FAST: A fully asynchronous and status-tracking pattern for geoprocessing services orchestration Huayi Wu a , Lan You a,b,n , Zhipeng Gui a,c , Shuang Gao a , Zhenqiang Li a , Jingmin Yu a a The State Key Laboratory of Information Engineering in Surveying, Mapping and Remote Sensing, Wuhan University, Wuhan, China b Faculty of Computer Science and Information Engineering, Hubei University, Wuhan, China c School of Remote Sensing and Information Engineering, Wuhan University, Wuhan, China article info Article history: Received 23 January 2014 Received in revised form 30 April 2014 Accepted 10 June 2014 Available online 18 June 2014 Keywords: Geoprocessing services orchestration Scientic workow WS-BPEL GeoChaining Asynchronous Status-tracking abstract Geoprocessing service orchestration (GSO) provides a unied and exible way to implement cross- application, long-lived, and multi-step geoprocessing service workows by coordinating geoprocessing services collaboratively. Usually, geoprocessing services and geoprocessing service workows are data and/or computing intensive. The intensity feature may make the execution process of a workow time- consuming. Since it initials an execution request without blocking other interactions on the client side, an asynchronous mechanism is especially appropriate for GSO workows. Many critical problems remain to be solved in existing asynchronous patterns for GSO including difculties in improving performance, status tracking, and clarifying the workow structure. These problems are a challenge when orchestrating performance efciency, making statuses instantly available, and constructing clearly structured GSO workows. A Fully Asynchronous and Status-Tracking (FAST) pattern that adopts asynchronous interac- tions throughout the whole communication tier of a workow is proposed for GSO. The proposed FAST pattern includes a mechanism that actively pushes the latest status to clients instantly and economically. An independent proxy was designed to isolate the status tracking logic from the geoprocessing business logic, which assists the formation of a clear GSO workow structure. A workow was implemented in the FAST pattern to simulate the ooding process in the Poyang Lake region. Experimental results show that the proposed FAST pattern can efciently tackle data/computing intensive geoprocessing tasks. The performance of all collaborative partners was improved due to the asynchronous mechanism throughout communication tier. A status-tracking mechanism helps users retrieve the latest running status of a GSO workow in an efcient and instant way. The clear structure of the GSO workow lowers the barriers for geospatial domain experts and model designers to compose asynchronous GSO workows. Most importantly, it provides better support for locating and diagnosing potential exceptions. & 2014 Elsevier Ltd. All rights reserved. 1. Introduction To build large-scale and complex geospatial simulation and analysis models, scattered geoprocessing services distributed on the web are integrated into a geoprocessing services workow (Brauner et al., 2009). The emergence and spread of these geoprocessing service workows improves the interoperation and collaboration of distributed geoprocessing functions, which signicantly enhances the capacity to derive geoinformation and knowledge over a network (Zhao et al., 2012b). As a special kind of Web Services Orchestration (WSO) (Peltz, 2003) in the geospatial domain, Geoprocessing Service Orchestration (GSO) provides a unied and exible way to implement a cross-application, long- lived, and multi-step geoprocessing service workow by coordi- nating geoprocessing services collaboratively. A web service can be implemented as either synchronous or asynchronous according to type of communication mechanism deployed. Most geoprocessing algorithms however, are unsuitable for provision as synchronous web services because a geoproces- sing algorithm includes multiple processing steps; each step might involve data and/or computing intensive calculations. The data and/or computing-intensive geoprocessing algorithm usually con- sumes much time. For an instance, a GSO workow for simulating a ooding process in the Poyang Lake region may run hours or even days when dealing with ne-resolution images. It is very common for a time-consuming geoprocessing algorithm to take more time than the typical Hypertext Transfer Protocol (HTTP) transaction time-out duration. Theoretically, asynchronous mechanisms would Contents lists available at ScienceDirect journal homepage: www.elsevier.com/locate/cageo Computers & Geosciences http://dx.doi.org/10.1016/j.cageo.2014.06.005 0098-3004/& 2014 Elsevier Ltd. All rights reserved. n Corresponding author. E-mail address: youlan@whu.edu.cn (L. You). Computers & Geosciences 70 (2014) 213228