波士顿咨询（BCG）：赢得DevOps速度之战（英文版）（13页）.pdf

《波士顿咨询（BCG）：赢得DevOps速度之战（英文版）（13页）.pdf》由会员分享，可在线阅读，更多相关《波士顿咨询（BCG）：赢得DevOps速度之战（英文版）（13页）.pdf（13页珍藏版）》请在皮匠网上搜索。

1、 2022 Boston Consulting Group1The Battle for Speed Is Won in the DevOpsArenaMARCH 07, 2022 By Andrew Agerbak, Jon Brock, Kaj Burchardi, Steven Alexander Kok, and Arne WeinerDevOps is critical for soware-intensive businesses. Here are five principlesto ensure that you can attain the necessary velocit

2、y.For soware-intensive businesses, the fast, effective delivery of new offerings is notonly a driver of competitive advantage but also a critical capability for long-termsurvival. This reality has prompted many companies to make substantialinvestments in DevOps, which includes practices, processes,

3、and tools to improve 2022 Boston Consulting Group2the quality and speed of the soware development and integration life cycle.Unfortunately, so far these investments have not oen translated into tangibleresults that satisfy internal stakeholders and customers.In our work with clients, we have seen fi

4、rsthand the challenges companies face ingetting the most out of their DevOps investmentsand how these efforts toimprove the soware-delivery operating model can succeed. The primary problemis that most companies have too many silos and handoffs in the development lifecycle (between customers and user

5、s, business, development, infrastructure andoperations, and security) to attain the necessary velocity in their sowaredevelopment and integration. To help companies invest effectively in DevOps,overcome these challenges, and improve speed to achieve outcomes, we havesynthesized our experience into f

6、ive key principles for success.Make Speed the Primary MetricBroadly speaking, most companies invest in DevOps for three main reasons: speed(sometimes framed as time to market), cost, and quality. But in our experience,speed is the most critical factor because, if done right, the other two will follo

7、w.Increased speed brings with it enhanced agility and efficiency (which togetherreduce costs) and the ability to iterate and improve quickly (which also boostsquality, as long as you have the right testing and feedback processes).Speed, or time to market, is sometimes too narrowly defined as time to

8、deployment (meaning the time it takes to deploy a new feature in the market). AThe fast, effective delivery of new offerings is not onlya driver of competitive advantage but also a criticalcapability for long-term survival. 2022 Boston Consulting Group3better definition incorporates the time needed

9、to achieve outcomes. There arethree core tenets of speed:To leverage these tenets and improve cycle time, companies must also providetools and services that minimize waste during development (such as simplifyingtoolchains and adding self-serve DevOps capabilities). In addition, its critical toenable

10、 and incentivize developers with a variety of training and clear DevOpsdevelopment practices (including test automation, proper branching strategy, and Creating outcome-oriented persistent teams that own these outcomesover time and are responsible for the associated assets end to end.When a team is

11、involved long term and held accountable, its more likely tofocus on longer-range goals such as improving product delivery speed. Developing principles and guardrails designed to increase productivityand ensure that problems dont get baked into solutions early on. Thisrequires introducing security, q

12、uality, and reliability engineering practicesearlier in the development life cycle. Other principles and guardrails includeputting architecture governance in place to keep code up to date (for example,by certifying the life cycle of a component so that no component goes toolong without refactoring),

13、 choosing which tools to standardize throughout thevalue chain, and determining how much to rely on soware as a service (SaaS)to cra solutions. Designing end-to-end services with self-contained components thatindividual teams can own. This enables a decoupled, microservices-styleinfrastructure and h

14、igh degree of team-level autonomy and automation.Another advantage is that if a component goes down, it does not bring downthe whole system, just a narrow set of services or capabilities. This drivesspeed to outcome by concentrating the automation and reliability controlswithin these components, mak

15、ing it easier to manage the life cycle; it alsoallows different teams to deploy at different speeds (rather than being limitedby the slowest part of the chain). 2022 Boston Consulting Group4feature toggles). Moreover, companies should streamline the operating model tomaximize the quality and speed o

