CASE STUDIES

MSFT Partners

MSFT Partner Case Study BioCatch

BioCatch using Microservices on Azure with the help of Sela’s SMEs

MSFT Partner Case Study Etoro

Sela Helping eToro reach their goals in Azure

MSFT Partner Case Study Payoneer

Sela using Open Source solutions to win Azure for Payoneer

Expert Services

Sela’s team of talented experts deliver knowledge all over the world – Now available per your demand

BioCatch using Microservices on Azure with the help of Sela’s SMEs

Consumers are more active online than ever, and so are cybercriminals. BioCatch has come up with a better way to catch bad guys, which involves analyzing huge volumes of real-time user behavioral data as transactions are underway. However, the company’s success outstripped its virtual machine-based architecture, and it decided to re-architect its solution using microservices, containers, and open source software. Its new Python-based platform takes advantage of containers, Apache Spark, and open source DC/OS running in Microsoft Azure Container Service. With its new architecture, BioCatch has improved performance exponentially, slashed costs dramatically, and can enhance its product daily versus quarterly.

Sela has a track record of over 3 decades in leading the consulting and technology knowledge in Israel. Sela, as an MSP helped BioCatch reaching their business goals on Microsoft Azure.

Reduce fraud with passive, continuous authentication

Behavioral biometrics studies how individuals use their mouse, tap their keyboard, and hold and touch their smartphones. BioCatch analyzes more than 500 such parameters to create a unique user profile that cannot be lost, stolen, or imitated.

What’s great about the BioCatch approach is that it protects the integrity of a user session through its duration. It’s totally invisible and doesn’t require secret questions or other interruptions to the user experience.

Behind the scenes, BioCatch is crunching huge amounts of data as it captures and analyzes thousands of events for every user session. An event can be a simple mouse movement, a change in a smartphone angle, or a measure of how much pressure a user applies when tapping an onscreen button.

Multiplied by millions of user sessions each month, BioCatch has a very large big-data application. When the company was founded in 2011, its engineers designed the BioCatch 1.0 analytical platform to handle up to 5 million simultaneous users—which sounded like a really large number at the time. However, when the BioCatch service was a runaway hit and traffic skyrocketed from 2 million to 8 million users in the first year, the company’s underlying architecture struggled to keep up.

“Although we were born in the cloud, with our solution running in Microsoft Azure from day one, we had a traditional, monolithic code base and a virtual machine-based infrastructure,” says Dekel Shavit, Director of Operations and Chief Information Security Officer at BioCatch. “We couldn’t make code changes quickly enough to keep up with customer needs and fraudster moves, and the cost of a VM-based infrastructure was too high. We needed lower costs and better performance on an elastic scale.”

Re-architect for containers and open source

A participant in the Microsoft BizSpark Program, which helps entrepreneurs get off the ground, BioCatch availed itself of Microsoft cloud experts for advice. They recommended switching to microservices and containers. “Containers, and specifically Azure Container Service, fit our needs like a glove,” Shavit says. “We were reassured that the technology was able to handle the needs of even our largest financial services customers.”

BioCatch worked together with Sela’s SMEs to select open source DC/OS as its container orchestrator, which builds on Apache Mesos as its cluster manager and the Apache Spark data compute framework. “By using Azure Container Service with open source DC/OS, we are able to use a single scheduler for both Apache Spark workloads and Docker containers, thus simplifying our architecture,” Shavit says.

In fact, BioCatch used open source software to rewrite its entire platform for microservices. “Open source software was most cost-effective and gave us the greatest performance optimization options,” Shavit says. “We migrated all our algorithms from .NET Framework MATLAB to Python, which makes it much easier to do real-time changes.”

The combination of Architects from the Microsoft team together with Sela’s SMEs worked with BioCatch to come up with the best tools for every need. “I was frankly shocked at the amount of open source knowledge that Microsoft had and its openness in recommending technologies that were best for us,” Shavit says. “With Microsoft, we can get the support we need, no matter which tools we’re using. I don’t need to have DC/OS, Redis, and Python experts but instead can call Microsoft and get immediate help with all these tools, because it knows how to run big-data open source applications in the cloud.”

The BioCatch team adopted a gradual approach to implementing BioCatch 2.0. It still runs Azure Virtual Machines and containers side by side until it can rewrite all aspects of its application. Still, with a sprint-oriented DevOps methodology, BioCatch moved from proof of concept to a production product in Azure Container Service in about two and a half months. “We found that the hard part wasn’t the coding but changing our operational processes,” says Shavit. “Instead of upgrading and deploying a monolithic code base once a month, we needed to create continuous innovation/continuous deployment processes and think in terms of smaller feature releases.”feature releases.

