Vision and how it works

Vision

CSE-Online vision is to make all computing resources available to the users at their fingertips in a transparent manner.

We define the computing resources available to the user consisting of all resources (data, application tools, and computing cycles) available to the user including those on the user's desktop. For data, this includes private data on his desktop computer and on all remote servers for which the user has login accounts, public data at public databases available online, and possibly shared data which are made available from others. For application tools, this includes the user own tools, CSE-Online tools, other third-party open-source or commercialized tools. For computing resources, this includes the user desktop CPU, remote servers for which the user has CPU allocations, and the computing grid such as the TeraGrid or other grid computing resources.

How CSE-Online works

CSE-Online is a virtual desktop environment that allows data, applications, and computing resources to be delivered to the user desktop transparently. It is a client-rich software framework.

How CSE-Online accesses application tools

Upon creating a workspace for the first time, the user fills up her CSE-Online desktop with a number of icons of Java applications she has chosen from the catalog.

These icons contain the knowledge about the application, namely the location of the executable (jar file), how to initiate it, its version, etc. Double click on the icon of the application. This initiates a sequence of processes, particularly:

1. Is there a version of the application on the local cache? If not, download the executable from its server and run it.
2. If yes, then check if the version on the local computer is current. If no, update the local one by downloading the new version then run it.
3. If the local version is up to date, then run it.

This way, the user always has the latest published version of the application and the developers do not have to worry about sending out bug fixed versions. This gives CSE-Online its online and on demand characteristics.

The CSE-Online software framework allows the applications to be independent from each others. Thus, they can be developed independently. They can communicate with each other via open application programming interfaces. It allows the users to bring their own tools as well as the third-party tools into the desktop environment and having these applications to communicate with other tools in the environment. This makes CSE-Online extendable and their applications are interoperable and modular.

The above is true for Java applications. What about server-based applications? In fact, 99% of software for computational science are server-based due to their computing intensive nature and many are commercialized and thus have license restrictions. How does CSE-Online fulfill such license restrictions?

CSE-Online currently supports a number of server-based applications such as Gaussian 03, GAMESS, AMBER, TheRate, Thermo, etc. These applications have different license restrictions. To access these tools, CSE-Online developed Java Graphic User Interfaces (GUI's) to create input files for these applications. Depending on the license restrictions, CSE-Online will invoke different mechanisms for initiating the server-based applications with the input files from the GUI's. For instance, TheRate is CSE-Online server-based tool for calculating rate constants of elementary reactions. It does not require significant computing resources and thus we provide it as a free service to the community. In this case, CSE-Online allows the user to request service, i.e. running TheRate on the CSE-Online server, using Web-Services technology. The users do not have to install TheRate anywhere. On the other hand, Gaussian 03 is a commercialized software for quantum chemistry which has site licence restrictions. The Gaussian 03 GUI creates the input file then sends it to the Resource Broker tool for submission. The users must configure the Resource Broker for the location of the Gaussian 03 executable on the server that they plan to run the job and the Resource Broker will submit the job if the users are authorized to use the server. This way CSE-Online can fullfill license restrictions of any server-based application.

Users can also develop Java GUI's for their own tools for integrating into CSE-Online to take advantage of the Resource Broker for submitting and monitoring jobs on their servers.

How CSE-Online accesses data

There are two kinds of data depending on their sources, namely files from file servers and records from databases.

For files, CSE-Online has an application called Global File Explorer (GFE), that allows the users to access files from their local desktops as well as all remote accessible file servers. Users can copy/move files between servers by drag-and-drop. GFE simplifies the task of managing files on multiple servers. GFE also allows users to share files with others (not using the same servers) easily. This share feature is currently being implemented.

For a record from databases, similar to GFE, CSE-Online provides a database client (DB Client) enviroment for accessing databases. DB Client distinguishes three kinds of databases depending on their accessibility, namely public, private, and shared databases.

For public databases, there are two types, those hosted by CSE-Online and those hosted by third-pary providers. CSE-Online hosts a number of public databases which are available to all users such as the Protein Data Bank, Mechanisms DB, and Flame DB. It uses EJB to connect to these databases. For public databases provided by third-party providers , it uses http request providing such databases allow such connections such as the National Cancer Institute small molecule and PubChem databases. Thus, DB Client allows the users to access these databases and retrieve records to use in other tools, such as visualization for further modeling, without having to go to their web-sites and to manually download the records.

For private databases, CSE-Online currently has the Quantum Chemistry DB Management System (Quantum Chem DBMS) that allows the users to manage results from quantum chemistry calculations. The users can install (or having a System Administrative person to install) a Postgres or MySQL database system on their servers and create database accounts for them. The DB Client will allow the users to create private Quantum Chem databases on these servers for storage/querry/retrieval of quantum chemistry results. For more information on creating private databases, click Here.

For shared databases, the DB Client allows the users to share their private databases with other CSE-Online users (not neccesary having access to their servers) by copying these databases to the collaborative space on the CSE-Online server. The DB owner can set different levels of access, namely (Read only) or (Read and Write) for those he/she decides to share the databases with.

How CSE-Online accesses computing resources

CSE-Online distinguished three types of computing resources that the users have, namely those from their local desktop computers, remote servers where they have allocations, and from the computing grids such as the TeraGrid.

Since CSE-Online is Java-based, it is running on the local user CPU. In this case, graphic application tools in CSE-Online such as MolBuilder, BioViewer, or PsiViewer are using the local graphic cards on the user desktop computers to achieve their performance. The virtual desktop continues to operate even in the event of Internet disruption. In fact, CSE-Online provides the users the offline mode where all local resources are accessible by Java applications in the CSE-Online environment.

For remote servers, both the GFE and the Resource Broker provide means for accessing files and application tools there.

For a computing grid such as the TeraGrid? , CSE-Online provides a Globus submission tool. This is a client interface using the Globus toolkit to submit jobs to any Globus enabled grid. Users must have a proper Certificate of Authority (CA) on such a grid in order to use the Globus submission tool. It is different from many science gateways which have community accounts and submit jobs for the users upon request from their portals. The Globus submission tool provides a direct means for users to submit jobs directly to the computing grid , not via any portal, and using their own CA's.

Work Anywhere At Anytime (WAAA)

As mentioned above, CSE-Online provides both online and offline modes so the users can use the environment with their accessible resources at any time. In addition, CSE-Online allows the users to export their current workspace configurations and preferences of all applications to an encrypted file which can be stored on a moble storage device such as an USB drive. With this portible user workspace configuration, the users can activate the same working environment on any computer. This gives CSE-Online the WAAA characteristics.

Below is the software architecture of CSE-Online, it provides a summary of the above discussion and illustrates how CSE-Online works.