自动化脚本驱动的网络管理系统SPLICE/NM：高效运维解决方案

需积分: 8 169 浏览量更新于2024-08-07 收藏 195KB PDF 举报

本文是一篇关于"基于脚本化操作的网管系统SPLICE/NM的设计与实现"的学术论文。作者Kiyohiko Okayama和Suguru Yamaguchi来自日本奈良科学与技术学院，而Hideo Miyahara则来自大阪大学工程科学学院。论文的背景指出，在当今互联网服务迅速发展和网络问题日益复杂化的背景下，传统的网络管理系统在问题检测和处理速度以及准确性上存在不足，这使得网络管理员面临巨大的压力。 SPLICE/NM系统的主要目标是通过引入自动化恢复操作来提高网络管理效率。它的核心思想是让网络管理员能够使用一种编程语言来编写描述他们在终端上执行的操作，例如配置更改、故障诊断或恢复策略。管理员生成的脚本被记录并存储，当系统检测到网络问题时，这些预先编写的脚本会自动执行，从而实现故障响应的自动化。这种方法的优势在于，它简化了管理过程，减少了人为错误的可能性，并允许管理员将精力集中在更高层次的战略决策上，而不是频繁的手动干预。此外，SPLICE/NM的设计也考虑到了技术的扩展性和灵活性，可以根据网络环境的变化和新出现的问题进行相应的调整和升级。论文详细探讨了SPLICE/NM的设计原则、架构、实现方法以及如何通过自动化脚本实现对网络设备的管理和监控。它可能还涵盖了测试案例、性能评估以及与传统管理方式的比较，以证明其在实际网络环境中的有效性。这篇文章对于那些关注网络自动化管理，尤其是希望通过脚本化技术提升问题解决速度和精确度的IT专业人士具有重要的参考价值，展示了在快速变化的互联网环境中，创新的网管解决方案是如何应对挑战并优化网络运维的。

control reflecting policies of all participating organizations,

distribution of servers with consideration of network

scalability, and so on.

In the current study, after defining the faults handled

by SPLICE/NM, the authors will discuss the issues of

cooperative management between the distributed servers,

and then make evaluations and identify problems regarding

a real prototype based on the SPLICE/NM design.

2. Fault Definition in SPLICE/NM

Network management covers a wide field, from the

network equipment structure to user account generation and

billing. Considering various existing technologies and ap-

proaches, it is not feasible to discuss network management

as a whole.

Therefore, in OSI the entire network management is

subdivided into five classes: structure management, per-

formance management, fault management, security

management, and account management. SPLICE/NM

concentrates on the fault management and is intended for

automation of fault recovery tasks as they are convention-

ally performed by managers, but since there are all kinds of

faults that can occur on the network, they are perceived

differently by different people.

For example, although network performance deterio-

ration might be considered a fault, it could be difficult to

subdivide network deterioration cases into those regarded

as a temporary condition and those that at a certain stage

can be viewed as a fault. Solving this problem would require

technology for performance management such as that for

monitoring network traffic on a permanent basis. In addi-

tion, in a large-scale network such as the Internet, for

various reasons, such as the presence of communication bot-

tleneck sections outside of the current organization, prob-

lems cannot be solved without spending time and money.

Therefore, in SPLICE/NM, we have limited the fault

types and concentrated on network connectivity. In other

words, we consider a fault as a condition when it is impos-

sible to reach a specific host or network, or a condition when

a specific service cannot be used.

In addition, the network equipment targeted by

SPLICE/NM is a gateway functioning as a connecting point

between networks, or a server offering network services.

The term management object will refer to such types of

equipment.

3. Fault Management Model in

SPLICE/NM

Fault management in a network discussed in the time

flow aspect can be broken down into the following phases:

1. Fault discovery

2. Fault diagnosis (identification of cause and devel-

opment of countermeasures)

3. Recovery operations

4. Recording of operations results

In a conventional accident management system, each

phase is managed by a manager or a user using this network.

Among these, the fault discovery phase has long been a

target of attempts at automation, with the resulting network

monitoring system using SNMP. In addition, there is the

trouble ticket system, which involves creating a cooperative

operation environment between multiple managers. This

system can be used for recording the recovery operations

results.

The SPLICE/NM system proposed in the current

study concentrates on the phases of fault diagnosis and

recovery operations, aiming at automating the operations

required at these phases. Figure 1 illustrates a fault manage-

ment model for the case of a SPLICE/NM system combined

with a network monitoring system and a trouble ticket

system.

In the fault management model shown in Fig. 1, the

management object is usually monitored by the network

monitoring system. If a fault occurs in a management

object, then it is detected by the network monitoring system

and reported to SPLICE/NM. SPLICE/NM performs diag-

nostics of the fault in the indicated management object, and

conducts recovery operations, with the results reported to

the trouble ticket system. During this process, if it is impos-

sible to fully recover from the fault, the trouble ticket

system contacts the manager, and the recovery operations

are delegated to the manager. Finally, on recovering from

the fault, the operations results are recorded by the manager.

For implementing the fault management model using

SPLICE/NM, it is important to know how the information

Fig. 1. Management model of SPLICE/NM.

剩余10页未读，继续阅读

weixin_38672794

粉丝: 5
资源: 924

自动化脚本驱动的网络管理系统SPLICE/NM：高效运维解决方案

Design and implementation of network management system SPLICE/NM based on scripted operations

JavaScript splice 数组操作（删除，插入）

基于脚本操作的网络管理系统SPLICE/NM设计与实现

自动化脚本驱动的网络管理系统SPLICE/NM设计与高效恢复

js利用Array.splice实现Array的insert/remove

javascript splice数组简单操作

splice基于Electron的前端文件处理工具

JS数组splice操作实例分析

Webkit中Array Splice与字符编码转换的实现

C++列表操作：splice()和unique()在算法设计中的应用

最新资源