Interfacing humans and e-infrastructures
The e-infrastructure is in place and researchers can now easily make use of the technology that underlies this complex and heterogeneous ecosystem. They will do this to address novel scientific theories and create more knowledge. But, how to access the resources in a user friendly, seamless and transparent way? I will briefly unveil a few aspects on how I think scientists can interact with the “cyber-jungle” in an increasingly efficient way.
By definition, an interface is a point where two systems, subjects, organizations, etc. meet and interact. Humans interact with computers through an interface, often called the user interface (UI). This space of interaction allows for an effective operation and control of the machine. In the same way, in order to interact with an e-infrastructure, such as the European Grid Infrastructure (EGI), users have at their disposal an interface. The technical expertise of the user, as well as which resources are being targeted, will ultimately dictate which UI to use.
There are a myriad of UIs out there, each one with its own features for accomplishing a certain task. In the end, a UI is (or should be) designed to be a transparent tool for the operator to attain a targeted productivity. Nevertheless, I perceive that typically the learning curve for a user to use command-line interfaces (CLI), which require commands to be typed on the computer keyboard, is much steeper than using “Windows, Icons, Menus and Pointing devices” (WIMPs), where a series of clicks on specific positions on the screen will trigger an action on the system. Graphical User Interfaces (GUIs) are mostly WIMP based, and are the preferred way for a novice user to interact. EGI has both GUI and CLI interfaces and they are the doorways to access the e-infrastructure. You can read a bit more about them in the EGI wiki.
Grid and distributed computing infrastructures are in constant development, though stability and standardization rules are being implemented and a plateau on the evolution trend will undoubtedly occur. This brings a high penalty factor for application developers. The effort time consuming task of adapting an application to an inherently heterogeneous environment, both in terms of physical/logical resources as well as being distributed across multi-domains (academia, private companies, countries, continents ...), should be avoided, or at least minimized.
An easy way for application developers to connect community and user-specific applications to the services of the European Grid Infrastructure is through the usage/re-usage of application programming interfaces (APIs). The purpose of these APIs is, similar to the way the UI facilitates interaction between the user and the e-infrastructure, to allow communication between the consumer program and the services provided by EGI and other distributed computing infrastructures. Nevertheless, the huge number of APIs an application has to use to interact with such heterogeneous infrastructure(s) can be cumbersome and thus becomes prohibitive to implement. A current list of EGI service APIs is published through the EGI wiki.
Simple API for Grid Applications (SAGA) represents an international effort to get a consensus on the interface application, developers could use to interact with heterogeneous ecosystems such as EGI. SAGA created an abstraction layer which has the underlying platform-agnostic mechanisms and adaptors to make the appropriate calls in an intelligent fashion. This layer enables the coupling of high end machines and puts them available in a seamless way for a scientific application, without having to change the application code. The API is in the process of being incorporated into the Unified Middleware Distribution - this in turn will provide EGI communities, a robust and sustainable tool to interact with the e-infrastructure services.
Science gateways are an emerging interface for communities to engage more actively with the e-infrastructure. Again, the rule of thumb is to lower the barrier that can exist between users and the evolving grid world. The success of bringing new communities, and sustaining current ones, into this “cyber-jungle” relies heavily on the user-friendliness and robustness of the available interfaces. Otherwise, it will be mind-boggling for researchers and their lifelong commitment of producing knowledge will be lost.
