Based on the fundamental premise that data, information andknowledge are three intricate, indispensable and interdependentelements, hyperscale systems — distributed computingenvironments that efficiently scale from a few servers to thousandsof servers — are being built to unearth untapped businessvalue with unprecedented processing power, speed and accuracy.

CCC InformationServices is at the center of this innovation withinthe insurance industry. We work to help customers convert data fromdiscrete facts into meaningful structures, and ultimately, intodomain knowledge. Enterprises seeking to capture a strategicadvantage and growth will need to harness the value buried deepinside this domain knowledge, which includes generating metrics,categorizing results, assessing values, making forecasts andpredictions, and enabling informed decisions.

The first thing to understand is that today's data is different.We think about it in the context of 3 Vs:

• |
  • |
    • |
      • |
        • |
          • Volume
          • Variety, and
          • Velocity.  

The vast amount of dataset (volume) that are generatednow are distributed, diverse, disparate and heterogeneous innature. They are structured, unstructured and semi- structured(variety). And, these data set are growing at exponentialpace (velocity). For example, a single blade in a GE gasor wind turbine, generates 500 gigabytes of data each day. Thatmeans in 30 days a single blade would generate as much content asthe print collection of the Library of Congress. For the record,there are about 4,500 GE gas turbines around the world, each withdozens of blades. There are examples like this everywhere,including within the P&C insurance industry.

The 3 Vs of the modern data characteristics generate enormouschallenges in the areas of storage, retrieval, security, sharing,analysis and reporting. On the flip side, they also open massivebusiness opportunities throughout different phases of datalifecycle. Besides the 3Vs nature of the data we also recognize thedensity of information embedded in them. That is where trueenterprise value lies. Without it any enterprise will be flyingblind.

The new paradigm: Data is liquid, and it's everywhere

The nature of today's hyperscale data has striking similaritieswith water: precious, abundant resources that are essential tosurvival. Hyperscale data, like water, is also storable,transportable, transformable, vulnerable and susceptible tocontamination when being processed for consumption. And much thesame way that massive resources and infrastructures are built toharness the power of water — dams/reservoirs, electricpower stations, security and backup systems, purification andfiltration systems, distribution channels, maintenance,monitoring systems, billing/accounting/monetization systems— we believe that hyperscale data needs massive storagestructures, security systems, high speed networks, processingmachines, monitoring and regulatory structure to extract andharness the vast power embedded in them.

The CCC hyperscale platform

With so much excitement about the potential in the area ofhyperscale data, predictive analytics, and deep learning andmultilayer neural networks, the insurance industry in many ways istaking the lead in getting real business value from the technologybuzz. We are immensely proud to be part of this new technology wavewith our partners. We understand that as the hyperscale databusiness evolves, so does the technology stack, and enterpriseapplication architectures are evolving from integration-centricenterprise service bus (ESB) architectures to application-centric,micro-services, platform-as-a-service (PaaS), multi-cloud, andAPI-driven architectures.

We understand APIs are the lynchpin to the success for many ofour business and technology partners: these APIs are used byapplications to access key services and data. These services can bemicro-services, cloud workloads, legacy SOAP services, or theInternet of Things (IoT). Using the CCC API foundation layer, wereinforce our technology partnership to create rich insights intohow safe drivers are using auto physical damage and injury data,telematics, big data, and analytics. We recognize APIs and APImanagement are essential to scale quickly and can supportintegration with partner companies' complex infrastructure. Inorder to process the hyper density data we are building ahyperscale processing platform.

Dissecting the next generation hyperscale platorm

CCC's hyperscale platform has several important foundationalcomponents. They include: 

Data ingress/egress layer. OurIngress/Egress layer has two separate yet equally important goals: Interactivity (real-timequeries) and data freshness (real-time ingestion).

This layer has been built to help our clients make sense oflarge amounts of data in real-time. Every day the platform ingeststens of millions of new events, and currently handles severalhundred million aggregated events. The platform has been designedto also support the ability to automatically and efficiently movelarge amounts of data in and out of platform; and to parallelizethe execution, across multiple execution nodes in the applicationframework layer.

Data as a Service (DAAS) layer. OurDAAS layer is built on the fundamental guiding principle thatunderlying raw business data will be very fluid— structured, unstructured, and semi-structured— and stored in diverse data storage types at a it's avolume with an exponential pace. Treating DAAS and not a dependencylayer, large scale applications can be built to be data storage,structure and location agnostic. By insulating the applicationsfrom the above three concerns that hamstrung applicationportability and scaling for several decades, DAAS enables thebusiness logic to become highly portable across diverseapplications platforms. Being location-agnostic enablesapplications to be highly available across different geographicregions as DAAS can be made available 24×7. Our DAAS layer has beenbuilt to address concerns on the both application level featuresand operational area, including application latency; linearscalability; high availability; and operational economics.

Application framework layer. Thehyperscale application layer has been designed to support an arrayof diverse applications that can access data stored in the DAASlayer in a storage-, structure-, and location-agnostic way. Itsupports high velocity data streaming apps, high volumetransactional and batch type apps. Petabytes scale businessintelligence and predictive analytical apps, and multilayer neuralnetwork — (deep learning apps).

Hyperscale data security. Lastly, withbig data comes big responsibility; its security. A multilayereddata security strategy — prevention, detection andpolicy/administration — can be both efficient andeffective at maximizing security controls.

Preventive controls stop intruders from gaining unauthorizedaccess to data. Detective controls centralize auditing andreporting across the organization so that either security breach orcompromised system can be swiftly detected and necessary actions betaken. And finally, policies/administrative controls can preventunlimited and ad-hoc access to application data while stillallowing legitimate administrative activity.

CCC's hyperscale data security layer has been implemented as athree-pronged approach: two level identification data encryption,and data de-identification.

The road ahead

Hyperscale data is still in its infancy and evolving. Itspotential is immense, but so are its challenges. We at CCC areworking on next generation initiatives to address the hugechallenges our partners face and deliver solutions that are uniqueto their business.

Dr. Reza Rooholamini is vice president of Architecture andTechnology at CCC Information Services Inc. For questions, or tolearn more, send email to [email protected].