DevOps vs. SRE: What's the Difference Between Them, and ...

And according to Google, that's where the main difference between DevOps and SRE lies. While DevOps is all about the “What” needs to be done, ... BlogCategories IndustryTrends R&D News&Updates Community Subscribe DevOpsvs.SRE:What’stheDifferenceBetweenThem,andWhichOneAreYou? HennIdan  ● 11thJul2018 9minread DevOpsorSRE?We’regoingoverthetwoconcepts,highlightingthedifferencesbetweenthemandtryingtounderstandhoweachonecametobe. DevOpsandSREseemliketwosidesofthesamecoin.Bothtitlesaimtobridgethegapbetweendevelopmentandoperationteams,withaunifiedgoalofenhancingthereleasecyclewithoutanycompromises. Andindeed,inmostcompanieswecanseethatthere’sarequirementforjustoneofthesepositions,withanoverlapinresponsibilitiesandabilities.Bothtitlesco-existinthesamespace,andbothareanessentialpartofthedevelopmentteam;sohowaretheydifferent,andwhatdoeseachonemean?Let’scheckitout. Psst!Strugglingtomaintainthereliabilityofyourapplicationswithyourcurrenttooling?OverOpsprovidesdeep,code-levelinsightabouteveryerrorandslowdownandhelpsteamsprioritizethemtoeliminateSev1issuesinproduction.   Development,OperationsandReliability BeforeDevOpswasimplemented,developmentandoperationteamsworkedastwoindependentsquads,eachwithitsowngoalsandobjectives.Thedifferencesandlackofcommunicationbetweentheseteamsoftenimpactedtheproduct,whichinreturnaffectedtheendusersandcompany. Inordertobettercommunicateandbuildbetterproducts,DevOpsbecameoneofthemostcriticalpositionsineverycompany. TheofficialdefinitionofDevOpsis“asoftwareengineeringcultureandpractice,thataimsatunifyingsoftwaredevelopmentandsoftwareoperation.”ThetermwasfirstcoinedbyAndrewShaferandPatrickDeboisbackin2008,andwhileittookafewyearsforittobecomeacommonconcept,nowadaysjustabouteverycompany,fromenterprisestostartups,arehiringDevOps. TheconceptofSiteReliabilityEngineer(SRE)hasbeenaroundsince2003,makingitevenolderthanDevOps.ItwascoinedbyBenTreynor,whofoundedGoogle’sSiteReliabilityTeam.AccordingtoTreynor,SREis“whathappenswhenasoftwareengineeristaskedwithwhatusedtobecalledoperations.” JustlikeDevOps,SREisalsoaboutcombiningdevelopmentandoperationteams,helpingthemseetheothersideoftheprocess,whileintroducingvisibilitytothecompleteapplicationlifecycle. Bothtitlesareadvocatesofautomationandmonitoring,withasimilargoaltoreducethetimefromwhenadevelopercommitsachangetowhenit’sdeployedtoproduction.DevOpsandSREsbothwanttodosowithoutcompromisingonthequalityofthecodeorproductalongtheway. GoogleitselfstatesthatSREandDevOpsarenotsodifferentfromoneanother:“they’renottwocompetingmethodsforsoftwaredevelopmentandoperations,butratherclosefriendsdesignedtobreakdownorganizationalbarrierstodeliverbettersoftwarefaster.” SowhydidGoogleneedtocreateitsowndefinition? TheDifferencesBetweenDevOpsandSREs Aswementionedbefore,theconceptofDevOpsisallaboutcombiningdevelopmentandoperations,definingthebehaviorofthesystemandseeingwhatneedstobedonetoclosethe“gap”betweenthetwoteams.Thetheorybehindthistitletalksaboutwhatneedstobedonetomakethetwoteamsworkasone. AndaccordingtoGoogle,that’swherethemaindifferencebetweenDevOpsandSRElies.WhileDevOpsisallaboutthe“What”needstobedone,SREtalksabout“How”thiscanbedone.It’saboutexpandingthetheoreticalparttoanefficientworkflow,withtherightworkmethods,toolsandsoon.It’salsoaboutsharingtheresponsibilitybetweeneveryone,andgettingeveryoneinsyncwiththesamegoalandvision. Tohelpfurtherexplainthedifference,Googlereleasedaseriesofvideosandpoststhattalkabouthowthetwotitlesdiffer.Inoneoftheseposts,writtenbytwoGoogleemployees:SethVargo,StaffDeveloperAdvocateandLizFong-Jones,SiteReliabilityEngineer,theyexplainthatSRE“embodythephilosophiesofDevOpswithagreaterfocusonmeasuringandachievingreliabilitythroughengineeringandoperationswork.” SethandLizrepresentedthesimilaritiesanddifferencesbetweenthetwothroughthetop5pillarsofDevOps,explainingwhattheymeanforSRE: #1ReduceOrganizationalSilos Largeenterprisesusuallyhaveacomplexorganizationstructure,withalotofteamsworkinginsilos.Eachteamispullingtheproductinadifferentdirection,notcommunicatingwiththerestofthecompanyandasaresult,failtoseethebigpictureasawhole.Thiscanleadtofrustration,asetbackindeploymentandhighcostsduetodelays. DevOps’jobistoreducethesilos,andtomakesuretherearen’tanyteamswithinteamswhoarenotalignedwiththerestofthecompany.Theyminimizeandbridgetheteamsintoonegroup,withasharedvision. SREsdon’ttalkabouthowmanysilosareinthecompany,butmoreabouthowtogeteveryonetodiscuss.Thisisdonebyusingthesametoolsandtechniquesacrossthecompany,whichinreturnhelpssharetheownershipacrosseveryone. #2AcceptFailureasNormal AlthoughtheconceptofDevOpsisabouthandlingandcopingwithissuesbeforetheyfail,failureissomethingthatwe,unfortunately,can’tavoid.DevOpsembracesthisbyacceptingfailureassomethingthatisboundtohappen,andwhichcanhelptheteamlearnandgrow. IntheworldoftheSREs,thisobjectiveisdeliveredbyhavingaformulaforbalancingaccidentsandfailuresagainstnewreleases.Inotherwords,SREswanttomakesurethattherearen’ttoomanyerrorsorfailures,evenifit’ssomethingthatwecanlearn. Thisformulaismeasuredwithtwokeyidentifiers:ServiceLevelIndicators(SLIs)andServiceLevelObjectives(SLOs). SLIsmeasurethefailuresperrequest,bycalculatingrequestlatency,throughputofrequestspersecond,orfailuresperrequestasmeasuredovertime.SLOsderiveoutofthisthreshold,percentageornumber,andrepresentthesuccessofSLIsoveracertainamountoftime. #3ImplementGradualChange Companieswanttomovefasterthanbefore.Theywantfrequentreleases,continuallyupdatingtheproductandkeepingteammembersontheirtoesaboutnewandrelevanttechnology. DevOpsareallforthischange,butinagradualandhandledway.BothDevOpsandSREswanttomovequickly,andGooglepointsoutthatSREsemphasizesreducingthecostoffailureastheydoso. #4LeverageToolingandAutomation Aswementionedbefore,oneofthemainfocalpointsforbothDevOpsandSREsisautomation.Bothtitlesencourageaddingasmuchautomationandtoolsaspossible,aslongastheyprovidevaluetodevelopersandoperationsbyremovingmanualtasks. #5MeasureEverything Anautomatedworkflowthatmovesfastissomethingthatneedsconstantmonitoring.DevOpsandSREteamsbothneedtomakesurethatthey’removingintherightdirection,andtheydosobymeasuringeverything. ThemaindifferencehereisthatSREsrevolvesaroundtheconceptthatoperationsisasoftwareproblem,whichledthemtodefineprescriptivewaysformeasuringavailability,uptime,outages,toil,etc. SREsalsoensurethateveryoneinthecompanyagreesonhowtomeasurereliability,andwhattodowhenavailabilityfallsoutofspecification.Thisincludescontributorsateverylevel,fromdevelopers,throughteammanagersandallthewayuptoVPsandexecutives. [easy-tweettweet=”What’sthedifferencebetween#DevOpsand#SRE?Itturnsout,notmuch…#samegamedifferentname”user=”overopshq”url=”http://bit.ly/2SyDyPw”template=”lite”] WhatDoesItMeanToBeReliable? Wetalkedaboutsharingresponsibility,acceptingfailureandmeasuringeverything.Now,weneedawaytomakesureeverythingisindeedworkingasitshould,andisreliable.Inotherwords,thereshouldbeaunifiedmethodtomeasurereliabilityateverylevel. SREsaremeasuringSLIsandSLOs,andDevOpsteamsmeasurethefailurerate,aswellasthesuccessrateovertime,andbothusuallydosowithdifferenttoolsandmethods.Whiletheseteamshaveanoverviewofwhat’sgoingon,it’snotcomplete.Reliabilityisnotjustabouttheinfrastructure,it’srelevanteverystepoftheway–fromapplicationquality,throughperformanceanduptosecurity. Failureandissuescanandwillhappenindifferentaspectsoftheapplication,andwhenitdoes,weneedtohavereliabledatatounderstandwhytheissuehappenedinthefirstplace,whatcausedit,andhowtofixit.Ifwebreakitdown,thisdatashouldinclude: Executionstackandbytecode Completevariablestate(overlayedonfullsourcecode) JVMState:Threads,environmentvariables Relevantlogstatements(includingDEBUGandTRACEinproduction) Eventanalytics(Frequency,failurerate,deployment,application) Andsincethisiscrucialinformation,wehavetomakesureit’sreliableandactionable.Thiscanbedonewiththehelpofsettingupalertsfordifferentscenarios,embracingamethodofpeercodereview,unittestsandsoon. Whilethesemethodshelppromoteasharedresponsibilitybetweeneveryone,theymightendupimpactingtheproduct’sperformance.Andthebiggertheorganizationis,thehigherthecostforfailure,whetherit’scustomersatisfaction,employeechurn,oradecreasedproductvalue. That’swhyit’simportanttominimizethemanualsystemwork,andautomatethecollectionofinformation.Andwhileyou’reatit,youalsoneedtostayontopofeverythingthat’shappeninginyourproduct.Inotherwords,youneedtherightdatatomeasurethereliabilityofyoursoftwarethroughouttheCI/CDworkflow. Signupforourtrial toseehowyoucanproactivelydetectnewandcriticalissueswiththecompletecontextrequiredtoresolve.Signupnow. FinalThoughts So,isthereadifferencebetweenDevOpsandSREs?Google,the“founder”oftheSREtitleclearlydefinedit,alongwithastraightforwardsetofexpectations.DevOps,asitseems,ismoreofa“freespirit”,withthedefinitionandperspectivesvaryingfromorganizationtoorganization. However,DevOpsandSREteamsarenotsodifferent.Bothhelpcombinedeveloperandoperationteams,whilesharingsimilarresponsibilitiesandfocusingonenablingautomationandreliability. Thebottomlineisthatit’sallaboutthedata.Youneedinformationinordertounderstandhowtomeasuresuccessandfailureandhowtogaincontinuousreliabilityacrosstheapplication. HennIdan HennisamarketingmanageratOverOpscoveringtopicsrelatedtoJava,Scalaandeverythinginbetween.Sheisaloverofgadgets,apps,technologyandtea. devops sre TroubleshootingApacheSparkApplicationswithOverOps OverOps’abilitytodetectpreciselywhysomethingbrokeandtoseevariablestateisinvaluableinadistributedcomputeenvironment. TroubleshootingApacheSparkApplicationswithOverOps LearnMore   NextArticle StackOverflow:TopJavaQuestionsWeDidn'tKnowWeHad-andHowtoAnswerThem 6minread TheFastestWaytoWhy. EliminatethedetectiveworkofsearchinglogsfortheCauseofcriticalissues.Resolveissuesinminutes. LearnMore Weareusingcookiestogiveyouthebestexperienceonourwebsite. Youcanfindoutmoreaboutwhichcookiesweareusingorswitchthemoffinsettings. Accept CloseGDPRCookieSettings PrivacyOverview StrictlyNecessaryCookies Poweredby GDPRCookieCompliance PrivacyOverview Thiswebsiteusescookiessothatwecanprovideyouwiththebestuserexperiencepossible.Cookieinformationisstoredinyourbrowserandperformsfunctionssuchasrecognisingyouwhenyoureturntoourwebsiteandhelpingourteamtounderstandwhichsectionsofthewebsiteyoufindmostinterestinganduseful. StrictlyNecessaryCookies StrictlyNecessaryCookieshouldbeenabledatalltimessothatwecansaveyourpreferencesforcookiesettings. EnableorDisableCookies Ifyoudisablethiscookie,wewillnotbeabletosaveyourpreferences.Thismeansthateverytimeyouvisitthiswebsiteyouwillneedtoenableordisablecookiesagain. EnableAll SaveSettings