Could NoSQL Replace Relational Databases in FMIS?

Introduction

The declining availability of arable land and the concern about world food production for the next years(Godfray et al., 2010) boosted the pursuit by ways to enhance the agricultural yield. Due to availability of improved electronic devices equipped with accurate Global Positioning System(GPS) at reasonable costs(Fountas et al., 2015), Information Communication Technology(ICT) devices were chosen by researchers and companies as the path to achieve this goal. As a consequence, a myriad of sensors and devices were developed(Aqeel-ur-Rehman, Abbasi, Islam, & Shaikh, 2014). They produce or collect an immense quantity of unstructured geo-referenced digital data about soil, plants, weed, pests, weather, equipment and machinery, management practices and others(Peets, Mouazen, Blackburn, Kuang, & Wiebensohn, 2012; Porter et al., 2014; Steinberger, Rothmund, & Auernhammer, 2009). These data after merging and processing offer valuable information to take accurate decisions that improve income, management practices, products quality, environmental protection and comply with governmental regulations(Fountas et al., 2015; Kaloxylos et al., 2012). However, processing an enormous amount of data to get information is not a trivial task and a computational tool is used to this end.

The computational tool responsible to transport electronically, store and process data in order to get useful information that serves to support planning, execution and evaluation of farm or field operations is referred to as Farm Management Information System (FMIS)(Kaloxylos et al., 2012; Nikkilä, Seilonen, & Koskinen, 2010). Higher will be the accuracy of information produced when more data is available, for this reason all data gathered from diversified sources, including third-parties, is stored historically and handled in FMIS(Nikkilä et al., 2010; Steinberger et al., 2009). Web Services(Nash, Korduan, & Bill, 2009) are the most opportune way for third-parties to access the data, however distinct sources dispatch and consume data in diversified storage and implementation formats such as spreadsheets, images, structured XML, plain text, etc., some of them exclusive of manufacturers(Porter et al., 2014). It represents a challenge for the FMIS, which needs to be able to process all this data in order to display helpful information to the decision maker.

Establish a data format standard and suggest a software architecture capable of accepting the variety of formats were the solutions found to overcome that challenge. Two well established data standard are AgroXML (Schmitz, Martini, Kunisch, & Mösinger, 2009), a German standard for data exchange between third-party farming systems, and ISOBUS (Oksanen, Öhman, Miettinen, & Visala, 2005), an international standard for communication between machinery and agricultural implements. Although they could facilitate data exchange and technology adoption (Aubert, Schroeder, & Grimaudo, 2012), other formats still exist. For this reason some authors proposed different solutions to convert data and adjust it to a specific format(Iftikhar & Pedersen, 2011; Porter et al., 2014). This favored the adoption of Service Oriented Architecture (SOA) in FMIS in order to adjust automatically the data. Some works dealing with this issue are (Fountas et al., 2015; Kaloxylos et al., 2012; Nikkilä et al., 2010; Wolfert, Goense, & Sorensen, 2014). The FMIS with SOA has a layer responsible for adjusting data according to the desired format, easing the working with on-the-go sensors(Peets et al., 2012) and automatic machinery(Blank, Bartolein, Meyer, Ostermeier, & Rostanin, 2013) from distinct manufacturers, in addition to third-party data sharing (Wolfert et al., 2014).