16、f feature builds. All of this requires effective,outcome-oriented governance.Focus on Continuous Learning and ImprovementDevOps should not be siloed in the tech function. Its an end-to-end effort thatneeds to include both tech and nontech participation. The goal is to improvebusiness outcomes by int

17、egrating continuous learning and improvement in dailyoperations across all functions and stakeholders involved in the sowaredevelopment life cycle (SDLC). Continuous improvement is critical for anycompany to compete at speed and at scalejust as a Formula One pit crew mustconstantly improve the car i

18、f the team is to have any hope of winning the race.To apply continuous learning to agility, efficiency, and quality, the company shouldelevate the importance of certain objectives and key results (OKRs):A good way of gauging whether a companys commitment to continuous learningand improvement is adeq

19、uate is to study the budget. In our experience, 10% toDevOps should not be siloed in the technologyfunction. Its an end-to-end effort that needs to includeboth tech and nontech participation. Speed to improve business outcomes Costs spent on unnecessary or misaligned technology development Delivery

20、quality, such as fewer bugs, lower security risk, and greater customersatisfaction 2022 Boston Consulting Group520% of the overall technology budget should be allocated to continuousimprovement of delivery productivity. These DevOps investments need to focus onthree main areas:Achieving this type of

21、 focused investment, in our experience, is very challenging ifa company is still working with a project-centric delivery model. Thats because theprojects are typically funded for scope, so there is limited incentive to improve thedelivery model. But if companies do not commit enough of their change

22、budget tothese goals, they will not achieve the envisioned speed to outcome necessary tocompete in the digital realm.Consider DevOps a Transformational UndertakingDevOps done right is far more than a technology implementation involvingprocesses and toolchains: its a transformational endeavor, involv

23、ing the wholetechnology operation (and oen the business), to improve the entire deliverymodel. The implications are felt far and wide.Organization. The company must decide how best to support and increaseDevOps capabilities. That means determining what to federate, by making surethat individual deve

24、lopment teams have the right resources; and what tocentralize, by ensuring that the central function has the capability to supportindividual development teams as they improve their methods of self-serving andself-managing their assets. The key concept here is to establish DevOps as aplatform, which

25、means that central teams are not a shared service. Instead, they Increasing the levels of automation across the SDLC (build, test, deploy,release, live application monitoring and management) Boosting the reusability of components and standardization of interfacesacross the architecture Refactoring t

26、he architecture to make it more modular where necessary toimprove speed to outcome 2022 Boston Consulting Group6focus on creating capabilities and services that individual development teams canuse to boost their productivity (and ultimately cycle time). As a result, theboundary between application d

27、evelopment and operations should eventuallybreak down and be owned within the teams: “you build it, you run it; you break it,you fix it.”To a large extent, using DevOps to improve delivery productivity requires thatcompanies manage the context of application development, management,deployment, and m

28、onitoring (instead of individual projects). By “context,” wemean all aspects that affect development, including management and leadershippractices, the operating model, architectural guardrails, and tooling. Managing thecontext also allows for a more effective transfer of ownership and accountabilit

29、y sothat its as close as possible to application development and operating teams,which in turn accelerates decision makingand fulfills the principles and promisesof agile.Process. Companies need to simplify their control framework and governance toreduce handoffs and inefficient touch points. Instea

30、d of checking and testing codethemselves, employees in the DevOps environment should manage highlyautomated machines (which rely on many critical third-party assets) that providethese controls. For example, the site reliability engineering (SRE) model has clearrules, governance, technical, and opera

31、tional practices to oversee the overallquality of the technology stack. (See the sidebar, “A Global Bank Establishes SRE.”)The challenge is to consistently manage the required controls at scale in a highlyfragmented application stack, including overall quality (such as SRE models),security (the “Sec

32、” of DevSecOps), as well as continuous monitoring of theinfrastructure and live applications (AI for IT operations AIOps, for example).BCG worked with a global bank to stand up a site reliability engineering(SRE) function, establishing several small teams that could apply sowareA GLOBAL BANK ESTABLI

