We are looking to use Vahana for one of our clients. Could someone help me in finding the minimum system requirements to setup a VRT in development mode and what it would to take for running production setup under HA for 50 TPS?
@naks @Mayank @Vikas_Dhillon @tushar.jain @Dilip_Vishwakarma
Hi Vinay,
What would be the business use case scenarios for the VRT.?
Hi @Dilip_Vishwakarma ,
Welcome to Vahana community. It seems you’re liking Vahana community too much. Thanks for your support. The use case could be anything. Does it matter as already mentioned I am expecting 50 T PS traffic on this installation
Hi Vinay,
Below Infra sizing can withstand infra upto 2 peak factor and can handle 300-400 Concurrent users. Following assumption are taken in considerations:
|Assumptions:
Concurrent usres: 300-400 users (Autoscaled)
Payload size < 100 KB.
Decimal internal API response time < 1 second
Bank Core system API’s response time < 3 second
Average Peak Size Factor: 2 times
Logs retention period: Production: 7 days | Lower Environments: 3 days
| Components | AWS Services | Count/Size | Instance Type | Min | Desired | Max | Pricing Type | Remarks |
|---|---|---|---|---|---|---|---|---|
| Domain | Route 53 (DNS) | 2 | ||||||
| SSL Certificates | 2 | Based on Bank’s requirement (1/URL) | ||||||
| Network Components | NAT Gateway | 1 | ||||||
| Application LoadBalancer | 2 | |||||||
| Security | IAM Users with required permission | |||||||
| KMS Encryption | ||||||||
| AWS Gaurd Duty | ||||||||
| WAF (Web Application Firewall) | ||||||||
| EKS Cluster | EKS Cluster (VRT) | |||||||
| VRT Application Worker Node Size | r6a.large | 2 | 2 | 6 | Spot | Autoscaled enabled | ||
| Custom Application/SDG Component Worker Node Size | r6a.large | 2 | 3 | 6 | Spot | Autoscaled enabled | ||
| ELK Node Size | r6a.large | 2 | 2 | 6 | On-demand | Autoscaled enabled | ||
| Nginx Ingress Controller Worker Node Size | t3a.medium | 1 | 2 | 4 | On-demand | Autoscaled enabled | ||
| Service Discovery Consul/Eureka Node Size | t3a.medium | 1 | 2 | 4 | On-demand | Autoscaled enabled | ||
| ElastiCache for Redis configuration Database | ||||||||
| Storage | EFS Storage | 250 GB | ||||||
| S3 Bucket for Storage | ||||||||
| Database | RDS Database for Postgres | 50 GB | db.r6g.large (2vCPU | 16GB RAM) | ||||
| Monitoring | Cloudwatch | With 3 Metrices (CPU, Memory & Disk Space) | ||||||
| Container Insights | Subject to Usage | |||||||
| Server | EC2 Jumpbox | 100 GB GP3 | t3a.micro | |||||
| Backup | AWS Backup | Subject to Usage |
Hope it answers your query!
Hi Daksh,
You have mentioned about the concurrent number of users and other details.
Can you help me to know how did you realised these numbers:
|Assumptions:
Concurrent usres: 300-400 users (Autoscaled)
Payload size < 100 KB.
Decimal internal API response time < 1 second
Bank Core system API’s response time < 3 second
Average Peak Size Factor: 2 times
Logs retention period: Production: 7 days | Lower Environments: 3 days
THe above confuguration is applied any where?
Hi Dilip,
Yes. These configuration can be applied anywhere. As we use autoscale strategy as per infrastructure perspective. Only VRT parameters should be optimized.