亚马逊(Amazon)面向服务架构(SOA)经验-优快云博客

一位亚马逊前员工分享了关于面向服务架构(SOA)实施过程中的挑战与经验教训，包括问题定位困难、潜在的服务攻击问题、监控与质量保证的融合、服务发现的重要性以及跨团队代码调试难度增加等。

摘要生成于 C知道，由 DeepSeek-R1 满血版支持，前往体验 >

Amazon前员工，现google员工在Google+上说google不懂得做平台，里头也提到以前Amazon实现面向服务架构的一些经验(不过是好几年前的经验了)：

-pagerescalationgetswayharder,becauseaticketmightbouncethrough20servicecallsbeforetherealownerisidentified.Ifeachbouncegoesthroughateamwitha15-minuteresponsetime,itcanbehoursbeforetherightteamfinallyfindsout,unlessyoubuildalotofscaffoldingandmetricsandreporting.

-everysingleoneofyourpeerteamssuddenlybecomesapotentialDOSattacker.Nobodycanmakeanyrealforwardprogressuntilveryseriousquotasandthrottlingareputinplaceineverysingleservice.

-monitoringandQAarethesamething.You'dneverthinksountilyoutrydoingabigSOA.Butwhenyourservicesays"ohyes,I'mfine",itmaywellbethecasethattheonlythingstillfunctioningintheserveristhelittlecomponentthatknowshowtosay"I'mfine,rogerroger,overandout"inacheerydroidvoice.Inordertotellwhethertheserviceisactuallyresponding,youhavetomakeindividualcalls.Theproblemcontinuesrecursivelyuntilyourmonitoringisdoingcomprehensivesemanticscheckingofyourentirerangeofservicesanddata,atwhichpointit'sindistinguishablefromautomatedQA.Sothey'reacontinuum.

-ifyouhavehundredsofservices,andyourcodeMUSTcommunicatewithothergroups'codeviatheseservices,thenyouwon'tbeabletofindanyofthemwithoutaservice-discoverymechanism.Andyoucan'thavethatwithoutaserviceregistrationmechanism,whichitselfisanotherservice.SoAmazonhasauniversalserviceregistrywhereyoucanfindoutreflectively(programmatically)abouteveryservice,whatitsAPIsare,andalsowhetheritiscurrentlyup,andwhere.

- debugging problems with someone else's code gets a LOT harder, and is basically impossible unless there is a universal standard way to run every service in a debuggable sandbox.

来自https://plus.google.com/112678702228711889851/posts/eVeouesvaVX