The challenge
Keep a production SaaS platform running while the stack changes
The company runs customer-facing applications on both Windows-based Amazon EC2 and container services. Its support team needed to identify failures early, distinguish application latency from infrastructure or database symptoms, and roll out changes across tenant groups with a clear rollback path.
Modernization had to fit the same operating model. Legacy deployment tooling, the existing AWS Managed Services (AMS) workflows and dependencies on database, networking and messaging services all had to be supported as changes moved into production.
How TechTiera runs and improves the AWS platform
Operate EC2 and ECS Fargate
Maintain Windows instances and standard machine images, apply patches, and manage Amazon ECR, task sizing, health checks and deployment circuit breakers. Monitor throughput, response times, errors and application health, and use five-minute synthetics to check login and critical pages.
Tune capacity against real traffic
Use Application Load Balancer request data to tune Amazon ECS target tracking. Investigate changes in Apdex, an application responsiveness measure, and allow for .NET startup time when adjusting task capacity.
Manage database and integration dependencies
Monitor Aurora PostgreSQL with Babelfish for CPU, connections, I/O and replica lag, and coordinate database changes with the DBA team. Support RDS Proxy, DynamoDB configuration, ElastiCache heartbeats, CloudFront, Route 53, VPC connectivity, Amazon S3, and Amazon SES and event-driven interfaces.
Control releases and recovery
Sequence rollouts across tenant groups and use blue/green releases and target-group rollback. Use service heartbeats, AWS Lambda and Systems Manager automation for recovery, with safeguards that pause automated restarts during deployments. Monitor AWS Backup, protected vaults, S3 versioning and restore readiness, including AWS Elastic Disaster Recovery where configured.
Manage access, security and cloud costs
Handle AMS incidents, service requests and controlled changes. Review time-limited privileged access, identity and certificate controls, GuardDuty and Security Hub findings, and CloudTrail events. Review budgets, anomalies and utilization, and recommend rightsizing, non-production schedules and Savings Plans where appropriate.
Modernize in parallel with operations
The program includes moving remaining web tiers to containers and consolidating alerts. The team maintains CloudFormation change sets and drift checks while moving infrastructure and pipelines toward Terraform and GitHub Actions. TechTiera is assessing Amazon ECS versus Amazon EKS, .NET modernization and options for the AMS operating model.
The SaaS operating model in practice
| Outcome | What it means in practice |
|---|---|
| Continuous incident coverage | An 11-engineer team monitors application and AWS health around the clock, escalates incidents, coordinates restoration and reviews significant incidents. |
| Repeatable release and rollback procedures | Teams can introduce changes in stages and reverse a release through the load balancer if production checks warrant it. |
| Modernization inside the support engagement | Platform engineering work progresses alongside daily operations. A file service backed by Amazon S3 is already in place, while infrastructure and pipeline transitions continue. |
Frequently asked questions
How does TechTiera monitor a SaaS application on AWS?
Engineers combine synthetic login and page checks every five minutes with application response time, throughput, errors, infrastructure signals and cache heartbeats.
How are changes released safely to a multi-tenant SaaS platform?
The team rolls out changes to tenant groups in sequence, uses blue/green deployments where applicable and keeps a rollback path through load balancer target groups.
What does the modernization program cover?
It covers containerization, recovery automation, infrastructure as code, CI/CD, identity and alert consolidation, with separate assessments for future container, .NET and AMS operating-model decisions.