How Twingo helped Oribi 10x their query speed by adjusting the data architecture

As the demand for Oribi's platform grew, data architecture had to be upgraded to support the new reading and writing loads.

INDUSTRY: Internet Analytics
SOLUTION: An easy Marketing Analytics Tool that tells you what to do next.
TECHNICAL USE  CASE:  S3, Spark, and Presto-based Data Architecture that impoved query speed by 10X.

Oribi is a Web Analytics Insights tool, designed to aid site owners and Marketers understand what is needed in order to increase conversions and make their digital assets, whether blog or ecommerce website, successful. For that to happen, Oribi gathers and monitors multiple recurring events on its customers’ sites. This makes Oribi a Big Data company with a gigantic pipeline of services and data components that are responsible for data gathering and data production, already stored and aggregated in a from which is easy for querying. Querying is done via a unique Dashboard that enables the user to identify the current trends on its site, and extract issues to review and areas to invest in.

The Challenge

The application on which the Dashboard, used by Oribi‘s customers, is based query the system’s backend, which is composed of a series of data tools and microservices. In time, as the demand for Oribi’s platform continued to grow along with our customers’ sites traffic, an adjustment of the data architecture was needed to comply with the mass data and the updated reading and writing loads.

One of Oribi‘s main features relied on a Relational Database Management System (RDBMS) based Database, which was customized to comply with the system’s needs. Only, like most RDBMS-based solutions, its scaling abilities were limited. Hence we started searching for a database that can significantly increase computing abilities.

Oribi Logo

10x

Faster Query Time

More

Data Processed

Better

End-User Experience

The Solution

 When Oribi approached us and after identifying the needs, a few possible solutions were evaluated. To make a final decision 2 POC’s were established in production, both running for 6 weeks along with the existing RDBMS-based solution, which continued working as usual.

The POC results were an operation matrix that enabled the best possible decision, defined by pre-set parameters of Performance, cost, learning curve, maintenance requirements, scaling abilities, update and rectify historical data abilities, and support of all of Oribi’s features.

“Ilia Gulman, Twingo CTO, helped us define the research, understand what it is we are looking for, both micro and macro, in building the Queries,” says Eynav Mass, Oribi VP R&D. “Since our environment is unique – we run with Kubernetes above spots – Twingo was a great help in adjusting to this environment.”

Eventually, an S3, Spark, and Presto solution were chosen, since it was most suitable for Oribi, both in the aspect of cost-affect and future business plans. Twingo continued to help us with proper adjustments and best-fitting configuration.

“Twingo helped us focus on the problem at hand, realizing the relevant solutions, and built a plan for every solution that enabled us to understand what we need to address and what can be reviewed later”, adds Eynav. “We started from scratch with no infrastructure”. 

Ilia Gulman, Twingo CTO, helped us define the research, understand what it is we are looking for, both micro and macro, in building the Queries." Eynav Mass, VP R&D, Oribi

The Products

 The transition to S3, Spark, and Presto-based Data architecture improved ten times the query speed, significantly increased the data volumes that can indicate insights, and simplified the whole process. Bottom line, the behind-the-scenes process improved the end-user experience significantly and enabled Oribi to manufacture the Data infrastructure to support its business growth.

“what I loved about Twing was how important it was for them to understand our needs. Golan and Ilia were here for us, to support and help, not with the intent to sell.” Mass adds in conclusion.

Twingo helped us focus on the problem at hand, realizing the relevant solutions, and built a plan for every solution that enabled us to understand what we need to address and what can be reviewed later. We started from scratch with no infrastructure". Eynav Mass, VP R&D, Oribi