BioCatch has always used a DevOps development methodology, but the addition of microservices has enabled its DevOps team to work in near-real time. “Before microservices, our DevOps guys were basically enhanced IT guys,” Shavit says. “They didn’t help push code into production. Today, they are pushing code like any other developer. We’ve empowered every developer to be a tester, scaler, and quality assurance person.”

Re-architect for cImprove performance while slashing costs dramaticallyontainers and open source

A participant in the Microsoft BizSpark Program, which helps entrepreneurs get off the ground, BioCatch availed itself of Microsoft cloud experts for advice. They recommended switching to microservices and containers. “Containers, and specifically Azure Container Service, fit our needs like a glove,” Shavit says. “We were reassured that the technology was able to handle the needs of eveWith its new architecture, BioCatch has seen average response times improve by more than 1,500 percent. “By re-architecting our platform using microservices and containers, we can scale much faster and more easily,” Shavit says. “Previously, if a customer made a change on its website that resulted in a fivefold increase in traffic, it took us a long time to ramp up our infrastructure to support that load. Today, we can ramp up in about 20 seconds, and it happens automatically.”n our largest financial services customers.”

A participant in the Microsoft BizSpark Program, which helps entrepreneurs get off the ground, BioCatch availed itself of Microsoft cloud experts for advice. They recommended switching to microservices and containers. “Containers, and specifically Azure Container Service, fit our needs like a glove,” Shavit says. “We were reassured that the technology was able to handle the needs of eveWith its new architecture, BioCatch has seen average response times improve by more than 1,500 percent. “By re-architecting our platform using microservices and containers, we can scale much faster and more easily,” Shavit says. “Previously, if a customer made a change on its website that resulted in a fivefold increase in traffic, it took us a long time to ramp up our infrastructure to support that load. Today, we can ramp up in about 20 seconds, and it happens automatically.”n our largest fiThis means that BioCatch can go after customers that it couldn’t approach before, add more behavioral characteristics to its product, support more end users, and, ultimately, better keep up with fraudsters, who never slow down. “We can now grow the business without worrying about technology limits,” Shavit says. “Every month, we process exponentially more data.”nancial services customers.”

Just as important, BioCatch has reduced costs dramatically by moving from virtual machines to a container-based architecture. The company is redirecting that savings into breaking into new markets and becoming even more data-oriented.

Innovate faster

Another benefit of deploying a microservices architecture is the ability to enhance the product at a faster pace. “Previously, we pushed out a quarterly release. Today, we are pushing out updates up to 15 times a day, for both functionality and performance improvements,” Shavit says. “Faster innovation is critical in our business, because fraudsters are fast. We need to react on a daily basis to address new threats. Also, our customers are always pushing us to do more. We don’t want to tell them, ‘We’ll get that to you next quarter.’”

With microservices in Azure, BioCatch can much more easily stay ahead of the bad guys

Original use case was publish on: https://customers.microsoft.com/en-us/story/biocatch


For more Case studies click here

Sela Helping eToro reach their goals in Azure

More than 6 million registered traders in over 140 countries have taken a new approach to investing their hard-earned money. Through the eToro Social Trading Network, traders and investors can buy and sell a wide variety of financial instruments, while following and automatically copying the investment decisions of other successful traders in real time. This makes their trading experience more social, simple, enjoyable, and transparent. In early 2015, eToro moved the bulk of its platform from on-premises datacenters to Microsoft Azure. Soon after that, it added Azure analytics tools, including Microsoft R Server and Azure Machine Learning, to expand its new CopyFunds offerings. With the machine learning version of CopyFunds, investors can automatically duplicate market strategies and actual investment decisions of successful traders in the eToro network to use in their own portfolios.

Sela has a track record of over 3 decades in leading the consulting and technology knowledge in Israel. Sela, as an MSP helped eToro reaching their business goals on Microsoft Azure

What if you could automatically align your investments with those of highly successful investors who have similar philosophies and risk tolerance, without having to pay a huge consulting fee? Millions of traders all over the world are doing just that through the eToro social investment network. Unlike traditional online trading—where investors must learn, understand, and develop their own individual investing strategy—the eToro platform offers CopyTrader technology that combines online investing tools with social media, so investors can share their own trading activities and see and copy the strategy of other members.

