Skip to Content

Virtualization, Thin Clients and Cloud Computing: Potential uses in LTER

Printer-friendly versionPrinter-friendly version
Issue: 
Fall 2010

Virtualization, Thin Clients and Cloud Computing: Potential uses in LTER

John Porter (VCR)

Times are changing – again! The Long-Term Ecological Research (LTER) Network has weathered several major changes in how digital information is managed. The start of LTER in the early 1980s coincided with the start of the movement from mainframe computers to desktop personal computers. Nonetheless, mainframe computers and even punch cards were the staple of early LTER computer use. With the advent of the IBM PC in 1983 and subsequent improvements, the stand-alone PC, with floppy disks, small hard drives and even (for retrieving data from the field) cassette tapes) were increasingly useful for augmenting or replacing mainframe computers, which were themselves being replaced by Unix workstations. With the 1990s came the ability to interconnect the PCs and services such as electronic mail gained currency within the network. Thanks to efforts of then LTERNET Information Manager Rudolf Nottrott, LTERNET was one of the first places where email could move across the proprietary systems used by many federal agencies and the larger research community. The network infrastructure went through an order of magnitude (or more) increase in importance with the advent of the first network browsers in 1993, and the burgeoning World-Wide Web. The mid-1990s saw the introduction of interfaces and tools that let relational databases be directly linked to web sites, leading to a huge increase in the capabilities of web based systems for managing information. By 2000 you started to see Content Management Systems (CMS) used to automate the nuts-and-bolts of web site development. However, we are about to go through another major change driven by increasing virtualization and advances in cloud computing.

I’ll start off talking about some specific changes I’ve seen at the University of Virginia (UVa) and in collaborative projects. First, UVa is in the process of shutting down all of its public computing laboratories. These labs consisted of PCs and Macs loaded with specific site licensed software required for use in classes. The rationale for having labs was that it would be impractical and time consuming to install, configure and test all the needed software on each individual student’s PC or Mac. With the labs gone, how will this continuing need be met? The answer is that they are replacing all the labs with the “UVa Hive” – a system that uses remote-desktop software to link to virtual machines running on a large computing cluster. To use the Hive, one connects to a secure web site in a browser, logs in and connects to a virtual machine using remote desktop viewer. The effect is that one sees a PC booting up (alas, currently running Windows Vista, but likely to be running something better soon), that has a “start menu” pre-loaded with all sorts of useful software (e.g., ArcGIS, statistical packages, Matlab). It is as if you used Remote Desktop on your laptop to connect to your office PC, except with the Hive you are connecting to a PC that exists only as a set of files on a VMWare server. Additionally, all the drives on your local PC are automatically mapped to network drives on the virtual machine, so that data can be easily accessed. The latter is important because when you logout from the virtual machine, it is erased, taking any local files with it.

What are the potential implications of this technology for LTER? Such a technology could be used to make available to LTER researchers a standard suite of analytical tools that are available in a few moments from any network-connected PC. Imagine a system where a single login lets you select a system pre-loaded with LTER-specific tools such as the Morpho metadata editor, the Kepler scientific workflow system, ready-to-go ODBC drivers for connecting applications like Excel to databases, form-based editors for adding data to databases or tools for generating data reports. The system would be available in the matter of about one minute (essentially the boot time of the virtual machine) to any LTER researcher, whether they were at their home computer or not. The computational load placed on the local computer is minimal, because the local computer (be it PC, Mac or Linux-based) is just running a remote desktop viewer. The speed of the resulting analyses depend on the speed of the server running the virtual machine, not the power of the local machine, so that “thin client” computers with very limited speed and memory work just as well as powerful workstations. UVa has set up the virtual machines to be transient (each new virtual machine is generated from a set of master files for a particular type of server, so changes during a session not saved to a network drive are lost), but it is also possible to configure the machines to be persistent, or even to share a single machine among multiple users simultaneously. At the VCR/LTER we currently use the latter model for sharing a virtual machine running software for downloading our wirelessly-connected data loggers. A status display is available to LTER staff both at the field station and at UVa through a remote desktop.

A second tool at UVa that we have been using a frequently is the “Collab” system – a SAKAI-based content management system. SAKAI is a content management system with some similarities to Drupal, Xoops and others. Where it is different is that, whereas most content management systems focus on users being able to add content to a web site, Collab focuses on making it easy for users to generate new web sites. Using web form based-interfaces, a completely new collaborative web site that includes authentication, a place to share files, a shared email address and archive, discussion groups, and calendars can be set up in less than 5 minutes. Once the site is set up, site members can populate it with content, just as if it were a conventional CMS-based web site. Using this tool, individual researchers, including graduate students, can easily set up web sites dedicated to specific projects, however large or small. For example I currently am owner of 20 collab web sites including ones for LTER data submissions, wireless network and data logger configurations, VCR/LTER Graduate Students and many more. Additionally, I am a member of 17 other web sites set up by others, dealing with specific topics or projects.

For LTER the implications of this technology, and related advances in other collaborative tools, are that it simplifies the routine mechanics of sharing data, email, figures and documents among a team of researchers. But what distinguishes it from past technologies is the ease with which such an environment can be created, even by relatively naive users.

Finally, there are all those tools provided by Google and others that are making increasingly sophisticated use of AJAX and other technologies that convert a web browser into a highly capable software platform for word processing, spreadsheets and an increasingly wide array of functions. These browser-based tools can serve as clients for web services or for more specialized and proprietary servers. Google Docs has been heavily used during the drafting stage of several collaborative manuscripts I’ve been involved in. Its ability to maintain a network accessible “master” document – while also tracking changes and identifying their authors has been a huge boon in producing collaborative manuscripts.

Google Docs is a primary example of “Cloud Computing” where most of the actual work is done by a server over the network, but the look and feel of the application is that it is a local application. The UVa Hive example is a slight variant on cloud computing where the web server is used to download a remote desktop viewer that subsequently runs independently of the web browser. All of the examples above only require “thin clients” – essentially a computer capable of running a modern web browser to operate. Just as the advent of web browsers opened new horizons for what can be used to aid LTER research, cloud computing and virtualization are again expanding what we can do. The trick will be to link the opportunities to researchers in ways that help them be more productive.