提升微服务测试技巧：《Mountebank实战指南》2019

需积分: 20 141 浏览量更新于2024-07-18 收藏 10.45MB PDF 举报

《Testing Microservices with Mountebank》是一本由Brandon Byars编著的专业指南，专注于通过服务虚拟化技术测试微服务。本书深入探讨了微服务应用程序设计的各个方面，以及最新的测试实践，旨在提升读者在微服务领域的技能，并优化其开发的软件应用。该书历史部分可能会回顾微服务测试的发展历程，从早期的概念引入到Mountebank这样的工具在微服务测试中的兴起，展示了技术演进对测试策略的影响。书中涵盖了丰富的主题，如： 1. **微服务架构与设计原则**：介绍了如何设计和构建可测试、高可用的微服务，包括拆分服务、接口设计、通信协议等。 2. **服务虚拟化与模拟**：Mountebank作为主角，讲解了如何使用它创建服务的模拟版本，以便隔离真实服务进行单元测试、集成测试或端到端测试。 3. **测试策略与最佳实践**：讨论了针对微服务场景的测试方法，例如契约测试、断言和故障注入，以及如何确保测试覆盖率。 4. **教程与示例**：书中包含实战案例和步骤，帮助读者理解和应用这些理论知识，实现有效的测试自动化。 5. **优惠与购买信息**：Manning出版社提供了在线购买此书的链接，以及批量订购时可能的折扣详情，鼓励读者通过官方渠道获取。 6. **设置与配置**：书中会详细解释如何配置和管理Mountebank，以及与其他开发工具和框架的集成。 7. **支持与资源**：提供用户手册、常见问题解答和社区支持，确保读者在遇到问题时能得到及时的帮助。 8. **版权声明**：强调所有版权权益属于Manning Publications Co.，未经许可禁止任何形式的复制或传播。《Testing Microservices with Mountebank》不仅是一本技术指南，还是一本实用的工具参考书籍，帮助开发者更好地理解和应对微服务环境下的测试挑战，提升团队的生产力和软件质量。

Likeallmodels,thehighwaymetaphorisimperfect,butithighlightssomeuseful

points.AsyousawearlierintheAmazonandNetflixexamples,microservicesoften

originatefromanorganization’sdesiretoincreasefeaturethroughput.Ahelpfulside

effectisthatasmallercodebasehasahigherspeedlimit,increasingvelocity.Butboth

oftheseadvantagesarenegatedifyoudon’tchangetheorganizationtoremove

congestion.

Inorganizationalterms,fixingoverutilizationiseasyinprinciple,thoughit’soften

politicallychallenging.Youcaneitherhiremorepeople(addmorelanestothe

highway)orreducetheamountofworkinprogress(throttleentrytothehighway).

Theotherreasonsforcongestionarehardertofix.It’scommontoseeorganizations

withlessupstreamcapacitythandownstreamcapacity.Onetollgateexamplewouldbe

increasingcoordinationbyrequiringreleasestogothroughacentralrelease

managementteamthatthrottlesdevelopmentthroughput.Accidentsareevenmore

common.Everytimesomeonediscoversabugorthebuildbreaksinacodebasethat

multipleteamsshare,youhaveanaccidentrequiringcoordination,andboth

throughputandvelocitysuffer.AdrianCockcroftcitedthisexactreasonfordriving

Netflixtowardmicroservices.

Microservicesprovideatechnicalsolutionforreducingcongestioncausedbyaccidents.

Bynotsharingthesamecodebase,brokenbuildsdon’taffectmultipleteams,effectively

creatingdifferentlanesfordifferentservices.Buttollgatesarestillaproblem,andto

fullyunlockthethroughputandvelocityadvantageswehopetogainthrough

microservices,wehavetoaddressorganizationalcomplexity.Thatcomesinmany

forms(forexample,fromoperationstodatabaseschemamanagement),butthere’sone

formofupstreamcongestionthat’sparticularlyrelevanttothisbook:ourQA

organization.Microservicesfundamentallychallengethewayyoutest.

1.2. THE PROBLEM WITH END-TO-END TESTING

Traditionally,acentralQAteamcouldpartnerwithacentralreleasemanagementteam

tocoordinateascheduleofchangesthatneededtobedeployedintoproduction.They

couldarrangeitsuchthatonlyonechangewentthroughatatime,andthatchange

couldbetestedagainsttheproductionversionsofitsruntimedependenciesbefore

beingreleased.Suchanapproachisperfectlyreasonableuptoapoint.Beyondthat—

andthisisoftenwhereorganizationsturntomicroservices—it’sinappropriate.

