challenge
Our client, a global tool manufacturing company, had launched a revolutionary jobsite management application that connects sites, people and equipment and helps users manage their tools and streamline equipment workflows. By making it easier for customers to manage their inventory of tools that were purchased, the app inherently increases customer loyalty and long-term revenue. This multi-tenant application has a robust search feature that accepts filters and partial keywords to find devices across many disparate data sources. One of the primary domain databases was under a constant load, creating performance issues across all users of the applications. This caused large searches from some tenants of the application to impact the performance of smaller tenants. Search results for an empty list could take as long as 45 seconds during mid-peak usage. Their immediate resolutions were costly and not making much impact, causing headaches and frustration for their smaller tenants. They needed a partner to come in and quickly help them achieve high availability and reliability so that the app could perform at its best, therefore helping its users provide a high level of customer service to their own customers.
solution
Our team took a multidisciplinary approach to solving the issues around performance and lack of scalability of our client’s application. We removed CRUD APIs from the .NET monolith that was running the majority of the application and refactored those endpoints to Amazon REST API Gateway routes that executed scalable AWS Lambda functions. We then re-architected the search engine of the application to utilize AWS OpenSearch (formerly ElasticSearch). We deployed a well-architected OpenSearch cluster with appropriate encryption, networking, ingress access, API Gateway integration and monitoring. We also created DevOps pipelines to execute API testing, load tests and validation of new indexes. We used fine-tuned indexes to provide the fastest possible searches and indexing to occur within one second of new data being added. We then created data pipelines to synchronize data from IoT devices, DynamoDB, On-Prem SQL databases and RDS SQL databases to the OpenSearch cluster. This was accomplished via DMS migrations and real-time synchronization, AWS EventBridge, Kinesis, Kinesis Firehose, DynamoDB Streams, ECS Tasks and AWS Lambda. We created the operations (from Jenkins Pipelines and AWS CloudWatch), AWS Architecture design and implementation and backend API solutions. Separate front end and mobile teams consumed the new Search API.
results
With a change to the search architecture, our client was able to scale with their customer base, reduce costs, maintain high availability and reliability and provide a better customer experience. This also helped to increase customer loyalty and retention. The new solution had data searchable within one second of a device reporting or data being added, with the average case being under 100 milliseconds. The new endpoint returned results from the largest data sets within 400 milliseconds (down from 45 seconds), reducing search responses by 99.2%. We also reduced monthly costs by hundreds of dollars. By introducing a cluster service with multiple nodes we increased availability threefold over the previous implementation. All of this was accomplished within four months.
Ensuring 100% Data Accuracy for a Global Tech Company
Our client, a global technology company, needed to launch an internal revenue reporting system for ingesting and tracking federal revenue data. The project required coordination across a large number of teams, ERP systems and virtual desktops, with significant manual work needed to fix data issues. Learn how we helped.
Faster AI-Powered Mortgage Origination Process for Major Bank
A large bank’s mortgage origination department struggled with a time-consuming, labor-intensive application process. Reviewing and analyzing submitted documents took too long, leading to longer processing times and limiting how many applications Document Specialists could handle. The bank sought to automate the process to improve efficiency for both employees and customers. Learn how we helped.
Scaling Dropship Operations to Increase Customer Satisfaction for Global Retailer
Our international clothing and lifestyle client saw increased demand for larger items, but their retail stores lacked storage space. They needed a dropship system that used their warehouses for storage and integrated with their existing processes and inventory system. Learn how we helped.
Unlocking Actionable Intelligence Through Data Transformation
A global payments company needed insight into how marketers were utilizing email contacts so they could convert website visitors into active users for additional revenue growth. However, this exposed long-standing technical debt in their marketing systems. Learn how we built an upgraded platform so they could make better use of their data.