benchmark

Documentation

Overview

Package benchmark package provides tools to run progressive or independent benchmarks against teleport services.

Index

• func ExportLatencyProfile(path string, h *hdrhistogram.Histogram, ticks int32, valueScale float64) (string, error)
• type Config
  • func (c *Config) Benchmark(ctx context.Context, tc *client.TeleportClient, suite Suite) (Result, error)
• type KubeExecBenchmark
  • func (k KubeExecBenchmark) BenchBuilder(ctx context.Context, tc *client.TeleportClient) (WorkloadFunc, error)
• type KubeListBenchmark
  • func (k KubeListBenchmark) BenchBuilder(ctx context.Context, tc *client.TeleportClient) (WorkloadFunc, error)
• type Linear
  • func (lg *Linear) GetBenchmark() *Config
• type Result
  • func Run(ctx context.Context, lg *Linear, host, login, proxy string, suite Suite) ([]Result, error)
• type SSHBenchmark
  • func (s SSHBenchmark) BenchBuilder(ctx context.Context, tc *client.TeleportClient) (WorkloadFunc, error)
• type Service
• type Suite
• type WebSSHBenchmark
  • func (s WebSSHBenchmark) BenchBuilder(ctx context.Context, tc *client.TeleportClient) (WorkloadFunc, error)
• type WebSessionBenchmark
  • func (s *WebSessionBenchmark) BenchBuilder(ctx context.Context, tc *client.TeleportClient) (WorkloadFunc, error)
  • func (s *WebSessionBenchmark) Config(ctx context.Context, tc *client.TeleportClient, cfg *Config) error
• type WorkloadFunc

Functions

func ExportLatencyProfile

func ExportLatencyProfile(path string, h *hdrhistogram.Histogram, ticks int32, valueScale float64) (string, error)

ExportLatencyProfile exports the latency profile and returns the path as a string if no errors

Types

type Config

type Config struct {
	// Rate is requests per second origination rate
	Rate int
	// MinimumWindow is the min duration
	MinimumWindow time.Duration
	// MinimumMeasurments is the min amount of requests
	MinimumMeasurements int
}

Config specifies benchmark requests to run

func (*Config) Benchmark

func (c *Config) Benchmark(ctx context.Context, tc *client.TeleportClient, suite Suite) (Result, error)

Benchmark connects to remote server and executes requests in parallel according to benchmark spec. It returns a benchmark result when completed. This is a blocking function that can be canceled via context argument.

type KubeExecBenchmark

type KubeExecBenchmark struct {
	// Namespace is the Kubernetes namespace to run the command against.
	Namespace string
	// PodName is the name of the pod to run the command against.
	PodName string
	// ContainerName is the name of the container to run the command against.
	ContainerName string
	// Command is the command to run.
	Command []string
	// Interactive turns on interactive sessions
	Interactive bool
}

KubeListBenchmark is a benchmark suite that runs successive kubectl exec against a Teleport Kubernetes proxy for a given duration and rate.

func (KubeExecBenchmark) BenchBuilder

func (k KubeExecBenchmark) BenchBuilder(ctx context.Context, tc *client.TeleportClient) (WorkloadFunc, error)

BenchBuilder returns a WorkloadFunc for the given benchmark suite.

type KubeListBenchmark

type KubeListBenchmark struct {
	// Namespace is the Kubernetes namespace to run the command against.
	// If empty, it will include pods from all namespaces.
	Namespace string
}

KubeListBenchmark is a benchmark suite that runs successive kubectl get pods against a Teleport Kubernetes proxy for a given duration and rate.

func (KubeListBenchmark) BenchBuilder

func (k KubeListBenchmark) BenchBuilder(ctx context.Context, tc *client.TeleportClient) (WorkloadFunc, error)

BenchBuilder returns a WorkloadFunc for the given benchmark suite.

