Uber Technology Stack

-


The Content of this blog post is based on: https://eng.uber.com/tech-stack-part-one/

Almost everyone in the western world is familiar with the app called Uber. Uber is without doubt the most popular service of reliable transportation. By focusing entirely on mobile platforms they are a so called next generation company, which takes advantages of how a simple application build up around a strong inovative idea can make an impact on millions of people around the world. However a good idea and a great UI can't stand alone, especially not when dealing with such a large user base. In this blog post we will try break down the technology stack used in Uber.

As Ubers user base has grown with such a rapid speed, they now face similar global scaling problems as some of the most successful software companies. The infrastructure and storage of Uber has therefore changed since their first launch and will most likely conteniue to do so as new solutions is found to optimize todays standards.

Programming Languages

At Ubers lower level architecture several programming languages is being used. The engineers primarily write in Python, Node.js, Go and Java as of today, however these four were not all used in the early days of Uber. When Uber first started the two main languages were Node.js and Python.

Java and Go were adopted for high performance reasons, especially Java which takes advantages of the open source eco System and is used for integration where as Go is primarily for effecientcy.

Inferstructure and Storage

Do to the size of Ubers user base and their many geographical locations Uber can't rely on one single data center alone. The Uber business therefore runs on a hybrid cloud model, using a mix of cloud providers and multiple data centers. Especially the many data centers and there locations are key to the success of a app like Uber, as if one data center fails another one takes it place. The different data centers is assigned to cities based on their distance to one another.

Uber Architecture, picture is from: https://eng.uber.com/tech-stack-part-one/

Uber has since 2014 focused on Schemaless which were build in-house on top of MySQL. The choice of a In-House Schemaless builded solution came as they would run out of avaliable space for storing data using PostgreSQL before the end of 2014. For their new solution Uber was in need of the following:
(The full list and In-depth description of the criteria below can be found at: https://eng.uber.com/schemaless-part-one/)

  • We needed write availability
  • We needed secondary indexes
  • We needed operation trust in the system, as it contains mission-critical trip data
  • We needed a way of notifying downstream dependencies
Based on the above criteria they looked into already known solutions such as Cassandra, Riak and MongoDB. However none of these forfilled all criteria and as time was a pressing concern it was decided to build a In-House solution. The goal is to have Schemaless replacing individual MySQL and PostgreSQL in the future, which is still currently in use along with the Schemaless solution and can be seen in the Uber Atchitecture model above.

Routing and Services

Uber is using a Service-oriented Architecture (SOA) and Uber rely on its many services to be able to communicate effeciently with each other. To acheive such they have used a combination of HAProxy which is a free and open source load balancer and HyperBahn which is a open source solution build In-House at Uber. SOA is a popular architecture used by larger online applications which is spread across many different platforms as the same functionality has to be provided and such can only be achived effeciently using services over a network with a centrelized backend instead of a build in back-end supporting each platform independently.


For more

For more see https://vinsloev.com/ or visit our YouTube channel Vinsloev on YouTube


Kommentarer

Send en kommentar

Populære opslag fra denne blog

Artificial Intelligence - How it could lead to a greener and healthier world

-
Billede
This article is inspired by:  https://blogs.ei.columbia.edu/2018/06/05/artificial-intelligence-climate-environment/ One of the most discussed topics of this decade is without doubt global warming and how the climate changes are impacting our way of life. The general tempature is rising and the weather is getting more extreme. Its therefore vital that we find the most effecient solutions within the next decades if we are to save the next generations from living with the mistakes of the past. Many solutions do already exists to decrease the overall consumption of oil and gas, however most of these are either very expensive to implement or require us to rethink our daily habits. It's therefore important that we are as effecient as possible when it comes to managing the impact of climate changes and implementing countering solutions. To achieve such a new tool has been presented which we are only starting the see the real possibilties of namly Artifical Intelligence.  To discu
Læs mere

Introduction to Neural Networks - Part 1

-
Billede
Why should you learn Machine Learning? It’s no secret that machine learning and AI is one of the most popular buzzwords of this decade, but is the hype justified? Lets take a closer look at what exactly we should expected from Machine Learning and AI followed by a closer look at what exactly Neural Networks really is and why they have become so popular. Even though Machine Learning and the algorithms used in todays applications was introduced several decades ago, it's first within the last 10-15 years that we have started to see the real benefit of these. The reason why the area of Machine learning has seen such a growth can be boiled down to the data available. The amount of data we are able to collect nowadays w ith the introduction of the internet, Social media platforms and Internet of things just to mention some..    has lead to a revolution of the Machine Learning field, as these algorithms are in need of large datasets to achieve high accuracy. Examples of areas where we
Læs mere