Liaison Technologies: Going Big (Data) with MapR Hadoop and Cisco UCS

Liaison Technologies provides cloud-based solutions to help organizations integrate, manage, and secure data across the enterprise. A few years ago Liaison noticed the increasing market trend of blending data integration and data management needs. In response, the company decided to add data management capabilities to its strategic cloud platform to enable big data applications.


Liaison Technologies’ big data management cloud platform requires a high-performance, flexible, secure, and robust infrastructure.

Challenges

  • Ensure high availability for data management services.
  • Maintain compliance with healthcare regulations
  • Provide sufficient computing power to execute complex algorithms quickly

Big data management involves organizing, administering, and governing large volumes of structured and unstructured data across the entire information lifecycle. As Liaison’s data architects began to design the company’s next generation platform, they realized that the existing platform lacked the performance and flexibility to meet the needs of big data.

One of Liaison’s first applications involved clinical trials. This research work on human populations, required for new treatments, produces a significant volume of data including patient demographics, doctor visits, doctors’ institution affiliations, lab results, and patient outcomes. A typical trial can comprise 5000 patients, hundreds of testing laboratories, and thousands of providers.

Because these services are vital and cannot go offline for any length of time, high availability is paramount. In addition, offering services in the healthcare market subjects Liaison to a number of regulations, such as the Health Insurance Portability and Accountability Act (HIPAA) and the Health Information Technology for Economic and Clinical Health Act (HITECH), so the platform has to support audits and other compliance activities.

Cisco UCS platform enables high-performance computing.

Solutions

  • Cisco UCS platform provides the computing power to process complex analysis algorithms in a dense, scalable form factor.
  • MapR Distribution offers a robust, secure, enterprise-grade framework for cloud-based data management services.
  • MapR Distribution flexibility allows Liaison to plug in the right software tools to deliver customized, high-value solutions.

MapR performed better than other Hadoop distributions
Liaison conducted a thorough evaluation of the available solutions, focusing on two specific technology needs: the software framework to store and process big data in a distributed system, and the computing hardware to support the software. Early in the evaluation process, Liaison’s technical team investigated Apache Hadoop, an open source software project that enables distributed processing of large data sets across clusters of servers.

But which distribution of Hadoop?
Liaison evaluated the top commercial distributions to determine which one best met its requirements. This included a proof of concept (POC) to measure the performance and reliability of Apache HBase versus MapR Distribution including Apache Hadoop. The MapR solution excelled in both performance and availability, with enterprise-grade features such as hardening and no single point of failure. Given these and other advantages, the final decision to go with MapR Distribution was easy.

For the hardware, Liaison chose the Cisco® Unified Computing System™ (Cisco UCS®) platform because of its need for high performance. Liaison’s data management services are more computing intensive than its data integration services. Take, for example, a clinical trial with patients named Jon Smith, John Smith, and Jonathan Smith. For the clinical trial results to be meaningful, researchers need to know if these entries refer to the same person or different people.

MapR provides iron-clad security for multi-tenant services. Brad Anderson, Liaison’s vice president of big data informatics, explains how Liaison addresses this issue: “Our machine-learning algorithms determine whether similar entries represent the same person with a high degree of confidence by accessing and comparing information from multiple sources including diagnoses, doctor visits, hospital records, and laboratory test results. Getting the answer from the large data set in a 5000-person clinical trial requires a lot of computing power, but it gives us an edge over competitors who require a number of manual steps to perform this kind of analysis.

"We chose the Cisco Unified Computing System because of its dense form factor, scalability, and high performance. Our computational power is now about 80 times greater than our legacy server infrastructure, an improvement of nearly two orders of magnitude."

Security is second to none

From the beginning, Liaison realized that a multitenant solution—several customers sharing the same computing cluster—was the optimum architecture, but only if Liaison could ensure the security of each customer’s information. MapR support for multitenancy was crucial to Liaison and was not available from the other distributions. “Our Volumes feature creates a logical separation of data, which allows us to securely host several customers on the same cluster,” says Anderson. “MapR gives us iron-clad security for our multitenant cloud-based service."

Results

  • Cisco UCS system offers 80 times the improvement in performance over the legacy server infrastructure.
  • Cisco and MapR delivers 99.999% availability.
  • Compliance activities are simplified thanks to enterprise-grade features of MapR Distribution
  • Liaison complies with EU data movement restrictions using MapR fine-grain control of disaster recovery operations.

Success is on the horizon
As a cloud services provider, Liaison requires extremely high availability. Any significant amount of downtime can hamper the company’s ability to deliver results to its customers on time, which can negatively impact revenue, customer loyalty, and brand value. ”In traditional environments, if you’re using Hadoop for data analysis, and it’s down a day, that’s not a disaster,” says Anderson. “For us, our Hadoop environment is mission critical. We simply can’t have downtime."

The flexibility of the Cisco and MapR solution meets the unique needs of individual customers, particularly in the area of database software. Clinical trial researchers want answers to queries such as, “Find all the doctors whose patients had glucose levels within a certain range and had good outcomes in the last six months.” A relational database might take hours or days to process that query, while a graph database can return the answer in a minute or two. No matter what the requirement, Liaison is confident about its ability to deliver. ”The MapR and Cisco platform allows us to choose the best tool for our customers’ use case,” says Anderson.

And because Liaison clients are subject to a host of regulations such as HIPAA and HITECH for healthcare, and Payment Card Industry Security Standard and Gramm-Leach-Bliley Act for financial institutions, “MapR makes it easier to demonstrate streamlined compliance with enterprise-grade features than if we had to build the same functionality into another distribution,” says Anderson.

Disaster recovery is another area where the Cisco and MapR platform shines. “As a cloud provider, the ability to safeguard our customers’ data is vital to our business,” explains Anderson. “With MapR we just take a snapshot and mirror it to another data center. MapR gives us fine-grained control of the movement of the backup data, which allows us to comply with regulations that restrict the movement of certain kinds of information out of the European Union. You don’t get that with other Hadoop distributions."

Liaison is optimistic about its prospects in the data management market. “We have a superior architecture, which makes it difficult for our competitors to keep up feature- and functionwise,” says Anderson. “UCS gives us plenty of computing power, plus the ability to tweak configurations for each customer use case. MapR lets us customize solutions by plugging in whatever analysis tools are needed. I can’t overstate the importance of our enterprise-grade platform as we enter this new and challenging marketplace."

More Information



DOWNLOAD PDF
AttachmentSize
mapr-liaison-casestudy-072215.pdf2.19 MB