33、SHES SRE 2022 Boston Consulting Group7engineering to infrastructure and operational problems. SRE teams useengineers with soware expertise to perform work historically done byproduction management groups.The banks SRE teams, each with different types of infrastructure anddevelopment experience, inte

34、ract with their environmentproduction,development groups, testing teams, userswhile applying seven keyprinciples: simplify and modularize, measure everything, degrade servicesgracefully, embrace risk, set service-level objectives, automate, and respondto failures systematically. These principles hel

35、p the SRE teams and otherstakeholders streamline their production management efforts.To help embed SRE culture into the banks teams, BCG designed acurriculum that addressed two broad areas: the hard skills necessary todefine how engineering teams design and interact with platforms, and theso cultura

36、l skills required to embrace risk and break organizational silos.Critically, the bank committed to investing in these teams and thiscurriculum to improve reliability over the long term.Applications and Infrastructure. Companies need to push development into atechnology stack that is modular, can be

37、automatically deployed, and has highlevels of automated quality controls (such as those offered by todays modern cloudproviders). This requires strong coordination and engagement between thedevelopment and operations teams to design and manage the application andinfrastructure stack. The development

38、 teams take responsibility for choosing anddesigning the stack, whereas operations teams focus more on providing the stackand its elements as a self-service capability to development teams.Services and Capability Sourcing. Its important to manage the evolution ofthird-party services. Take, for examp

39、le, when a company outsources themanagement of on-premises infrastructure or application support andmaintenance. In these cases, the company and outsourcing partner need todevelop a plan to gradually automate these servicesand put the contractual 2022 Boston Consulting Group8terms and incentives in

40、place to make this happen instead of only squeezing themfor cost.Dont Leave Data BehindDevOps is not just about code; its also about data. Orchestrating the workflow tomanage complex AI and machine-learning data pipelines requires the same rigoras managing code. Moreover, good test data is crucial t

41、o ensure the effectiveness ofquality control across different environmentsincluding production, developmentteams, testing teams, and users. But in our experience, data oen doesnt get theproper attention in DevOps efforts focused on configuration management, testingautomation, and continuous soware d

42、evelopment.Its important to align improvements in data management with DevOps andarchitecture efforts to ensure end-to-end improvements in speed. (For example,even when choosing commercial off-the-shelf soware packages that use their owndata stores to operate, companies need to control their own dat

43、a and not allow thevendor to become the master repository for customer data.) Without thisalignment, its difficult to manage test data across different environmentsconsistently. And if data teams are forced to resolve these issues separately, thedelivery model will fragment.DevOps guardrails and pra

44、ctices also have an impact on data. For example,developers are sometimes uncertain about what data needs to be logged forperformance-monitoring purposes. Their response is oen to become overlycautious about what data to login some cases, overloading the services andnetwork, creating latency and pote

45、ntially instability issues.DevOps is not just about code; its also about data. 2022 Boston Consulting Group9This is why its essential to consider early on how the end-to-end stack(architecture choices, data management, and tools to manage the stack) will workthroughout the development life cycle and

46、 into live production.Invest in People and PartnersThere is a general scarcity of strong engineering talent (especially with modernDevSecOps and cloud capabilities). Even when companies can find a good outsidecandidate, they rarely find someone who deeply understands their businesscontext, history,

47、and how to drive holistic changes in the companys operatingmodel. So, in addition to battling for the right external talent, companies shouldinvest in the people already working for them, focusing on training andempowerment.In our experience, having good DevOps practices attracts talent, while the l

48、ack ofan effective SDLC can drive attrition. We worked with one client where developerssaw their code go live in as little as a week and another client where it took a year.Not surprisingly, the former was able to hire and retain talent much more easily.Companies also need to create incentives so th

49、at vendor partners will invest incontinuous improvementespecially among the vendors supporting run servicesand managing legacy on-premises infrastructure. Unfortunately, these contracts areoen managed primarily for cost. Companies keep pressuring their vendorsmargins, reducing their partners incenti

50、ves to invest or coinvest in modelimprovements.The service model itself can also discourage vendors from making theseimprovements. For example, a vendor providing testing services has no inherentincentive to automate these services. Similarly, if a partner provides applicationmaintenance and support