Uniphore stands at the forefront of AI innovation, being the first company designed for scale that seamlessly integrates artificial intelligence into every facet of the enterprise experience. Their robust multimodal AI and data platform harmonizes voice, video, text, and data through a sophisticated blend of generative AI, Knowledge AI, Emotion AI, and workflow automation, acting as a reliable co-pilot for businesses. However, as Uniphore’s operations expanded, they encountered significant challenges with their legacy Windows Server infrastructure, which struggled to keep pace with their growth. The aging system not only incurred rising maintenance costs but also raised security concerns due to its end-of-support status, prompting a critical need for modernization while ensuring business continuity.
To tackle these challenges, Uniphore turned to Amazon Web Services (AWS) to modernize their Windows workloads via containerization and cloud-native solutions. By leveraging Amazon Elastic Kubernetes Service (Amazon EKS) for containerization and Amazon Elastic Compute Cloud (Amazon EC2) for compute resources, along with Amazon Elastic File System (Amazon EFS) and Amazon S3 for storage, Uniphore successfully achieved a remarkable 30% reduction in costs while enhancing their operational capabilities. The migration from a Windows on-premises setup to a Linux environment on AWS optimized their call-center analytics workloads, allowing them to train and fine-tune large language models (LLMs) more effectively. This transition enabled the creation of a unified, high-performance ecosystem tailored specifically for AI and machine learning operations.
Business challenge
Initially, Uniphore’s U-Assist platform operated on 50 bare-metal Windows Server 2008 R2 instances, which soon became overwhelmed as the business rapidly expanded. Frequent storage shortages led to application cache failures and operational disruptions, with data exceeding 60 TB and aging hardware reaching its limits. The existing applications lacked scalability and vendor support, complicating efforts to meet increasing business demands. Additionally, the end-of-support status of Windows Server 2008 R2 raised significant security concerns, as it no longer received security updates or compliance support. These mounting challenges highlighted the urgent need for a modernization strategy.
Solution implemented
Uniphore embarked on a comprehensive modernization strategy utilizing AWS services across three pivotal areas:
1. Application modernization
The modernization process began with the application stack, where the team refactored Java-based services to be operating system-agnostic and implemented containerization using Docker. This required addressing Windows-specific dependencies and externalizing configurations for compatibility with containers. The newly containerized applications were then deployed to Amazon EKS, facilitating true application portability and resilience across cloud environments through features like automatic scaling and self-healing.
2. Custom data migration solution
To ensure a smooth transition, Uniphore developed a custom migration solution utilizing Type 2 hypervisor technology. This innovative approach hosted the AWS DataSync agent, enabling seamless integration with Windows-based storage. High-speed internet connectivity facilitated efficient data transfer to Amazon EFS, enhancing reliability and availability.
3. Cloud infrastructure setup
The modernized infrastructure employed Amazon EC2 and Amazon EKS for compute and container orchestration, while Amazon Virtual Private Cloud (Amazon VPC) ensured secure network isolation. Storage needs were met through Amazon EFS and Amazon S3, with DataSync orchestrating the transfer of data from legacy servers to Amazon EFS, which was subsequently processed by applications running on Amazon EKS. The processed data was stored in Amazon S3, with backups managed through AWS Backup.
Figure 1: Uniphore’s data migration architecture with DataSync and AWS services
Technical implementation
Uniphore executed the migration in a structured, three-phase approach to ensure a seamless transition from legacy infrastructure to a modern, cloud-native environment. Each phase was designed to minimize risk, maintain data integrity, and ensure business continuity.
Phase 1: Hypervisor and DataSync agent deployment
The initial phase involved setting up a Type 2 hypervisor on existing hardware to host the DataSync agent, establishing secure connectivity between on-premises Windows Server instances and the AWS environment. Key actions included:
- Provisioning and configuring the hypervisor for stable operation.
- Optimizing the network layer with dedicated high-speed internet and VPN tunnels.
- Validating connectivity between on-premises storage and Amazon EFS targets.
Phase 2: Secure data migration and validation
With the infrastructure in place, the team migrated over 60 TB of structured and unstructured data using DataSync to ensure reliable transfers. To guarantee data integrity, the following measures were implemented:
- Incremental transfers with checksum validations.
- Real-time monitoring of transfer jobs through Amazon CloudWatch.
- Iterative performance tuning to optimize throughput.
- Validation of migrated data in a separate AWS test environment.
Phase 3: Production cutover and application deployment
The final phase involved transitioning to production workloads on AWS. The legacy Windows-based application stack was containerized and deployed to Amazon EKS, allowing for dynamic scaling independent of the underlying operating system. Key steps included:
- Configuring EKS clusters with autoscaling node groups.
- Setting up Amazon VPC for secure traffic flows.
- Managing application configurations and secrets with AWS Systems Manager Parameter Store.
- Implementing A/B testing for zero-downtime deployment.
- Decommissioning legacy systems post-validation in the cloud environment.
Best practices
Throughout the modernization journey, several best practices were adopted:
- Infrastructure as code (IaC) with Terraform for version-controlled deployments.
- CI/CD pipelines using GitLab for rapid and safe updates.
- Integration of DATADOG and CloudWatch for comprehensive monitoring.
- Enforcement of AWS Backup policies for disaster recovery.
Key benefits
The modernization initiative yielded six significant benefits:
1. Scalability and performance
Before modernization, Uniphore’s rigid bare-metal servers often led to performance bottlenecks. Post-migration:
- Auto Scaling allowed infrastructure to adjust based on real-time demands.
- High availability and consistent performance were ensured during traffic spikes.
- Application latency significantly decreased due to optimized orchestration.
2. Flexibility and portability
The legacy applications were tightly bound to Windows OS, complicating upgrades. After refactoring:
- Applications became OS-agnostic and could run on any Kubernetes-compatible platform.
- Developers gained the flexibility to test and deploy in isolated environments.
3. Enhanced security and compliance
Operating unsupported systems posed serious risks. Post-migration:
- Improved access control and network segmentation were achieved through AWS tools.
- All workloads ran on actively supported OS versions with enhanced security measures.
4. Efficient data management
Data growth had outpaced on-premises capabilities. Post-migration:
- Amazon EFS provided a scalable file system, while Amazon S3 offered high durability.
- DataSync ensured fast and secure data transfer with integrity checks.
5. Reduced operational overhead
Managing on-premises infrastructure was costly. After migration:
- Hardware maintenance and OS upgrades became non-issues.
- IaC and automated CI/CD pipelines improved reliability and consistency.
6. Significant cost savings
The initiative resulted in a 30% reduction in operational costs:
- Cost savings stemmed from shutting down legacy resources and optimizing instance usage.
- Containerization allowed for efficient resource utilization.
