A STUDY OF THE INTEROPERABILITY BETWEEN TERAGRID AND CNGRID BY EXPERIMENTING RE

这个子项目是许多利用 由NIH/NCRR资助的中心赠款提供的资源。子项目和 研究者（PI）可能从另一个NIH来源获得了主要资金， 因此可在其他CRISP条目中表示。所列机构为 研究中心，而研究中心不一定是研究者所在的机构。 TeraGrid和CNGrid之间的互操作性研究，通过实际规模的生物信息学应用实验1. TeraGrid被称为世界上最大的网格基础设施，用于支持美国的开放式科学研究。目前部署的TeraGrid为多学科科学和教育社区提供了极大的计算和存储容量[1]。中国国家网格（CNGrid）[2]在其第一阶段（2002-2005）作为中国网格技术研究和开发的国家级试验平台，旨在成为支持第二阶段（2007-2010）开放e-Science活动的生产环境。该项目是由中国政府在高技术863计划下资助的，第一阶段奖励1300万美元，第二阶段奖励3600万美元。目前的CNGrid环境是围绕8个全国性的超级计算中心的互联而构建的，容量为20 Tflops/s，存储容量为200 TB。这两个网格测试平台的互操作有利于双边合作的科学应用，在更大规模的共享资源池，统一的访问接口不同的底层软件栈，和潜在的更好的服务质量。2.目的本项目的主要目的是研究TeraGrid和CNGrid之间的互操作性问题和解决方案，通过实验在这两个测试平台上运行的实际规模的生物信息学应用程序。从该项目中获得的经验作为CNGrid互操作活动的基础，CNGrid互操作活动是下一阶段CNGrid软件的子任务之一[3]。此外，该项目还可以为未来全球主流网格项目的互操作做出贡献，这在OGF上被称为GIN [4]。研究成果主要体现在两个层面上：a）在应用层面，我们将探索国际上典型网格应用测试平台的优势; B）在中间件层面，我们将针对公共服务互联的技术问题，包括认证和授权、数据管理、作业提交、资源发现和监控等，给出概念验证解决方案。两个生物信息学应用程序RAPTOR [5]和Treeback [6]将首先作为评估两个测试平台互操作性的基准应用程序。RAPTOR是最好的蛋白质结构预测程序之一。TreePack（以前称为SCATD）是一个基于蛋白质骨架结构树分解的侧链预测程序。它们都可以在SMP机器、集群系统和大规模并行处理平台上运行。RAPTOR将通过一个门户网站向公众开放，供生物科学家使用，该门户网站为网格测试平台提供前端服务。在获得足够的初步实验经验后，本项目的PI和Co-PI将与CNGrid应用合作伙伴一起开展大规模的生物信息学研究活动。TeraGrid采用CTSS作为中间件，由Globus Toolkit（v2和v4）、Condor和其他实用程序组成。CNGrid拥有自己的软件栈，名为GOS [3]，采用面向服务的方法，符合许多开放标准，包括WS-I基本配置文件，WS-Security和SAML。实际上，CNGrid软件将重点放在VO级管理服务上，可以将TeraGrid站点作为其VO的成员。具体而言，该项目可以分为以下任务和预期的时间表：（1）在选定的TeraGrid和CNGrid站点上部署应用程序，并调查应用程序如何在两个软件栈（CTSS和CNGrid GOS）上执行。(M1)（2）开发应用级网关（例如通过CNGrid门户或桌面应用），允许CNGrid节点访问TeraGrid生物信息资源，基于CNGrid软件客户端库，但使用TeraGrid帐户。（M2）（3）中国电网CA加入PMA等国际PKI联盟。探索TeraGrid站点上CNGrid证书的认证和授权方法。成功地将生物信息学工作提交到具有CNGrid证书的TeraGrid站点并进行监控。(M3)（4）执行步骤2，3的相反部分，以允许从TeraGrid（M4）访问CNGrid资源。（5）对跨两个测试平台运行的应用程序的性能和开销进行一系列基准测试，确定并整合国际网格测试平台的需求，优点和问题。（M5-M6）4.资源请求的合理性执行并行生物信息学实验应用程序需要5，000个SU。10 G存储空间用于应用程序和基因组数据库，10 G暂存空间用于测试数据。5.参考文献[1] TeraGrid，http://www.teragrid.org/ [2] CNGrid，http://www.cngrid.org/ [3] Xie，N Xiao，Z Xu，L Zha，W Li，H Yu，CNGrid Software 2：Service Oriented Approach to Grid Computing，the proceedings of the UK e-Science All Hands Meeting，2005 -allhands.org.uk [4] GIN，http://forge.gridforum.org/sf/go/projects.gin/wiki [5] Jinbo Xu，Ying Xu，Dongsup Kim，Ming Li.报告员：线性规划优化蛋白质线程，创刊号，生物信息学与计算生物学杂志，2003年4月[6]徐金波。通过树分解的快速蛋白质侧链包装。RECOMB 2005年。

This subproject is one of many research subprojects utilizing the resources provided by a Center grant funded by NIH/NCRR. The subproject and investigator (PI) may have received primary funding from another NIH source, and thus could be represented in other CRISP entries. The institution listed is for the Center, which is not necessarily the institution for the investigator. A Study of the Interoperability between TeraGrid and CNGrid by Experimenting Real-size Bioinformatics Applications 1. Introduction TeraGrid is known as the world's largest grid infrastructure for supporting open scientific research in the US. Current deployment of TeraGrid provides extremely large computing and storage capacity for multi-disciplinary scientific and education communities [1]. The China National Grid (CNGrid) [2], serving as a nation-scale testbed for grid technology research and development in China in its first phase(2002-2005), aims to become a production environment for supporting open e-Science activities in its second phase(2007-2010). It is sponsored by Chinese government under the Hi-tech 863 program, with an award of $13 million for first-phase and $36 million for second-phase. Current CNGrid environment has been built around the interconnection of eight national-wide supercomputing centers with a capacity of 20Tflops/s and 200TB storage. Interoperation of these two grid testbeds is beneficial to bi-lateral collaborative scientific applications, in terms of larger scale of pooled resources for sharing, uniform access interfaces for disparate underlying software stacks, and potential better quality of service. 