EvoApp's Bermuda platform addresses the challenges of big data by enabling real-time analysis of massive datasets through iterative, ad-hoc querying of both structured and unstructured data in a cloud-native environment. Bermuda leverages virtual machines, cloud storage, and in-memory processing to handle terabytes of data across servers with sub-second response times. This allows for flexible, low-cost analysis of large datasets for applications such as dynamic pricing, predictive maintenance, and fraud detection.
2. The Problem We’ve Solved
EvoApp is the first to simultaneously address all four elements of the world’s growing Data Challenge:
As data volume and velocity grow, ad-hoc queries cost more while real-time performance suffers -
becoming too slow for human interaction on large datasets.
The Data Challenge The Solution
EvoApp Bermuda TM
(patent pending)
• Volume: Leverage virtual machines and cloud-
Volume Velocity
scale storage systems
• Velocity: Accommodate terabytes of
information across multiple servers
• Time: Conduct scan-intensive, real-time
queries without pre-defining specifics
• Cost: Combine the scalability of NoSQL with in-
Query Time Query Cost memory performance and cloud economics
EvoApp in 2011 EvoApp Today
3. EvoApp’s Bermuda Advantage
Bermuda enables BI-class, real-time analysis of massive datasets:
• Iterative, ad-hoc querying for unstructured and structured datasets
• Cloud-native, lower ownership and deployment costs
• Deployable in public and private clouds
• Flexible, low-cost, cloud native
• Runs on standard hardware
Data Volume in Terabytes
Substantial Cost/Resource Advantage
• Significant capital investment • Batch processing
• Rigid deployment requirements • Optimized for throughput
10 20 30 40 50
Query Time in Seconds
4. How It Works
Remote Network Storage
Collection of data items
Key Characteristics:
1. Unique data packaging strategies
for optimal memory saturation.
2. Maximizes use of main memory at
scale. No disk drive interaction.
Data retrieval
3. Data loading time minimized.
Parallel Compute Nodes Query evaluation & distribution
Calculation Calculation Calculation Calculation 4. Intelligent query distribution at
maximum scale.
5. Sub-second response drives real-
time user experience
Merge Calculation
Node 1 Node 2 Node 3 Node n
In-memory data items Query Result External
Query
5. Sample Use Cases
E-Commerce Manufacturing Finance
The Company: Real-time pricing The Company: A major aircraft The Company: A major international
optimization solutions provider for the manufacturer serving commercial and provider of SaaS accounting solutions for
airline and hospitality industries. defense clients. small and medium businesses.
The Problem: The Problem: The Problem:
• 99% of current implementations are • Each aircraft has thousands of sensors • Managing regulatory compliance of
on-premise or hosted emitting a constant stream of time- their SaaS and mobile services
• Millions of online purchase varying data that is recorded to the transactions (e.g.
information, price views and plane’s black box and uploaded after invoicing, payment, superannuation)
cancellations each flight. • Sometimes fraud may go undetected
• Serious scalability problems due to the • Data is processed in batches to help due to the inability of batch processes
volume and velocity of data and the technicians predict/analyze the to handle real-time, ad-hoc queries
need for real-time queries airplane’s maintenance needs across these large datasets
The Solution: Bermuda’s real-time The Solution: Bermuda’s real-time, ad- The Solution: Bermuda’s ad-hoc querying
analytics capability enables: hoc analytics capabilities enable: capabilities enable:
• Adjusting dynamic pricing to maximize • Predicting maintenance events in- • Analyzing patterns in the data to alert
customer revenue at point-of-purchase flight, hours before landing – reducing human auditors about possible fraud
• Detecting anomalous events (e.g. turnaround time • Real-time exploration of potential
attempts to game pricing through • Improving the efficiency of costly fraud cases resulting from these alerts
denial-of-service attacks) maintenance resources