type Linear

type Linear struct {
	// LowerBound is the lower end of rps to execute
	LowerBound int
	// UpperBound is the upper end of rps to execute
	UpperBound int
	// Step is the amount of rps to increment by
	Step int
	// MinimumMeasurements is the minimum measurement a benchmark should execute
	MinimumMeasurements int
	// MinimumWindow is the minimum duration to run benchmark for
	MinimumWindow time.Duration
	// contains filtered or unexported fields
}

Linear generator

func (*Linear) GetBenchmark

func (lg *Linear) GetBenchmark() *Config

GetBenchmark returns the benchmark config for the current generation.

type Result

type Result struct {
	// RequestsOriginated is amount of requests originated
	RequestsOriginated int
	// RequestsFailed is amount of requests failed
	RequestsFailed int
	// Histogram holds the response duration values
	Histogram *hdrhistogram.Histogram
	// LastError contains last recorded error
	LastError error
	// Duration it takes for the whole benchmark to run
	Duration time.Duration
}

Result is a result of the benchmark

func Run

func Run(ctx context.Context, lg *Linear, host, login, proxy string, suite Suite) ([]Result, error)

Run is used to run the benchmarks, it is given a generator, command to run, a host, host login, and proxy. If host login or proxy is an empty string, it will use the default login

type SSHBenchmark

type SSHBenchmark struct {
	// Command to execute on the host.
	Command []string
	// Random whether to connect to a random host or not
	Random bool
}

SSHBenchmark is a benchmark suite that connects to the configured target hosts and executes the provided command.

func (SSHBenchmark) BenchBuilder

func (s SSHBenchmark) BenchBuilder(ctx context.Context, tc *client.TeleportClient) (WorkloadFunc, error)

BenchBuilder returns a WorkloadFunc for the given benchmark suite.

type Service

type Service string

Service is a the Teleport service to benchmark.

const (
	// SSHService is the SSH service
	SSHService Service = "ssh"
	// KubernetesService is the Kubernetes service
	KubernetesService Service = "kube"
)

type Suite

type Suite interface {
	// BenchBuilder returns a function that executes a single benchmark call.
	// The returned function is called in a loop until the context is canceled.
	BenchBuilder(context.Context, *client.TeleportClient) (WorkloadFunc, error)
}

Suite is an interface that defines a benchmark suite.

type WebSSHBenchmark

type WebSSHBenchmark struct {
	// Command to execute on the host.
	Command []string
	// Random whether to connect to a random host or not
	Random bool
	// Duration of the test used to determine if renewing web sessions
	// is necessary.
	Duration time.Duration
}

WebSSHBenchmark is a benchmark suite that connects to the configured target hosts via the web api and executes the provided command.

func (WebSSHBenchmark) BenchBuilder

func (s WebSSHBenchmark) BenchBuilder(ctx context.Context, tc *client.TeleportClient) (WorkloadFunc, error)

BenchBuilder returns a WorkloadFunc for the given benchmark suite.

type WebSessionBenchmark

type WebSessionBenchmark struct {
	// Command to execute on the host.
	Command []string
	// Max number of sessions to have open at once.
	Max int
	// Duration of the test used to determine if renewing web sessions
	// is necessary.
	Duration time.Duration
	// contains filtered or unexported fields
}

WebSessionBenchmark is a benchmark suite that connects to the configured target hosts via the web api and executes the provided command.

func (*WebSessionBenchmark) BenchBuilder

func (s *WebSessionBenchmark) BenchBuilder(ctx context.Context, tc *client.TeleportClient) (WorkloadFunc, error)

BenchBuilder returns a WorkloadFunc for the given benchmark suite.

func (*WebSessionBenchmark) Config

func (s *WebSessionBenchmark) Config(ctx context.Context, tc *client.TeleportClient, cfg *Config) error

type WorkloadFunc

type WorkloadFunc func(context.Context) error

WorkloadFunc is a function that executes a single benchmark call.