Subject: Re: C432 JGFSI Reviews (NEW) Resent-Date: Fri, 12 Nov 1999 09:20:49 -0500 Resent-From: Geoffrey Fox Resent-To: p_gcf@npac.syr.edu Date: Tue, 9 Nov 1999 14:12:00 -0500 (EST) From: Cheng Zhong Xu To: gcf@npac.syr.edu Dear Prof. Fox, (Please abandom the last email if you receive more than one today. I forgot to include the revised paper in the last email. ) Following is a summary of our responses to the review comments. Most of the comments are quite valuable and constructive and are considered in the revision. Referee 1 said: >>1 "The true benefits of Traveler are not clear--if you are using >>1 shared memory idea, why is this more interesting than other >>1 approaches such as JINI" In the revision, we highlighted, as suggested by referee 2, that Traveler provides clients with an agent wrapper to simplify agent-oriented programming. Its resource brokerage mechanism provides more than JINI resource location services. >>7 "Section 3.1 mentions that 'Brokers are organized in a hierarchical way', >>7 but it does not mention how this is achieved. Agent naming is a big >>7 problem, and the paper does not clarify how this is achieved in this >>7 particular system." It is true that we omit the details of the hierarchical broker system. Since the focus of current work is on mechanism of mobile agents for global computing, rather than resouces scheduling, the Traveler assumes a single resource broker. Agent naming on a single-broker system is by no means an issue. Issues on hierarchical broker systems are going to be addressed in the future extension work. >>11 "On page 10, the migration of an AgentTask is identified, but what >>11 happens to the data associated with this operation? Data migration is >>11 not mentioned or clearly identified" AgentTask is run under the support of DSA threads and AgentTask moves together with DSA threads (adaptivity of virtual machine). Distributed shared array will be migrated with the DSA threads. Referee 2 said: >> "... the statement regarding the dependence on lifetime reliable network >> connections isn't true for applications distributed across different >> administrative domains. There are numerous projects that have used global >> computing architectures, such as DSP for SETI and DES cracking where work >> is parceled out and then connectivity isn't an issue for long periods >> of time. " While some of existing global computing architectures like Web and DES crakcing are state-less, most of traditional distributed applications are connection/state-oriented. As discussed in Related work section, the existing state-less global computing infrastructures are "pull" based. The innovality of the paper is in the first employment of mobile agents for "push" based global computing. Referee 3 commented: We appreciate very much the general comments by the referee. Some of the issues he/she pointed out, like resource supply/demand trading and the impact of mobile agent technologies on intranet computing, will be studied in the future. -------------------------------------------------------------- | Cheng-Zhong Xu, Assistant Professor | | | | Wayne State University Phone: (313) 577-3856 | | Dept. of Electr. & Computer Engg. Fax: (313) 577-1101 | | Detroit, MI 48202 czxu@ece.eng.wayne.edu | | USA http://www.pdcl.eng.wayne.edu/~czxu | -------------------------------------------------------------- --------------------------------------------------------------------- Name: CPE-Final.pdf CPE-Final.pdf Type: Acrobat (application/pdf) Encoding: BASE64 Description: CPE-Final.pdf Name: CPE-Final.doc CPE-Final.doc Type: Winword File (application/msword) Encoding: BASE64 Description: CPE-Final.doc