“In the eToro platform, you can identify network members who are in line with your investment philosophy, risk appetite, and/or specific industry sectors of interest, and immediately start learning from, interacting with, and even automatically copying those individuals in real time,” says Israel Kalush, VP of Engineering at eToro.

eToro was founded in 2007 with the promise of making trading available for the masses by transforming it into a simple, transparent, and enjoyable experience. Three years later, the company launched eToro OpenBook social investment platform, combining online investing with social media to help investors at all levels apply the approach used by the most successful traders. With a wide range of mobile and web-based social trading tools, eToro appeals to every level of trading expertise and helps new investors gain valuable insight into how successful investors operate.

The results can be quite remarkable. In 2015, more than 80 percent of the 124 million copied trades transacted through eToro posted a profit. Of course, past performance is no guarantee of future results and eToro continuously reminds members that trading involves risk.

To help reduce that risk, copy traders can set a stop-loss threshold, so if their losses exceed that amount, the system automatically closes the copy process. The system also assigns a risk score to each trader, so users can copy those who match their risk appetite. Furthermore, the platform does not allow copying high-risk traders (risk score 8 and above), which helps protect novices. In addition, eToro encourages beginners to practice through a virtual portfolio before using real money for their trades. The firm complies with regulatory standards from the Financial Conduct Authority in the UK and Cyprus Security and Exchange Commission.

Seeking cloud scale opportunities

In 2015, eToro moved to Microsoft Azure to increase reliability, scalability, and development agility, and to reduce support requirements to pave the way for its next level of growth. The ability to run open source solutions in Azure also played a big role in the decision.

“We use several open source solutions, including Redis Cache, and our front end is based on Node.js and AngularJS. So it’s a big plus to be able to use those technologies inside the Azure platform,” says Kalush. “Every developer we have comes with a strong knowledge of open source technologies. The ability to leverage that knowledge and go with open source where it makes sense, and with Microsoft native solutions where they make sense, makes our lives much easier.”

Working with Microsoft and with the help of Sela’s experts, it took the firm about a year to develop the new eToro platform. Part of that time was spent rebuilding some of the existing elements to optimize them for the cloud. But a large part of that time was spent creating new features and capabilities to take advantage of the Azure platform.

Harnessing the wisdom of successful traders

Not long after moving to Azure, eToro began developing its latest breakthrough product, called CopyFunds, which are customized portfolios based on a specific theme or industry (also known as thematic investments). “Investors only have to decide what theme they want to invest in—such as banking, technology, or more recently, cryptocurrencies—and how much money they want to invest,” says Kalush. “From that point on, the system does the work.”

Investors can choose from two types of CopyFunds:

  • Market CopyFunds™ consist of specially selected instruments such as stocks, commodities, or ETFs managed either by eToro or a select group of third-party partners. These portfolios allow investors to track an array of sectors around a defined market strategy.
  • Top Trader CopyFunds™ consist of the best-performing and most sustainable traders in the eToro network, who are selected by various means, based on the CopyFund strategy and risk factors. For example, the MomentumDD CopyFund uses Microsoft R Server on Azure HDInsight, and Azure Machine Learning technologies.

“Instead of investing directly in markets and in assets, our machine learning CopyFunds invest in talent that exists within our community, basically copy trading that talent,” says Kalush. “So when you open the CopyFunds portfolio, you will see that it's composed solely of traders in our community. The magic behind it is that the CopyFund, using Microsoft R Server and Azure Machine Learning, essentially handpicks the talent in that group so you have the wisdom of the crowd working for you instead of trying to beat the market by yourself.”

eToro uses Microsoft R Server to unlock knowledge related to trends, behaviors, and predictions in conjunction with Apache Spark for Azure HDInsight, which it uses to execute deep-learning algorithms that help uncover new insights. Through the algorithm, the CopyFund selects the group of investors and decides how to invest and how much money to allocate for each of those individual’s portfolios, factoring in how much risk each investor contributes to the CopyFund as a whole. The algorithm is executed periodically to keep the portfolio of traders efficient and current.

“We are obviously picking the best talent that we have in our network, but our algorithm also takes into consideration that each individual component is a risk factor,” says Moti Goldklang, Director of Trading Systems R&D for eToro. All the while, investors have complete visibility into what's going on in the CopyFund and can see both the investors and the actual trades that were executed by them.

