in Blog

November 08, 2021

Building a data lake on cloud (AWS, Azure, GCP)


Artur Haponik

CEO & Co-Founder

Reading time:

6 minutes

On our blog, we frequently talk about data storage. And the truth is, for the vast majority of AI-related purposes, you need a data warehouse implementation. This solution allows you to store structured and organized data so that it can be easily used for future purposes. But sometimes, you need something more flexible. You need a solution that allows you to store various forms of data, even unstructured ones. And this is where data lakes step into the game and save the day. Today, we’re going to talk about building a data lake on the cloud. We will also take a look at three common solutions–Amazon AWS, Microsoft Azure, and GCP.

Nowadays, building a data lake on the cloud is your best bet when you’re looking for a decent data lake solution. Why? Well, data lakes on the cloud are secure, relatively easy to set up, and more affordable than the traditional on-premises option. However, before we switch to the analysis of the three most common data lake on cloud solutions, let’s talk some more about what a data lake actually is.

Read more about Data Lake Architecture

Building a data lake on the cloud – what are data lakes?

Shortly put, data lakes, just like data warehouses, are storage repositories. However, there are a couple of differences that you have to be aware of. Firstly, because data lakes allow you to store diverse types of data (structured, semi-structured, unstructured), there is no need to transform your data before it can be loaded into a data lake.


As its name indicates, here, you can store every type and kind of data in every format, just like a normal lake “stores” water, plants, sand, stones, wood, and fish. However, the extendible nature of a data lake can be dangerous. When unsupervised and not maintained properly, a data lake can turn into something called a data swamp.

And again, this hydrological metaphor is well-grounded. Data swamps are disorganized, messy, and, therefore, useless when you’re thinking about business-related usage.

It might be also interesting for you: Delta Lake on Databricks – Reliable Data Lakes at Scale 

Benefits of a data lake on the cloud

When building a data lake on the cloud, you can easily store big data in its raw, untransformed format. You don’t have to set up and invest in costly IT infrastructure. Everything is stored neatly in the cloud. Data lakes enable you to store data for the future, maybe even not yet determined purposes and applications. Mind you, due to this variegated nature of files and formats within a data lake, this type of storage repository usually requires more storage room than a data warehouse.

On the other hand, it’s quicker to manipulate, update, and access a data lake.

data warehouse

Additionally, we should mention that the data lake eliminates the need for the data silos, as data stored within them is centralized. As you may know, a data silo is also a repository of data, but its primary feature is that it is isolated from other repositories. It is usually managed exclusively by one department or office, which can become a challenge when the need to exchange information between departments happens.

With a data lake on the cloud, you can access, gather, and manage all data within one organization, no matter where it’s located and what it is for. And because the whole thing is located in the cloud, there are no problems with accessing it at any given time. All you need is a stable internet connection.

Now that you know what data lakes are all about and what are their major benefits let’s take a look at your possible options. You can generally opt for many data lakes on the cloud, but three services come to the forefront. And these are:

  • Amazon AWS
  • Microsoft Azure
  • Google Cloud Platform (GCP)

What do you need to know about these three services?

Building a data lake on the cloud: Available solutions

Amazon AWS

When you decide to opt for Amazon AWS, you get access to a broader AWS ecosystem that comprises many additional tools and features, e.g., AWS Lambda microservices (functions), Amazon OpenSearch Service (successor to Amazon Elasticsearch Service) for robust search capabilities, Amazon Cognito for user authentication, AWS Glue for data transformation, and Amazon Athena for data analysis[1].

Amazon facilitates the whole process of building a data lake on the cloud and adjusting it to your needs. They automatically configure the core AWS services allowing you to tag, search, share, transform, analyze, and govern specific subsets of data. The AWS solution deploys a console that users can access to search and browse available datasets. Here, you will find additional details.

And here’s how AWS architecture looks like:

AWS Architecture

source: AWS Amazon

Microsoft Azure

What does Amazon’s major competitor have in store for data lake users? Their service is called Azure Data Lake, and it includes all the capabilities required to make it easy for developers, data scientists, and analysts to store. Microsoft boasts to remove all the complexities of ingesting and storing data while making it faster to get up and running with batch, streaming, and interactive analytics[2].

Azure freely integrates with data warehouses, so you can make the most of both worlds. Microsoft’s data lakes are secure, scalable, and built to the open HDFS standard. As a result, there are no limits to the size of data and the ability to run parallel analytics. Take a look at Azure’s scheme:

data lake diagram

Source: com


Finally, we have Google’s option:


Google Cloud enables you to migrate your Apache Spark and Hadoop-based data lakes to their cloud service. What are the benefits of Google’s solution? For starters, migration can happen very quickly.

According to Google, you can govern purpose-built data and analytic open-source software clusters such as Apache Spark in as little as 90 seconds. Secondly, Google Data lakes can be teamed with Apache Spark, BigQuery, AI Platform Notebooks, GPUs, and other analytics accelerators. And lastly, Google brags their solution is up to 57% cheaper than an on-premises Hadoop data lake.

If you want to know more about GCP’s solution, go here. And this is how GCP’s architecture looks like:

data lake architecture



And don’t forget, if you need help with data engineering services in your company, we will gladly help you pick and implement the best data management solution, whether it’s a data warehouse or a data lake. Drop us a line for details!


[1] Data Lake on AWS. URL: Accessed Nov 2, 2021.
[2] Data Lake. URL: Accessed Nov 2, 2021.


Data Engineering