2. Objective The main purpose of this project is to study interoperability issues and solutions between TeraGrid and CNGrid by experimenting real-size bioinformatics applications that run cross these two testbeds. Experiences gained from this project serve as a basis for CNGrid Interoperability Activity, one of sub tasks of next-phase CNGrid software [3]. Furthermore, this project could also contribute to future interoperating of world-wide main-stream grid projects, which is coined as GIN [4] at OGF. The deliverable results will reside on two levels: a) At the application level, we will explore what benefits could be gained on international testbeds for typical grid applications; b) At the middleware level, we will give proof-of-concept solution of technical issues regarding interconnection of common services, including authentication and authorization, data management, job submission, resource discovery and monitoring, and so on. 3. Tasks Two bioinformatics applications, RAPTOR [5] and Treeback [6], will be firstly taken as benchmarking applications for evaluating the interoperability of two testbeds. RAPTOR is one of the best protein structure prediction program. TreePack (called SCATD before) is a side-chain prediction program based on tree-decomposition of a protein backbone structure. Both of them can run on a SMP machine, a cluster system and a massively parallel processing platform. The RAPTOR will be made public available for use by bio-scientists via a web portal that serving a front-end to grid testbeds. After gaining enough experiences from initial experiments, PI and Co-PI of this project will work together with CNGrid application partners, to conduct large-scale bioinformatics research activities. TeraGrid employs CTSS as the middleware which consists of Globus Toolkit (v2 and v4), Condor and other utilities. CNGrid has its own software stack, named GOS [3], which adopts a service oriented approach compliant to many open standards including WS-I basic profile, WS-Security and SAML. Actually CNGrid software puts focus on VO-level management services that could take TeraGrid sites as members of its VOs. Concretely, this project could be divided into the following tasks with an expected schedule: (1) Application deployment on both selected TeraGrid and CNGrid sites and investigating how applications are executed on both software stacks ( CTSS and CNGrid GOS). (M1) (2) Develop application-level gateway (e.g. via a CNGrid portal or a desktop application) that allows CNGrid nodes to access TeraGrid bioinformatics resources, based on CNGrid software client libraries but using TeraGrid account. ( M2 ) (3) CAs of CNGrid join international PKI federations like PMA. Explore an approach for authentication and authorization with respect to CNGrid certificates on TeraGrid sites. Successfully submit and monitor bioinformatics jobs to TeraGrid sites with CNGrid certificates. (M3) (4) Doing the vice versa part of step 2,3 , to allow the access of CNGrid resources from TeraGrid (M4) (5) Conducting a series of benchmark tests on performance and overhead of applications run cross sites both from two testbeds, identify and consolidate requirements, benefits and issues for international grid testbeds. (M5-M6) 4. Justification for Resource Requests 5,000 SUs needed for executing parallel bioinformatics experimental applications. 10G storage for applications together with genome databases, 10G scratch space for test data. 5. References [1] TeraGrid, http://www.teragrid.org/ [2] CNGrid, http://www.cngrid.org/ [3] X Xie, N Xiao, Z Xu, L Zha, W Li, H Yu, CNGrid Software 2: Service Oriented Approach to Grid Computing, the proceedings of the UK e-Science All Hands Meeting, 2005 - allhands.org.uk [4] GIN, http://forge.gridforum.org/sf/go/projects.gin/wiki [5] Jinbo Xu, Ying Xu, Dongsup Kim, Ming Li. RAPTOR: Optimal Protein Threading by Linear Programming, the inaugural issue, Journal of Bioinformatics and Computational Biology, April 2003 [6] Jinbo Xu. Rapid Protein Side-Chain Packing via Tree Decomposition. RECOMB 2005.