효율적이고 신뢰성 있는 네임 서비스 시스템의 구조 및 실험

심영철,박준철,강호석,이준원 한국통신학회 2004 韓國通信學會論文誌 Vol.29 No.3B

네임 서비스 시스템은 도메인 네임을 인터넷 주소로 또는 역으로 인터넷 주소를 도메인 네임으로 변화하는 서비스를 제공하며 웹, 이메일, 파일전송 등의 모든 인터넷 응용이 가장 먼저 사용하는 인터넷에서 가장 중요한 서비스 중의 하나이다. 본 논문에서는 상위 레벨의 대규모 도메인을 가정하여, 이 도메인에 대한 서비스를 제공하는 네임서버를 중심으로 성능 향상 및 신뢰성 제고의 방안을 제시하고, 테스트베드 구축 띤 성능 실험을 통해 이의 실현 가능성을 검증한다. 제안한 구조는 이중화된 네임서버를 가정하여 부하 분담 및 고신뢰성 달성이라는 두 가지 목표를 동시에 이루고자 하였다. 또한 매스터 네임서버의 데이터 갱신을 위해 동적 갱신 방식을 사용하였고, 슬레이브 네임서버고의 데이터 전송을 위해 IXFR(Incremental Zone Transfer)을 사용하여 서버의 부담을 최소화하고자 하였다. 테스트 베드의 실험 결과를 실제 구축될 시스템에 적용할 경우, 향후 서버에의 부하 증가에 효율적으로 대처할 수 있으며 시스템의 다운이나 오동자의 경우에도 연속적인 서비스 수행이 가능하다고 판단된다. Domain Name System(DNS), one of the most important Internet services, handles mapping from host names to Internet addresses and vice versa, and precedes many Internet applications such as Web, e-mail, file transfer, etc. In this paper, we propose a structural design of a generic name server system providing name services for a huge domain for the purpose of improving the performance as well as the reliability of the system. We demonstrate the validity of the design by implementing and running a testbed system. Our testbed employs a couple of master name sowers for distributing the service overhead over two, rather than one, servers and for achieving high availability of the system as a whole. We suggest the use of dynamic update to add and delete records from a zone for which the name server has authority. The slave name servers located remotely then get a new, updated copy of the zone via incremental zone transfers(IXFRs). The experiments with the implemented testbed show that the proposed structure would easily manage increasing demands on the server power, and be highly available in the face of transient faults of a module in the system.

Global computing 시스템에서의 문제점들과 스케쥴링

沈永喆 弘益大學校 科學技術硏究所 2000 科學技術硏究論文集 Vol.11 No.-

A large scale virtual parallel system which is composed of a large number of computers on the Internet and executes parallel programs is called a metacomputing system or a global computing system. In this paper we first introduce the global computing research which has already been performed or is being performed now and then discuss important issues that should be considered when we design and implement global computing systems. Among these issues scheduling becomes one of the most important issues and we propose the structure for a scalable distributed scheduler system. The proposed scheduler system has the composite structure of a tree and a graph and the characteristic of being both efficient and easily controllable.

특별강연 : 재테크 완전정복

심영철 대한내과학회 2005 대한내과학회 춘계학술대회 Vol.68 No.10

통신망 관리를 위한 소프트웨어의 개발에 관한 연구

沈永喆 弘益大學校 科學技術硏究所 1994 科學技術硏究論文集 Vol.4 No.-

As communication networks are becoming larger, more complex, and more heterogeneous, the problem of managing them is becoming a more important issue. OSI standards for the network management are widely accepted and are used or will be used for many data communication networks and telecommunication networks. An OSI management system consists of a manager and agent processes. An agent process collects status information from a managed system and send them to the manager. It receives management operations from the manager and executes them or sends them to the managed system. In this paper we proposed a software architecture for an agent process. An agent process consists of 4 modules: an OSI agent module, a physical managed object module, a logical managed object module, and a managed system interface module. We also presented a tool environment for automatically generating these modules.

