Scaling applications with go

SCALING APPLICATIONS
WITH GO
-Vimlesh Sharma

 Introduction
 What does Go offer?
 Where go can be used?
 Go in Production
 Go in comparison to other language
 Case studies
TABLE OF CONTENTS

 Go is a general-purpose programming language, like Python,
Java, or C
 It is commonly referred to as golang.
 A programming language developed at Google in 2007
 Main contributor were Robert Griesemer, Rob Pike, and Ken
Thompson.
INTRODUCTION

 Go is relatively
 compact language
 easier to learn
 Go language spec is webpage (http://golang.org/ref/spec )
that prints as a 56 page PDF compared to others
 Java SE – 768 pages http://docs.oracle.com/javase/specs/
 Scala – 183 pages http://www.scala-
lang.org/docu/files/ScalaReference.pdf
 Go documentation is better than other languages.
A SIMPLE LANGUAGE THAT IS EASY TO
LEARN AND READ.

 Go combines ease of programming in an interpreted &
dynamically typed language.
 Also it provides high efficiency and safety of a statically
typed, compiled language.
var intA int
intA = 32 // intA has 32 and is static type int
intB := 42 // intB has value 42 and dynamic type int
STATICALLY TYPED, BUT WITH A DYNAMIC
FEEL

 Like C and C++, Go compiles to native machine code so that
we don’t need environments such as CLR and JVM to run Go
applications.
 Go compiler compiles programs very quickly that helps
particularly when compiling large applications
 ‘Gocc’ is a go compiler’s compiler
 https://code.google.com/p/gocc/
COMPILED TO NATIVE MACHINE CODE

 Go is concurrent system
 Concurrency is a way to structure a program by breaking it
into pieces that can be executed independently.
 Concurrency vs. parallelism
 Concurrency is about dealing with lots of things at once
 Parallelism is about doing lots of things at once
 Not the same, but related
 Concurrency is about structure, parallelism is about execution
 Concurrency provides a way to structure a solution to solve a
problem that may (but not necessarily) be parallelizable
GO SUPPORTS CONCURRENCY

 In Go, concurrency is a first-class citizen in the core language
 Go introduces Goroutines, which lets you run functions
concurrently.
 You can communicate and pass values between Goroutines
using Channels.
 Goroutines and Channels are great features of Go that let you
leverage concurrency and parallelism
LANGUAGE-LEVEL CONCURRENCY FEATURES

 Go’s standard libraries are exemplary.
 Go also provide 3rd party packages.
 Go’s package management system is designed to work with
modern scenarios
 e.g. sharing it with developers overs GIT repository) unlike NPM,
NuGet, RubyGem
EXPANSIVE STANDARD LIBRARY

 Go provides lots of tools for managing documentation,
testing, package management, and more.
 go doc : can be used to pull documentation for anything in
codebase.
 go test : This is one of the easiest testing setup
 go get : This is used to pull the shared source code from GIT
GREAT TOOLS

 Server application
 Command-line tools - https://github.com/codegangsta/cli
 Web applications
 Games - http://go-ngine.com/
 Scientific computing -
https://archive.fosdem.org/2014/schedule/event/hpc_devroo
m_go/
 Android & IOS development (in upcoming versions)
 Many More...
USAGE

 Go produces statically linked binaries that are self sufficient.
 Single binary makes deployment super-easy.
 This is great when compared to other dependency setup that
is required for most other application.
SINGLE BINARY

 Go has excellent support for cross compilation.
 gonative is a simple tool which creates a build of Go that can
cross compile to all platforms. –
http://github.com/inconshreveable/gonative
 http://dave.cheney.net/2013/07/09/an-introduction-to-cross-
compilation-with-go-1-1
CROSS COMPILATION

 There is excellent support for
 Live profiling capability for CPU
 Memory
 Goroutines
 OS threads
 Lock contention etc
 Read more at http://blog.golang.org/profiling-go-programs
 Use go toolchain and profile the application on-demand.
BUILT-IN PROFILING

C++ Java Javascript Scala Go
Type Safe
Garbage
Collection
Compilation Slow Slow NA Slow Fast
Concurrency
Ease of
Programming
Efficiency Highest Moderate Slow Good Close to C++
GO IN COMPARISON TO OTHER LANGUAGE

BENCHMARKING ON JSON RESPONSE.
Nodejs 32.8% https://www.techempower.com/benchmarks/#section=data-r9&hw=i7&test=json

CASE STUDY FROM
GOPHERCONINDIA 15

 Ruby is a complex language
 Memory consumption and speed
 Lot of caching data in memory for speed, GC in Ruby consumed a lot
of time ultimately hindering performance.
 Concurrency
 The safest way to do concurrency in Ruby is through the use of
multiple processes, so app uses many workers and hence memory
consumption.
FROM A RUBY MONOLITH TO MICROSERVICES
IN GO

 Instead of refactoring the entire application, moving selected
component to a micro services architecture in go increased
the efficiency.
 Main Application + Go Micro services boost application
performance.
 https://sourcegraph.com/blog/live/gopherconindia/1126565
68167
SOLUTION/OBSERVATION

 N1QL pronounced as “nickel”.
 N1QL is a query language developed in Go for Couchbase, a
distributed NoSQL database.
 With N1QL scalable application can be developed quickly
N1QL: A QUERY LANGUAGE IN GO

SIGNIFICANT LOWER DEVELOPMENT COST.
C++ 210 days Go 150 days C++ 400 bugs Go 250 bugs

 Benchmarks show that Go compares favorably with C++ and
Java
 In C++, they had to experiment with a bunch of different
libraries
 In Java, they had to fiddle around with a bunch of parameters
like the min heap size
 Go worked well out of the box -
https://sourcegraph.com/blog/live/gopherconindia/1116380
91937

 Extract dominant color from images.
 Generate Insight based on product color
 Pain Point:
 They were having high system usage.
 Avg mem. consumption was more than 70%
PROCESSING IMAGES WITH GO

Existing architecture
Modified architecture

 Serves more request
 Server usage never exceeds more than 50-60%
 Easier deployment
 Only golang standard library used -
97747

EMBD
http://www.raspberrypi.org

 Requirement
 Static binary.
 Efficient Code.
 Zero latency(real time)
 Cross Platform
 Initial choice was Python (Dropped – Restricted library)
 EMBD a package entirely written in Go
 High Performance.

LIST OF COMPANIES USING GO IN
PRODUCTION

 Golang’s 3 hours online jumpstart course
 http://tour.golang.org/welcome/1
 Testimonials
 https://sendgrid.com/blog/convince-company-go-golang/
 http://www.scriptrock.com/blog/our-experience-with-golang
 Further reading
 http://golang.org/doc/
 https://talks.golang.org/
 http://thenewstack.io/a-closer-look-at-golang-from-an-architects-
perspective/
QUICK LINKS

Scaling applications with go

More Related Content

What's hot

Similar to Scaling applications with go

Recently uploaded

Scaling applications with go