Large Data architecture

&http://www.aliyun.com/zixun/aggregation/37954.html ">nbsp; Some data architects have long since begun to deal with the huge amount of data generated by the collaboration between IoT (IoT) and the private network of smart devices. Integration Expert Brian Anderson is the general manager of software and process development company Notionovus. He worked as an embedded system programmer for the Caterpillar company early in his career. At the time, he was primarily responsible for building the application system for the robot Vehicle network in the workshop.

During his 25 years as a Caterpillar company, Anderson has also served as Engineering Director, manufacturing engineer and 6 Sigma Black belt. In other words, before IoT, he began to deal with the data problem.

Potential at Work consulted Anderson about how the architect would prepare the existing environment for IoT inherent mass expansion and massive integration.

What are your suggestions for architects who are trying to lay the groundwork for a structured approach to IoT data integration?
Anderson: Integration professionals need to focus on four key process metrics (KPIs): speed, reliability, cost, and security. These four customer-facing metrics always recur when people talk about good or bad integration experiences. However, when discussing IoT, the key indexes are mainly on the security. The device belongs to the owner, and the owner is bound to have privacy concerns.

The winners of the IoT competition arena will be those that provide the best user-friendly data controls, and can tightly lock the audience into viewing content. Losers will be businesses that secretly use the data generated by their devices to pry into the privacy of their users. In addition, look for tools that can reuse and leverage existing code libraries. This will reduce the demand for highly specialized personnel.

The architect's job involves more global thinking and anticipation of future integration requirements so that flexibility can be placed in its planning model. For example, if data architects and data integration professionals are busy dealing with a problem on a particular interface, you are wasting money rather than planning for the future.

How do you apply the Lean method to unstructured IoT?
Anderson: Stick with tools that help build transparent integration. An end-to-end visibility into the value chain of information is critical to measuring KPIs. If we are committed to continuous improvement, we cannot use the black box system. Avoid developing redundant application systems and recreate established connection points.

Patches, updates, and bug fixes are not customer-specific and have added value, so focus on minimizing the complexity of the release code. Because of the high cost of bad data, placing data validation in software can improve value. Validation is the only non-value-added option when the system and its users generate no error data.

How can data architects adapt the enterprise data management strategy to IoT?
Anderson: Enterprise data management policies need to regain access to archiving, security, and use cases for future and existing data streams. All the increased data flows on the IoT will have a significant impact on the current operation. You need archiving and data retention policies in place to maximize the use of massive amounts of data flow while minimizing the footprint of your data management system.