Gart

Client:

A leading Thai jewelry manufacturer.

• Focus:

Azure Cloud Cost Optimization | AI Vision | Hybrid Cloud | Scalable Infrastructure | Large Data Processing

The Challenge

A leading Thai jewelry manufacturer wanted to optimize production by analyzing worker activity across 200-250 workstations. However, processing 2TB of daily video data for real-time insights presented a massive computational challenge. They needed a secure solution for extracting key details (feature extraction & classification) to identify inefficiencies and make data-driven decisions.

• 2-2.5 TB Daily Video

Processing a staggering 2TB of video data daily from each workstation poses a significant computational challenge.

• AI-powered Insights

The client needs to extract key details from video data (feature extraction and classification) to identify patterns and inefficiencies in worker performance.

• Secure Video Data

Safeguarding sensitive worker activity and production data is crucial (compliance & security).

• Cost Optimization

The initial Network Video Recorder (NVR) solution proved to be quite expensive to operate. With 200 workstations each recording 9 hours of video daily, this amounted to approximately 1800 hours of video data that needed to be processed overnight. The client required an optimized solution for handling such large volumes of data in a cost-effective manner.

Our Solution

We calculated that the NVR solution proposed by the developers is quite expensive. It would require a lot of CPU to make the project financially viable.

Gart Solutions proposed a cloud-native, cost-efficient solution leveraging Azure Spot Virtual Machines (VMs) for AI video processing.

Spot VMs use unused Azure capacity at a deep discount, making them ideal for compute-intensive, interruptible workloads like batch video processing. This makes them ideal for workloads like video processing that can tolerate short interruptions.

Here’s a breakdown of the estimated cost savings:

1. Original Cost (On-demand pricing): $5,363 per month (as shown in the table)
2. VMSS Cost with Spot VMs: $430 per month (significant reduction from $4,700)
3. Total Estimated Cost with Spot VMs: $1,100 per month

Projected Savings: $4,263 per month

By utilizing Spot VMs for the VMSS instances, we can achieve a substantial cost reduction of approximately 81% compared to the initial estimate. This translates to savings of $4,263 per month.

Additional Benefits:

• Scalability: Azure’s cloud infrastructure allows scaling Spot VMs up or down as needed, ensuring efficient resource utilization.
• Performance: Spot VMs deliver high performance ideal for demanding workloads like video processing.

Considerations: While rare, Spot VMs can be interrupted when needed by Microsoft for other purposes. But our DevOps architect designed the architecture to handle these interruptions gracefully, ensuring minimal disruption to client’s AI vision processing tasks.

Using Azure’s scalable cloud infrastructure, we built a secure AI vision system. This system seamlessly processed the massive video data, extracted crucial insights, and provided real-time feedback for optimizing production processes.

• Designed a scalable and secure cloud architecture on Azure to handle the massive data volume and real-time processing requirements.
• Developed pipelines for efficient data ingestion, storage, and processing

Results

By implementing Azure’s scalable cloud infrastructure and AI vision system, the Thai jewelry manufacturer achieved significant productivity gains with data-driven insights. This translates to tangible benefits:

• Reduced Operational Costs:

  The client achieved a substantial cost reduction of 81% compared to the initial Network Video Recorder (NVR) solution. This translates to a staggering monthly savings of $4,263. By leveraging Azure Spot VMs, the overall operational cost dropped from an estimated $5,363 per month to a highly optimized $1,100 per month.

• Improved Production Efficiency:

  The real-time insights extracted from video data allowed the manufacturer to identify inefficiencies and optimize production processes. This resulted in increased productivity, leading to a positive impact on their bottom line.

• Real-Time Operational Insights

  AI-powered feature extraction provided actionable analytics from video feeds, enabling rapid identification of workflow inefficiencies, which improved production throughput.

• Scalable & Resilient Infrastructure

  The Azure-based infrastructure scaled elastically and handled transient Spot VM interruptions with zero impact on critical video processing workflows.

Client Overview

Our client is a leading entertainment software platform that operates globally, leveraging AWS to manage its extensive infrastructure and high traffic. The platform supports real-time audio and media streaming, so scalability, cost-efficiency, and performance optimization were critical components of its success.  

Previously, Gart Solutions optimized AWS infrastructure and implemented automation solutions. As our customers experienced significant growth and operations expansion, the need to maintain optimal infrastructure performance and costs became clear. 

Challenges 

Initially, the client relied on AWS CloudWatch for infrastructure monitoring. However, as the system grew more complex, several limitations became evident: 

1. Difficulty in Monitoring Complex Infrastructure: The expanding infrastructure requires real-time monitoring of multiple services, applications, and costs. While AWS CloudWatch was effective for basic monitoring, it lacked the customization necessary for the client’s increasingly complex needs.
    
2. Unclear Billing Information: AWS’s billing structure can be opaque, and the client found it challenging to track costs across different services. This made it harder to optimize resource usage and manage expenses efficiently.
    
3. Developer Debugging Limitations: The native AWS dashboards were not user-friendly for the development team, making debugging time-consuming and difficult. A more intuitive solution was needed to allow quick identification and resolution of issues.
    
4. Refactoring Existing Solutions: When we initially began working with the client’s existing monitoring setup, significant refactoring was required. As new infrastructure features were added, a centralized system that could monitor infrastructure performance, application metrics, and associated costs was needed. 

Solution 

To address these challenges, Gart Solutions implemented a centralized monitoring solution that combined Grafana with AWS CloudWatch. This hybrid approach offered the best of both worlds, using CloudWatch for comprehensive data collection and Grafana for flexible, user-friendly visualizations. 

