Rerouting Grid Applications Across Firewalls

نویسندگان

  • J. Tan
  • D. Abramson
  • C. Enticott
چکیده

The Grid refers to “a distributed computing infrastructure for advanced science and engineering” [4]. Grid applications require advanced computations through distributed processing, with an explicit focus on supporting virtual organizations (VOs). By participating in a VO, participating institutions are able to share and coordinate computations, commonly with the Internet as the underlying communication infrastructure. Unfortunately, on the Internet, there is always the risk of unauthorized access or even sabotage to data, transactions or computational resources. Thus, networks on the Internet are usually protected with firewalls and other measures. This presents a dilemma for VO setups since this reduces the level of collaboration that is possible. Furthermore, current Grid applications often assume that network connectivity is mostly unrestricted. Accordingly, a number of virtual organizations have failed because the member sites have security policies that clash, encountering more restrictions than anticipated in the design of their applications. Moreover, since the participants in a VO are autonomous to one another, their network security measures are likewise quite varied. These issues may be taken together as a security interoperability problem [3]. Somewhere in the interconnected networks involved, a mismatch occurs between what is allowed on one end and what is not allowed on the other. We studied the likely scenarios resulting from security interoperability and developed methods that facilitate connectivity for grid application components despite such restrictions. We refer to our project as Remus, in which we are building a rerouting and multiplexing system that enables intra-firewall connectivity while minimizing the need to modify or rebuild either the grid applications or the firewalls. Our solutions are built on a number of standard secure communication protocols such as SSH and SOCKS. We present our experiments here and the results of tests using Globus and the Nimrod/G middleware. We considered three types of conflicts that can hinder communications between two grid application components. A node conflict occurs when security restrictions block connections to the target node at any port. A port conflict occurs when security restrictions block connections to one or more target ports only. We may also find that, while a port conflict exists, there is instead another vacant port available. In such situation, the connection still cannot be made to the intended component until that vacant port is bound to some mechanism that will facilitate communication. We can also encounter a protocol conflict, when a successful connection can be made to the node at a given port but the service listening behind that port is unable to communicate using the protocol expected by the connecting component. We developed solutions that can handle each of those types of conflicts using various tested means through tunneling, port forwarding and communicating by proxy. Remus is also designed to accommodate, as much as possible, the network connections exactly as expected by the grid application components. Furthermore, it is recommended that rerouters use deliberately only those channels that are authorized by existing security policies. We tested the applicability of Remus to Globus [5][6] and Nimrod [1]. Both are middleware that sit above and harness the computational power of several loosely-coupled processors. While Globus provides a generic facility for resource sharing and distributed computations, Nimrod specifically uses distributed resources for the computations of parametric applications [1]. Nimrod/G [2] is the Grid-enabled version of Nimrod that uses various gridmiddleware services that support various standards, such as Globus [5]. Nimrod/G computations are not the same as Globus computations because Nimrod/G uses agents. When Nimrod/G uses remote resources, agents execute in them, allowing Nimrod/G greater flexibility in job management. Nimrod agents coordinate with Nimrod/G server components in order to perform data transfers to download input data or to upload output data. Nimrod agents may J. Tan, D. Abramson and C. Enticott

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

منابع مشابه

The WISENT Grid Architecture: Coping with Firewalls and NAT

In energy meteorology research, scientists from several domains such as physics, meteorology and electrical engineering work together to obtain information needed to characterize energy production from regenerative energy sources such as wind and solar power. For this purpose, several scientific applications were developed to process large data sets from heterogenous data sources in complex and...

متن کامل

Globus Grid and Firewalls: Issues and Solutions in a Utility Data Center Environment

The “Grid” has become the next incarnation of the utility computing vision: seamless access to data and services anytime, anywhere. Sharing a utility infrastructure requires high standards of security. Grid platforms such as the Globus Toolkit, the Grid Engine, Platform or the forthcoming Open Grid Services Architecture (OGSA) include various security mechanisms from encrypted communication, us...

متن کامل

Online Steering of HEP Grid Applications

Online steering and visualization of scientific applications is a wellestablished method for accelerating research and saving resources. However, for Grid environments no appropriate, secure online steering tools exist. As a part of the HEPCG (High Energy Physics Community Grid) project we are developing the online steering system RMOST for Grid applications, which is specifically targeted towa...

متن کامل

Recovering Internet Symmetry in Distributed Computing

This paper describes two systems to recover the Internet connectivity impaired by private networks and firewalls. These devices cause asymmetry in the Internet, making peer-to-peer computing difficult or even impossible. The Condor system is one of those that are severely impaired by the asymmetry. Compared to normal peer-to-peer computing applications, Condor has stricter requirements, which a...

متن کامل

An energy-aware grid-based routing scheme for wireless sensor networks

As an important field of emerging technology, wireless sensor networks (WSN) offer many new possibilities for applications such as target tracking and environmental surveillance by allowing the observer to move around freely. However, disseminating sensing data to the mobile observer raises significant design challenges for the routing scheme. In addition, WSN often operate under certain energy...

متن کامل

ذخیره در منابع من


  با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید

برای دانلود متن کامل این مقاله و بیش از 32 میلیون مقاله دیگر ابتدا ثبت نام کنید

ثبت نام

اگر عضو سایت هستید لطفا وارد حساب کاربری خود شوید

عنوان ژورنال:

دوره   شماره 

صفحات  -

تاریخ انتشار 2006