Support for the Software Development Using Extended Relational Database Systems

沈永喆 弘益大學校 科學技術硏究所 1994 科學技術硏究論文集 Vol.5 No.-

좋은 품질의 소프트웨어를 높은 생산성으로 생산할 수 있기 위해서 소프트웨어 공학의 분야는 데이타베이스, 인공지능, 그래픽스 등과 같은 타 분야에서 개발된 기술들을 많이 활용하고 있다. 이중 데이타베이스의 영향이 가장 크다고 할 수 있다. 본 논문에서는 먼저 소프트웨어공학이 데이타베이스로부터 어떠한 자원을 필요로 하는지 알아보고 다음 데이타베이스가 소프트웨어 공학을 어떻게 도울 수 있는지 분석한다. 특히 소프트웨어 공학의 기법 중 프로세스 관리, 도구 통합, 재사용, 자동 프로그래밍, 메트릭스를 사용한 소프트웨어 개발 등에 데이타베이스가 어떻게 활용될 수 있는지 분석한다.

통합 보안 관리를 위한 기반구조의 개발

沈永喆,姜昊錫 弘益大學校 科學技術硏究所 2002 科學技術硏究論文集 Vol.13 No.-

In this paper we describe the infrastructure for the policy-based integrated security management so that the security of networks can be effectively managed. The proposed infrastructure is based on the manager-agent concept. The central security manager stores the security policy in the database and monitors the network topology and the security configuration of nodes in the network in real-time. If the security configuration of the network does not conform to the security policy under the current network topology, the manager generates and executes management operations on the network nodes so that their security configuration may conform to the security policy. An agent program is installed in each network node such as hosts and routers and it informs the manager of any security related events occurring in the node and receives/executes the management operations received from the manager. We also describe the UML-based information modeling method of a network, the required operations for the security management, the secure communication protocol, and the security policy specification language.

CMIP을 이용한 침입관제 프레임워크 설계

沈永喆,朴恩英 弘益大學校 科學技術硏究所 2001 科學技術硏究論文集 Vol.12 No.-

Recently the requirement for security is increasing because the number of intrusions is increasing. It is important to make all the security system cooperate in a large scale network, and the affects the overall structure of the whole system. To address these issues, it is required to develop a unified system which detects and analyzes intrusions in a large network. We use the approach in a network management for the interaction and cooperation of intrusion surveillance systems. In this thesis, we build ER-model for representing both the security policy and the network topology, and define MOs and MIBs of components of an intrusion surveillance system. We also propose an extension to the Common Intrusion Specification Language (CISL), the language adopted by CIDF, to model request among components of our system.

Partnership For win-win Growth & Success (태평양 R&D 사례 위주)

심영철 한국생물공학회 2004 한국생물공학회 학술대회 Vol.2004 No.10

다변수 VSS에 관한 연구 ( ON MIMO VARIABLE STRUCTURE SYSTEM )

심영철,최종덕,변증남 대한전자공학회 1980 대한전자공학회 학술대회 Vol.1980 No.1

TINA-C ODL 개발 환경의 디자인

변경근,심영철 弘益大學校 科學技術硏究所 1996 科學技術硏究論文集 Vol.7 No.2

TINA-C not only provides basic concepts and a language for specifying distributed applications which provide services in telecommunication networks but also defines a distributed processing environments(DPE) on which distributed applications can be deployed and executed with location and heterogeniety transparency. But an actual environment with which to develop and test distributed applications according to TINA-C concepts dose not exist. In this paper we describe how we can use OMG CORBA to build a distributed development and execution environment for TINA-C. OMG CORBA and TINA-C environments differ in two respects: (1)definition language for distributed applications and (2)distributed servers. The difference in definition languages can be overcome by introducing a translator from the TINA-C definition language to the OMG definition language and adding to the underlying kernel or NCCE certain functions to support stream communications with QoS and transactions. OMG CORBA dose not provide any servers. Therefore, distributed servers such as a trader, an authentication server, and system management servers should be implemented. This paper explains these approaches.3