviewing costs weekly, identifying any non-utilized resources such as terminated servers with persistent EBS volumes or unreleased IPs to save costs.

We have direct AWS accounts for our environment and multiple dedicated accounts for QA, production, and staging workloads.

Amazon EC2 Auto Scaling has significantly enhanced our infrastructure’s reliability, scalability, and cost efficiency. By automatically adjusting the number of EC2 instances based on traffic and server health, we’ve achieved:

• High Availability: Unhealthy instances are automatically replaced, reducing downtime and ensuring continuous service delivery.

• Optimized Performance: During traffic spikes, Auto Scaling adds instances seamlessly, maintaining application responsiveness.

• Cost Efficiency: By scaling in during low-traffic periods, we avoid paying for idle resources, aligning with FinOps best practices.

• Operational Simplification: Reduced manual intervention for server provisioning and health monitoring, allowing the team to focus on higher-value tasks.

• Integrated Monitoring: Coupled with tools like CloudWatch, Instana, and Grafana, we gain real-time insights and proactive alerts for faster incident response.

The best feature I appreciate about Amazon EC2 Auto Scaling is its health check functionality; when a server becomes unreachable or enters an unhealthy state, it automatically triggers an alert, and the load balancer responds by spinning up a new server, ensuring that traffic is distributed effectively.

For predictive scaling capabilities in Amazon EC2 Auto Scaling, I deploy all servers to the limit for a week to monitor CPU utilization, and I calculate the load trends to manage the minimum and maximum through time-based auto scaling.

We are using CloudWatch, and have integrated it with various application performance monitoring tools such as Instana, AppDynamics, and Grafana. We set thresholds in these tools to alert us if CPU utilization exceeds 70%, allowing our dedicated team to rapidly respond to any issues.

The main benefit of implementing Amazon EC2 Auto Scaling is that service availability for end users reaches 99.99%, with effectively zero downtime. Even if multiple servers become unreachable, Amazon EC2 Auto Scaling automatically spins up new servers to replace them, all without human intervention.

We utilize two types of scaling policies: time-based and CPU-based. When our CPU utilization reaches 70%, Amazon EC2 Auto Scaling automatically adds a new server to handle the generated load.

Current Auto Scaling is mostly reactive. Introducing predictive or AI-driven scaling could anticipate traffic spikes and improve responsiveness.

I have been working with Amazon EC2 Auto Scaling for almost five to six years.

Two or three years ago, we did escalate concerns to AWS support when our servers were not spinning up despite setting appropriate min and max thresholds.

I would rate the technical support of AWS a nine, as their team resolves issues effectively and meets our expectations.

Positive

More than a year ago, I worked on EKS, and in my current new project, there is no EKS.

During the initial setup of the environment with Amazon EC2 Auto Scaling, we faced challenges with port issues leading to failed health checks, which were ultimately traced back to a bug in our JAR file that was fixed by our developer team.

I cannot pinpoint specific features for future inclusion, as my current project's requirements are well met by the AWS services we are using.

I do not have experience with other cloud services. I have only worked with AWS, where I am also certified. I cannot compare it directly because I have not used the other platforms.

I do not have specific suggestions, but if some features were similar to Lambda's serverless deployment method, allowing us to simply upload our deployable files directly to the load balancer, it would streamline processes.

On a scale of one to ten, I rate Amazon EC2 Auto Scaling a nine.