Youstillneedconfidencethattheentiresystemwillworkwhenyoureleasepartofit.

Gainingthatconfidencethroughtraditionalendtoendtestingandcoordinated

releasescouplesalloftheservicestogether,movingthecouplingbottleneckfromthe

developmentorganizationtotheQAorganization(figure1.6).

Figure 1.6. Centralized QA processes recouple releases together, causing a

bottleneck.

Coordinatedintegrationtestingbetweenservicesrecouplescodebasesthathavebeen

decoupledthroughservicedecomposition,destroyingthescalingandrapiddelivery

advantagesofmicroservices.Assoonasyoucouplethereleasesofservicestogether,

youhavereintroducedthecommunicationandcoordinationoverheadyouweretrying

toavoid.Itdoesn’tmatterhowmanyservicesyouhave;whenyouhavetoreleasethem

atthesametime,youhaveamonolith.

Theonlywaytotrulyscalethetechnologyorganizationistodecouplereleases,sothatit

candeployaserviceindependentlyoftheservice’sdependencies(

figure1.7).This

requiresafundamentalrethinkingoftheteststrategyformicroservices.Endtoend

testingdoesn’tcompletelydisappear,butrelyingonitasagatetoreleasingsoftware

becomesanothersacrificeonthepathtowardmicroservices.Thequestionremains:

howdoyougaintheconfidenceyouneedinyourchangesbeforereleasingthem?

Figure 1.7. Independent testing works to avoid release congestion.

andfaketherest,assumingstandardinterfaces.Intraditionalmockingandstubbing

libraries,youwouldstuboutadependencyandinjectthatintoyourobject’s

constructor,allowingyourteststoprobetheobjectundertestinisolation.Withservice

virtualization,youvirtualizeaserviceandconfiguretheserviceundertesttousethe

virtualizedservice’sURLasaruntimedependency(figure1.9).Youcansetupthe

virtualservicewithaspecificsetofcannedresponses,allowingyoutoprobetheservice

undertestinisolation.

Figure 1.9. Testing using service virtualization

Servicevirtualizationletsyoudoblackboxtestingoftheservicewhiletightly

controllingtheruntimeenvironmentinwhichitoperates.Althoughitfallsshortofthe

endtoendconfidencethatintegrationtestsgiveyou,itdoesmaketestingmucheasier.

Ifyouneedtotestwhatyourshoppingcartwilldoifyoutrytosubmittheorderwhen

youareoutofinventory,youdon’thavetofigureouthowtochangetheinventory

systemtorunyourtest.Youcanvirtualizetheinventoryserviceandconfigureitto

respondwithanoutofinventorymessage.Youcantakefulladvantageofthereduced

couplingthatthenarrowserviceinterfaceprovidestodramaticallyreducetheamount

oftestsetuprequired.

Deterministictestsareteststhatalwayspassorfailwhengiventhesamecodetotest.

Nondeterminismisatester’sworstenemy.Everytimeyoutryto“fix”abrokentestby

rerunningitbecauseitworkedlasttime,thedevilonyourshoulderdoesalittledance

whiletheangelonyourothershoulderletsoutabigsigh.Automatedtestscreatea

socialcontractforateam:whenatestbreaks,youfixthecode.Whenyouallowflaky

tests,youchipawayatthatsocialcontract.Allkindsofbadbehaviormightoccurwhen

teamsloseconfidencethattheirtestsaregivingthemmeaningfulfeedback,including

剩余253页未读，继续阅读

lq1600k

粉丝: 3
资源: 5

提升微服务测试技巧：《Mountebank实战指南》2019

mountepy：用于创建HTTP（微）服务测试的实用程序。 基于Mountebank

mountebank：在线测试翻倍

testing java microservices

mountebank-util

docker-mountebank

Mountebank 配置详解

mountebank-mock-server

mb-graphql:mountebank测试使GraphQL翻倍

使用Mountebank进行微服务测试

Mountebank配置与接口管理指南

最新资源

mountepy：用于创建HTTP（微）服务测试的实用程序。基于Mountebank