当前位置:首页 > 报告详情

AWS 云 WAN MCP 服务器:利用 GenAI 变革网络运营.pdf

上传人: 明**** 编号:1013603 2025-12-21 19页 475.62KB

1、 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Andy TaylorSenior Specialist Solutions Architect,NetworkJose JuhalaSenior Solutions ArchitectNET33

2、1-AWS Cloud WAN MCP Server:Transform network operations with GenAI 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.How did we get here?Imagine the scene.Troubleshooting a large and very complex network architecture:Dual Cloud Wan Core Networks Over multiple regionsEach Core Networ

3、k had a lot of firewalls:o East/West-between regions and segments inspectiono North/South-centralised egress(to the Internet)and to and from on premise locations in each regionMultiple Transit Gateways,including peered Transit Gateways,some with Inspection VPCsTools to handFar too stubborn to not ge

4、t all the answers myself and find a way to do this in a repeatable mannerMaybe a little GenAI assisted coding Plenty of coffee Collaboration was key to successCollaborating on a task with a LOT of learning really benefited from diverse optionsCollaboration saved my sanity!2025,Amazon Web Services,In

5、c.or its affiliates.All rights reserved.Troubleshooting by hand turned to code 2025,Amazon Web Services,Inc.or its affiliates.All rights reserved.Options and Next StepsToolEffortRepeatabilityAWS ConsoleHard to repeat clickopsI have a poor memoryAWS CLILots of scripts being chainedI still have a poor

6、 memoryPython(Boto3)More upfront workAn opportunity to learn a lot and repeatableHow can I get data hop by hop in a repeatable way?Immediate GoalsAre there any security control in path,and will they allow the flow?Analyse the Cloud WAN Policy including Network Function GroupsGet prefixes/route for e

word格式文档无特别注明外均可编辑修改,预览文件经过压缩,下载原文更清晰!
三个皮匠报告文库所有资源均是客户上传分享,仅供网友学习交流,未经上传用户书面授权,请勿作商用。
根据报告的内容,全文主要内容概括如下: - **网络架构挑战**:处理大型、复杂的网络架构,包括双云WAN核心网络、多个防火墙和多个中继网关。 - **工具与挑战**:使用AWS Console、CLI和Boto3等工具,但面临重复性和记忆问题。 - **目标**:实现端到端网络路径分析,创建类似traceroute的工具,并使用AI辅助复杂故障排除。 - **方法**:开发Python脚本,逐步构建功能,使用代理和/或LLM分析数据。 - **AI代理**:介绍Agentic Loop,包括推理、行动和执行过程。 - **MCP协议**:介绍MCP协议,用于连接AI模型和数据源。 - **实践应用**:通过MCP服务器和代理文件实现稳定角色、工具使用策略和本地提示/知识连接。 - **学习与尝试**:强调通过学习和试错实现目标的重要性。 关键点: - 复杂网络架构需要高效工具和重复性解决方案。 - AI代理和MCP协议在故障排除中发挥关键作用。 - 学习和试错是成功的关键。
AI助力新篇章?" AWS云WAN的AI秘籍!" 网络运维的变革之路!"
客服
商务合作
小程序
服务号
折叠