Large Scale Processing and Searching of Engineering Design Content Using Open Source Technologies with Mike Haley.
A common issue faced by the over 8 million users of Autodesk's design tools is finding good reusable design content. As CAD has evolved over the last 15 years, it has transitioned from being about lines, arcs and circles to being a more object-oriented process. By this we mean that the designers building a model generally construct the model from already available parts.
Typically the designs for these parts may either be generic placeholders for the real world part or they may actually be manufacturer specific designs. The latter is becoming a lot more common today as architects and engineers strive for greater efficiencies. Given the enormous variations on all parts representing real world objects, the process of finding the correct design file for a part is becoming an enormous burden for the majority of our users.
The Autodesk Content Search solution is a new product targeted at solving this problem. Content Search is a vertical search engine with parametric and ontology aware filtering capabilities for rapidly finding appropriate part information and design content. In building this solution we were faced with numerous challenges ranging from building the core parametric search technology to processing and analyzing large quantities of metadata and CAD content in a scalable fashion.
The development team managed to turn around a first version of this product in only 4 months largely due to using open-source technologies such as: Lucene, Hadoop, DWR, and Active-MQ. We are now beginning to make extensive use of the Amazon compute-cloud and storage services for scaling out the storage and processing capabilities.
In this talk I will present the architecture of this solution, describe many of the key choices we made with respect to OSS technology and finish with a brief demonstration of the system- Association for Computing Machinery (ACM)
Mike Haley directs the engineering of Content Search and distributed systems technology at Autodesk. He leads an engineering team focused on scalable computing solutions, search and indexing techniques, machine learning and contemporary web technologies all focused at producing an effective solution for Autodesk customers to find and use design information online.
Formerly he was the architect of the Buzzsaw Collaborative Project Management system at Autodesk now used by over 150000 people to manage construction projects worldwide. Prior to joining Autodesk he has worked in the fields of multimedia networking, medical computer graphics, embedded systems and large scale telephony systems.
He has over 10 years of experience in server technology and an extensive background covering 20 years of work in software design, computer graphics, networking and real-time computing.
Mike holds an MS in Computer Science with distinction from the University of Cape Town, South Africa.