• Grafana Integration for Custom Monitoring
   
• AWS CloudWatch for Data Collection
   
• Centralized Dashboards and Cost Monitoring
   
• Automated Alerts and Log Monitoring
   
• Refactoring and Infrastructure Scaling 

Why Grafana Over AWS Dashboards? 

Although AWS CloudWatch offers robust infrastructure monitoring features, Grafana was chosen for several reasons: 

• User-Friendliness: Grafana’s intuitive interface allowed the client’s team to easily access and interpret monitoring data without requiring extensive AWS expertise. 

• Customization: Grafana’s dashboards were highly customizable, enabling the creation of tailored views for different teams. This was particularly important for the development team, who needed custom dashboards for troubleshooting. 

• Cost Transparency: Grafana provided greater control over how cost data was displayed, making it easier to identify inefficiencies, whereas AWS’s built-in dashboards often resulted in unclear billing metrics.
• Free and Open Source: Grafana’s open-source nature meant there were no additional licensing costs, aligning with the client’s cost-effectiveness goals. 

Results

The implementation of the centralized monitoring solution resulted in several key benefits for the client: 

• Enhanced Monitoring Capabilities:  

The integration of Grafana with AWS CloudWatch provided a single-pane-of-glass view into all infrastructure and application layers, enabling the team to make informed decisions and respond quickly to issues. 

• Improved Cost Management for $19,9k of Budget Optimization:  

The custom Grafana dashboards delivered real-time insights into resource usage and cloud spending, helping the client optimize AWS costs and avoid unnecessary expenses.
  

• Faster Debugging and Issue Resolution:  

Custom dashboards made it easier for the development team to identify and resolve issues, reducing downtime and improving system stability. 

• Scalability and Future-Proofing:  

The solution was designed to scale with the client’s infrastructure, ensuring that new services and features could be monitored seamlessly as they were added.

Conclusion 

By implementing a centralized monitoring solution leveraging Grafana and AWS CloudWatch, Gart Solutions helped the client gain real-time visibility into their infrastructure and application performance. This solution provided both technical and financial benefits, from enhanced monitoring capabilities to significant cost optimizations ($19,9k per month). 

Our cooperation has succeeded in overriding client requirements. Our customer reviewed us 5 stars and left a testimonial on Clutch.

About the Client

ReSource International is an Icelandic company that specializes in environmental solutions. They have developed elandfill.io, a digital solution for monitoring and managing landfill operations.

Challenge

1. Need for a scalable solution to monitor and manage the elandfill.io platform.
2. Requirement for a cloud-agnostic approach to support future growth.
3. Ensuring the platform could be managed by users with varying technical expertise.
4. Identifying and addressing issues swiftly to minimize downtime and operational disruptions.
5. Need for automation in monitoring processes and notifications.

Solution

To address these challenges, we developed the Resource Management Framework (RMF) – a unified system for managing and monitoring digital solutions for landfills.

The RMF was designed with scalability in mind, allowing for future growth without being tied to a single cloud provider. The platform was initially hosted on Hetzner, which posed a limitation for future scalability. We developed a cloud-agnostic solution imposed certain constraints on the DevOps team but ultimately provided greater flexibility.

The platform’s modular structure includes various components, one of which is monitoring. This component consists of a dashboard that shows the status of the system, including applications, their versions, and their last seen status
 

This dashboard displays:

• List of installed applications
• Application versions
• Status of each service
• Time of last status update (Last seen)

Management Dashboard

This user-friendly interface allows non-technical personnel to monitor production environments. It helps quickly identify issues, such as login failures, by pinpointing whether the problem lies with the UI, backend, database, or performance.

Each new service is automatically added to the dashboard, enabling the platform to expand as needed.

Notifications and alerting

We configured alerts for critical services to notify users of issues such as memory usage exceeding predefined limits. These alerts can trigger scripts to resolve problems automatically.

Integration with Microsoft Teams provides notifications on deployments, ensuring that all team members are aware of changes and potential impacts.

By structuring and representing the collected data, we laid the foundation for the next level of automation. This includes building an alert system that not only notifies but also performs actions based on the alerts, like running scripts to fix issues.

For instance, when a notification about CPU usage spike is received, a corresponding action or script is triggered to resolve the issue.

Results

• Created a universal monitoring system that allowed the project to scale internationally.
• The elandfill.io platform has been successfully implemented in Iceland, France, Sweden, and Turkey.
• Improved capabilities for predicting methane emissions and optimizing landfill management.
• Simplified the process of complying with regulatory requirements for methane emissions monitoring and reporting.
• Ensured flexibility in cloud provider selection thanks to the cloud-agnostic approach.

Predictive Models

By analyzing historical data on methane emissions, waste deposition rates, and environmental factors, the platform develops predictive models to forecast future emissions trends. This enables proactive planning and implementation of measures to reduce methane emissions, such as optimizing landfill cover placement or implementing alternative daily cover practices.

Regulatory Compliance

Landfill operators must meet regulatory requirements for methane emissions monitoring and reporting. eLandfill streamlines compliance efforts by automating data collection, analysis, and reporting processes, ensuring operators remain compliant with regulatory standards and deadlines.

International Expansion

Thanks to our universal RMS and monitoring system, the project has become international, with new sites being installed and Proof of Concept (PoC) trials conducted with several waste management groups worldwide.

The solution is designed for landfills globally, addressing common challenges and facilitating digital transformation and leaner management. Current expansions include sites in Iceland, France, Sweden, and Turkey.

Read Nicolas Proietti’s full feedback on our collaboration