book

Efficient Go

by Bartlomiej Plotka

November 2022

Intermediate to advanced

502 pages

13h 50m

English

O'Reilly Media, Inc.

Book available

Read now

Unlock full access

Why I Wrote This BookHow I Gathered This KnowledgeWho This Book Is ForHow This Book Is OrganizedConventions Used in This BookUsing Code ExamplesAcknowledgmentsFeedback Is Welcome!O’Reilly Online LearningHow to Contact Us
Behind PerformanceCommon Efficiency MisconceptionsOptimized Code Is Not ReadableYou Aren’t Going to Need ItHardware Is Getting Faster and CheaperWe Can Scale Horizontally InsteadTime to Market Is More ImportantThe Key to Pragmatic Code PerformanceSummary
Basics You Should Know About GoImperative, Compiled, and Statically Typed LanguageDesigned to Improve Serious CodebasesGoverned by Google, Yet Open SourceSimplicity, Safety, and Readability Are ParamountPackaging and ModulesDependencies Transparency by DefaultConsistent ToolingSingle Way of Handling ErrorsStrong EcosystemUnused Import or Variable Causes Build ErrorUnit Testing and Table TestsAdvanced Language ElementsCode Documentation as a First CitizenBackward Compatibility and PortabilityGo RuntimeObject-Oriented ProgrammingGenericsIs Go “Fast”?Summary
Beyond Waste, Optimization Is a Zero-Sum GameReasonable OptimizationsDeliberate OptimizationsOptimization ChallengesUnderstand Your GoalsEfficiency Requirements Should Be FormalizedResource-Aware Efficiency RequirementsAcquiring and Assessing Efficiency GoalsExample of Defining RAERGot an Efficiency Problem? Keep Calm!Optimization Design LevelsEfficiency-Aware Development FlowFunctionality PhaseEfficiency PhaseSummary
CPU in a Modern Computer ArchitectureAssemblyUnderstanding Go CompilerCPU and Memory Wall ProblemHierachical Cache SystemPipelining and Out-of-Order ExecutionHyper-ThreadingSchedulersOperating System SchedulerGo Runtime SchedulerWhen to Use ConcurrencySummary
Memory RelevanceDo We Have a Memory Problem?Physical MemoryOS Memory ManagementVirtual Memorymmap SyscallOS Memory MappingGo Memory ManagementValues, Pointers, and Memory BlocksGo AllocatorGarbage CollectionSummary
ObservabilityExample: Instrumenting for LatencyLoggingTracingMetricsEfficiency Metrics SemanticsLatencyCPU UsageMemory UsageSummary
Complexity Analysis“Estimated” Efficiency ComplexityAsymptotic Complexity with Big O NotationPractical ApplicationsThe Art of BenchmarkingComparison to Functional TestingBenchmarks LieReliability of ExperimentsHuman ErrorsReproducing ProductionPerformance NondeterminismBenchmarking LevelsBenchmarking in ProductionMacrobenchmarksMicrobenchmarksWhat Level Should You Use?Summary
MicrobenchmarksGo BenchmarksUnderstanding the ResultsTips and Tricks for MicrobenchmarkingToo-High VarianceFind Your WorkflowTest Your Benchmark for Correctness!Sharing Benchmarks with the Team (and Your Future Self)Running Benchmarks for Different InputsMicrobenchmarks Versus Memory ManagementCompiler Optimizations Versus BenchmarkMacrobenchmarksBasicsGo e2e FrameworkUnderstanding Results and ObservationsCommon Macrobenchmarking WorkflowsSummary
Root Cause Analysis, but for EfficiencyProfiling in Gopprof Formatgo tool pprof ReportsCapturing the Profiling SignalCommon Profile InstrumentationHeapGoroutineCPUOff-CPU TimeTips and TricksSharing ProfilesContinuous ProfilingComparing and Aggregating ProfilesSummary

Sum ExamplesOptimizing LatencyOptimizing bytes.SplitOptimizing runtime.slicebytetostringOptimizing strconv.ParseOptimizing Memory UsageMoving to Streaming AlgorithmOptimizing bufio.ScannerOptimizing Latency Using ConcurrencyA Naive ConcurrencyA Worker Approach with DistributionA Worker Approach Without Coordination (Sharding)A Streamed, Sharded Worker ApproachBonus: Thinking Out of the BoxSummary
Common PatternsDo Less WorkTrading Functionality for EfficiencyTrading Space for TimeTrading Time for SpaceThe Three Rs Optimization MethodReduce AllocationsReuse MemoryRecycleDon’t Leak ResourcesControl the Lifecycle of Your GoroutinesReliably Close ThingsExhaust ThingsPre-Allocate If You CanOverusing Memory with ArraysMemory Reuse and PoolingSummaryNext Steps

Content preview from Efficient Go

Chapter 1. Software Efficiency Matters

The primary task of software engineers is the cost-effective development of maintainable and useful software.

Jon Louis Bentley, Writing Efficient Programs (Prentice Hall, 1982)

Even after 40 years, Jon’s definition of development is fairly accurate. The ultimate goal for any engineer is to create a useful product that can sustain user needs for the product lifetime. Unfortunately, nowadays not every developer realizes the significance of the software cost. The truth can be brutal; stating that the development process can be expensive might be an underestimation. For instance, it took 5 years and 250 engineers for Rockstar to develop the popular Grand Theft Auto 5 video game, which was estimated to cost $137.5 million. On the other hand, to create a usable, commercialized operating system, Apple had to spend way over $500 million before the first release of macOS in 2001.

Because of the high cost of producing software, it’s crucial to focus our efforts on things that matter the most. Ideally, we don’t want to waste engineering time and energy on unnecessary actions, for example, spending weeks on code refactoring that doesn’t objectively reduce code complexity, or deep micro-optimizations of a function that rarely runs. Therefore, the industry continually invents new patterns to pursue an efficient development process. Agile Kanban methods that allow us to adapt to ever-changing requirements, specialized programming languages for mobile platforms ...