• PAYPAL LEVERAGES HPC TO
      SUPPORT COMPLEX
      ANALYTICS REQUIREMENTS

PAYPAL

“We needed a more robust alternative that would enable us to watch events for a longer period of time, with shorter latency so that we could react faster,” notes Ryan Quick, PayPal’s principal architect for global platform infrastructure. “We also needed wider bandwidth to consume more data in the same time increment, and faster I/O would enable that while augmenting analytics.” In addition to stellar random and sequential I/O performance, PayPal wanted seamless InfiniBand integration as the company’s previous network topologies had proven too slow, and scaling was overcomplicated. Watch the video below to learn how HPC technology enabled PayPal to meet their robust requirements.

The eBay company PayPal, based in San Jose, California, enables global e-commerce by supporting more than 143 million active accounts and 26 currencies. Because PayPal continually collects and correlates massive amounts of information to optimize the convenience, flexibility, and security of online shopping, the company relies heavily on leading-edge technologies to enhance its various websites and platforms. With thousands of code paths and hundreds of transaction types, it became clear that traditional scale-out technology couldn’t provide sufficient real-time functionality to support PayPal’s analytics requirements.

#HPCMatters to PayPal
Ryan Quick, Principal Architect for Advance Technologies at Paypal, talks about how #HPC transforms online payment transactions.
CHALLENGES

  • To provide sufficient real-time functionality to support PayPal’s analytics requirements
  • To follow transaction workflows and provide deep-line analytics in “real-enough” time
  • To find a solution that eliminates data silos by unifying application data on a single platform with a single storage engine

SOLUTION

  • DDN SFA high-performance storage engine to keep pace with escalating hyper-scale analytics requirements

BUSINESS BENEFITS

  • Expedited ROI
  • Ability to solve hyper-scale data problems in real time
  • Price-performance, ease of use, and versatility in addressing a variety of data-intensive workloads

TECHNICAL BENEFITS

  • Converged InfiniBand with Ethernet data network avoids the need for a separate storage and data network that provides an eight-time performance boost
  • Data files and objects are written directly to disk while preserving valuable cache for random I/O transactions
  • The ability to accelerate multiple data access applications on the same centrally stored files without need for copying or serial locking
  • Centralized management of storage coupled with parallel file systems provides the POSIX compliance expected by data-intensive applications like databases and NoSQL systems, and provides access to that same data across the entire cluster simultaneously

RESULTS

  • Expedited resolution of real-time, hyper-scale data problems
  • Increased ROI, price-performance, ease of use, and versatility in addressing a variety of data-intensive workloads

MORE CLOUD & WEB SUCCESS STORIES

PRODUCTS & SOLUTIONS USED

INDUSTRY