Enterprise e-commerce
and the data under it
AtomicNorth provides enterprise e-commerce solutions for both B2B and B2C markets, with scalable platforms and custom features designed to meet complex business needs.
The storefront is the easy half. What decides whether it scales is the analytics estate behind it.
On the client site in the engagement below.
Served by the same platform.
Legacy warehouse data into HDFS clusters.
Reported after the analytics rebuild.
A US e-commerce firm, rebuilt on Hadoop
Recently, a US-based e-commerce business firm has been benefitted by AtomicNorth while executing a new business analytics solution. The solution balanced cost-efficiency with analytical capability for growing data management requirements.
The client maintained over 250 thousand products on its official website serving more than one million users. Previously utilizing Informatica and conventional MySQL databases, they needed scalable expansion.
Rather than upgrading existing infrastructure, AtomicNorth recommended developing a fresh Cloudera Hadoop-based business solution that would be scalable, customizable and flexible while reducing capital expenditure.
Five steps, each with a confront
Each stage below had a specific problem to solve. They are written the way they were solved, not the way they would look on a method slide.
Migration from Informatica to Cloudera Hadoop
The confront: what is the best way to drift terabytes of complex system data? The solution involved configuring Cloudera Hadoop infrastructure while maintaining real-time data streaming from multiple channels. We employed Scoop scripts to drift all legacy data from the current Oracle data warehouse to the latest Hadoop Distributed File System, managing 15+ terabytes of tangled data into different HDFS clusters.
Creating data processing engines
Processing engines built on the new cluster so the transformation work runs where the data already sits, rather than pulling it back out to a separate tier.
Incremental data transformation
The firm needed automated ETL conversion for various data categories including user information, product inventory, product delivery, logistics and employees' details, run incrementally rather than as a nightly full load.
Advanced analytics
Apache Spark over the cluster for the analytical workload, so cost calculations return in seconds rather than in a batch window.
Integration with Tableau
The assimilation is made at the virtualization layer, which allows real-time generation of reports and outlines of processed Hadoop data.
What the client reported
These are the figures the client reported after the work. They are theirs, in their environment, and are not a forecast for anybody else's.
Seconds, not batches
Enables cost calculation results within seconds.
Latency
Inward leads up 20%+
Reported increase in inbound leads after the rebuild.
Commercial
Better offer quality
Optimized insurance offerings through improved data quality.
Data quality
Traffic up ~20%
Approximate reported increase in website traffic.
Reach
What it was built on
- Cloudera Hadoop
- Apache Spark
- Tableau
- Informatica
- MySQL
- Oracle
- Scoop
Hear it straight from our clients
We have trusted AtomicNorth as our quality Oracle partner over the years. They have always looked into each of our requirements meticulously, whether it be into Oracle EBS or Oracle Cloud, and provided high-quality resources to achieve our goals.
AtomicNorth has remained a trustworthy and our go-to technology partner over the years. They have delivered quality services when it matters the most and their consultants have always been highly methodical in their problem-solving approach.
We have been engaged with AtomicNorth for last four years and it has been quite a satisfactory journey for us. The engagement that started with Oracle EBS reimplementation expanded to include other technologies as well.
Integrated service spectrum
A team with extensive knowledge and deep expertise in the leading platforms customers are deploying today as part of their digital transformation efforts.
Technical Audit
What the estate actually looks like against what the documentation claims, written down.
Discovery Ideation
The problem framed and sized before anybody proposes a platform to solve it.
Tech Strategy and Architect
A target architecture and the order of work to reach it, owned by the people who will build it.
The figures above are one client's reported outcomes in one estate. Treat them as an illustration of the kind of change, not as a projection.
Tell us where thestorefront slows down.
Catalogue size, traffic pattern, and what reporting people are waiting on. That is usually enough to say whether the platform or the data layer is the problem.