Kalush says, “We started a year ago with almost no machine learning experience, and now we are running quite an impressive machine learning operation, thanks to the Azure platform itself and the expertise and advice that our partners from Microsoft provided along the way.”

As part of the CopyFunds project eToro built a simulation platform, using Azure Web Apps and Microsoft Power BI for both the presentation layer and for data manipulation, to analyze the performance of the platform. This allows eToro to run various types of machine learning experiments and analyze how each affects the performance of the algorithmic CopyFunds.

“What made the biggest difference for us in this project was that Microsoft really treated us as a partner,” says Kalush. “They took our company, which at that point relied on legacy systems and technologies, and helped us through a not trivial ramp-up to the cloud tech stack. Through the entire journey, we found Microsoft to be way more than just a vendor and much more like a partner that helped us pick the right technologies, design our solutions, and made the journey to the cloud much more feasible, economical, and pleasant.”

Original Use Case was published on: https://customers.microsoft.com/en-us/story/etoro

Re-architect for cImprove performance while slashing costs dramaticallyontainers and open source

A participant in the Microsoft BizSpark Program, which helps entrepreneurs get off the ground, BioCatch availed itself of Microsoft cloud experts for advice. They recommended switching to microservices and containers. “Containers, and specifically Azure Container Service, fit our needs like a glove,” Shavit says. “We were reassured that the technology was able to handle the needs of eveWith its new architecture, BioCatch has seen average response times improve by more than 1,500 percent. “By re-architecting our platform using microservices and containers, we can scale much faster and more easily,” Shavit says. “Previously, if a customer made a change on its website that resulted in a fivefold increase in traffic, it took us a long time to ramp up our infrastructure to support that load. Today, we can ramp up in about 20 seconds, and it happens automatically.”n our largest financial services customers.”

A participant in the Microsoft BizSpark Program, which helps entrepreneurs get off the ground, BioCatch availed itself of Microsoft cloud experts for advice. They recommended switching to microservices and containers. “Containers, and specifically Azure Container Service, fit our needs like a glove,” Shavit says. “We were reassured that the technology was able to handle the needs of eveWith its new architecture, BioCatch has seen average response times improve by more than 1,500 percent. “By re-architecting our platform using microservices and containers, we can scale much faster and more easily,” Shavit says. “Previously, if a customer made a change on its website that resulted in a fivefold increase in traffic, it took us a long time to ramp up our infrastructure to support that load. Today, we can ramp up in about 20 seconds, and it happens automatically.”n our largest fiThis means that BioCatch can go after customers that it couldn’t approach before, add more behavioral characteristics to its product, support more end users, and, ultimately, better keep up with fraudsters, who never slow down. “We can now grow the business without worrying about technology limits,” Shavit says. “Every month, we process exponentially more data.”nancial services customers.”

Just as important, BioCatch has reduced costs dramatically by moving from virtual machines to a container-based architecture. The company is redirecting that savings into breaking into new markets and becoming even more data-oriented.

Innovate faster

Another benefit of deploying a microservices architecture is the ability to enhance the product at a faster pace. “Previously, we pushed out a quarterly release. Today, we are pushing out updates up to 15 times a day, for both functionality and performance improvements,” Shavit says. “Faster innovation is critical in our business, because fraudsters are fast. We need to react on a daily basis to address new threats. Also, our customers are always pushing us to do more. We don’t want to tell them, ‘We’ll get that to you next quarter.’”

With microservices in Azure, BioCatch can much more easily stay ahead of the bad guys

Original use case was publish on: https://customers.microsoft.com/en-us/story/biocatch


For more Case studies click here

Sela using Open Source solutions to win Azure for Payoneer

Customer profile - Payoneer is a leading ISV in Israel, enabling millions of businesses from more than 200 countries to facilitate seamless, cross-border payments. Recently ranked in the top 100 of Inc. 5000’s Financial Services companies (second to PayPal), Payoneer supports customers such as Airbnb, eBay and Amazon and transacting billions of dollars each month.

Sela has a track record of over 3 decades in leading the consulting and technology knowledge in Israel. Sela, as an MSP helped BioCatch reaching their business goals on Microsoft Azure.

Payoneer business problems that were tackled

Sela was approached by Microsoft to help a big potential AWS customer, Payoneer. Our SMEs, together with Microsoft Architects and Payoneer engineers identified the following challenges: Using Sela’s Managed Services, your organization may leverage its IT services and extend far beyond current capabilities. Using various scenarios you may choose to
1.  Identify new business models and challenges
2.  High operational fee due in payment transactions
3.  Scaling issues for supporting large data volumes

In order to be able to offer their payment service in China, Payoneer (https://www.payoneer.com/main/) had to meet some Chinese market regulations:

  • Timeline considerations were very limiting, and the time they had to finalize their deployment was not sufficient.
  • Payoneer wanted to use Cloudera’s solution for their big data requirements but they faced a lot of issues when trying to integrate Cloudera’s solution.

Payoneer had a few problems:

  • Constant analysis of big data
  • Generation of reports for management

How did Sela helpd?

Sela has a unique positioning as an SI, a Software House and a Training Partner. This enables us to give a 360 to the client from the Infrastructure, through DevOps, to the Data and Development, not only giving them the best solution, but also to make a transformation within the teams, by providing a comprehensive knowledge transfer and a state of mind transformation. We at SELA believe we should forget about “Lift & Shift”, it is not a unique enough solution and it leaves the conversation around pricing. You can Win Compere in some cases, but even if you do that, other vendors will win back eventually. We found that the best competitive solution and the most robust one is “Lift & Transform”. We find the best business and technology solution against the current IaaS AWS architecture. Usually it involve in Azure PaaS elements, in most cases a combination of Big Data, AI, Containers and specific Azure native components (AKS / Azure SQL / Cosmos DB / Data lake / ML / etc.). This “Transformation” from IaaS into PaaS and Services brings the best of Azure, both in technology and pricing and increases the client’s stickiness in scale. In addition, SELA has developed unique IPs of its own: OneDrop (http://msp.sela.co.il/onedrop.html), Dr. Dump (http://www.sela.co.il/Products#_drdump) and 3 DevOps tools (Scrat - http://www.sela.co.il/Products#_scrat, WimBi http://www.sela.co.il/Products#_wimbi, WiMigrator - http://www.sela.co.il/Products#_wimigrator).

The Solution Areas

The solution covered combines the following areas: Open Source, Modern Workplace, Data and AI, and Infrastructure.

The Main Benefits Achieved

  • Cost Reduction by implementing predictive and corrective measurements to the transaction process
  • By building a new AI model with Cloudera on Azure, Payoneer was able to gain new business insights from data to improve the business model
  • Payoneer originally based their solution on AWS (account of more than 1M USD)
  • Our solution helped them reduce their costs to an average of 40K USD per month over Azure in a period of less than 3 months.
  • The service we provided enabled Payoneer to meet their strict timeline and as a result – they can offer their services in the Chinese market and gained a contract of over 3M USD.

The main technologies that were involved in the solution

  • SELA’s experts delivered the following infrastructure solution for Payoneer (All the below technologies were deployed over Payoneer’s new Azure subscription):
    • Established a Spoke and Hub network configuration
    • The Hub includes central management tools:
      • Cloudera Director
      • Cloudera Repo
      • Linux Repo
      • Chef repo
      • Redundant connection to Payoneer’s datacenters.
    • The Spokes serve as separators across Payoneer’s different environments – Development, Staging and Production


  • For a streamlined DevOps best practices, SELA’s experts created:
    • Custom templates to adjust Cloudera Director to Payoneer’s solution and their mode of operation.
    • Developed automation for creating the Hub and the different Spokes and for establishing Cloudera based on Cloudera Director’s automation.
    • Created local Repos (not auto-synced) - this enabled Payoneer to prevent internet access from the different Spokes, a requirement set by their Information Security Dept.
    • Connected Payoneer’s on-premise datacenter management systems to the Azure-based solution for authentication and configuration management needs.

Payoneer’s impression from working this us

“Working with SELA proved to be a key factor in our success in meeting our challenge of the Chinese contract. SELA’s experts helped us design and implement a solution which needed to be very custom to our needs and deployed to Azure. The team accompanied us during all the phases of activity, from architecture and infrastructure design, the implementation and deployment of the design and its configuration. The value SELA brought to the table surpassed the technological domain and impacted our business directly by enabling our expansion into new markets. Working with SELA was a very good experience for us, through the engagement SELA kept as transparent as possible and their team made sure they don’t leave before a comprehensive knowledge transfer was carried out – so we can more easily maintain our system.”

- Yaron Weiss, VP IT-Operations & corporate security, Payoneer

For more Case studies click here