Skip to main content

Introduction to Data Lakes

Data lakes provide a complete and authoritative data store that can power data analytics, business intelligence and machine learning

datalakes

Introduction to data lakes

What is a data lake?

A data lake is a central location that holds a large amount of data in its native, raw format. Compared to a hierarchical data warehouse, which stores data in files or folders, a data lake uses a flat architecture and object storage to store the data.‍ Object storage stores data with metadata tags and a unique identifier, which makes it easier to locate and retrieve data across regions, and improves performance. By leveraging inexpensive object storage and open formats, data lakes enable many applications to take advantage of the data.

Data lakes were developed in response to the limitations of data warehouses. While data warehouses provide businesses with highly performant and scalable analytics, they are expensive and proprietary and can't handle the modern use cases most companies are looking to address. Data lakes are often used to consolidate all of an organization’s data in a single, central location, where it can be saved “as is,” without the need to impose a schema (i.e., a formal structure for how the data is organized) up front like a data warehouse does. Data in all stages of the refinement process can be stored in a data lake: raw data can be ingested and stored right alongside an organization’s structured, tabular data sources (like database tables), as well as intermediate data tables generated in the process of refining raw data. Unlike most databases and data warehouses, data lakes can process all data types — including unstructured and semi-structured data like images, video, audio and documents — which are critical for today’s machine learning and advanced analytics use cases.

Why would you use a data lake?

First and foremost, data lakes are open format, so users avoid lock-in to a proprietary system like a data warehouse, which has become increasingly important in modern data architectures. Data lakes are also highly durable and low cost, because of their ability to scale and leverage object storage. Additionally, advanced analytics and machine learning on unstructured data are some of the most strategic priorities for enterprises today. The unique ability to ingest raw data in a variety of formats (structured, unstructured, semi-structured), along with the other benefits mentioned, makes a data lake the clear choice for data storage.

When properly architected, data lakes enable the ability to:

power data science

Power data science and machine learning

Data lakes allow you to transform raw data into structured data that is ready for SQL analytics, data science and machine learning with low latency. Raw data can be retained indefinitely at low cost for future use in machine learning and analytics.

centralize

Centralize, consolidate and catalogue your data

A centralized data lake eliminates problems with data silos (like data duplication, multiple security policies and difficulty with collaboration), offering downstream users a single place to look for all sources of data.

quickly

Quickly and seamlessly integrate diverse data sources and formats

Any and all data types can be collected and retained indefinitely in a data lake, including batch and streaming data, video, image, binary files and more. And since the data lake provides a landing zone for new data, it is always up to date.

democratize

Democratize your data by offering users self-service tools

Data lakes are incredibly flexible, enabling users with completely different skills, tools and languages to perform different analytics tasks all at once.

Data lake challenges

Despite their pros, many of the promises of data lakes have not been realized due to the lack of some critical features: no support for transactions, no enforcement of data quality or governance, and poor performance optimizations. As a result, most of the data lakes in the enterprise have become data swamps.

reliability-issues-icon

Reliability issues

Without the proper tools in place, data lakes can suffer from data reliability issues that make it difficult for data scientists and analysts to reason about the data. These issues can stem from difficulty combining batch and streaming data, data corruption and other factors.

slow-performance-icon

Slow performance

As the size of the data in a data lake increases, the performance of traditional query engines has traditionally gotten slower. Some of the bottlenecks include metadata management, improper data partitioning and others.

security-features-icon

Lack of security features

Data lakes are hard to properly secure and govern due to the lack of visibility and ability to delete or update data. These limitations make it very difficult to meet the requirements of regulatory bodies.

For these reasons, a traditional data lake on its own is not sufficient to meet the needs of businesses looking to innovate, which is why businesses often operate in complex architectures, with data siloed away in different storage systems: data warehouses, databases and other storage systems across the enterprise. Simplifying that architecture by unifying all your data in a data lake is the first step for companies that aspire to harness the power of machine learning and data analytics to win in the next decade.

How a lakehouse solves those challenges

The answer to the challenges of data lakes is the lakehouse, which adds a transactional storage layer on top. A lakehouse that uses similar data structures and data management features as those in a data warehouse but instead runs them directly on cloud data lakes. Ultimately, a lakehouse allows traditional analytics, data science and machine learning to coexist in the same system, all in an open format.

A lakehouse enables a wide range of new use cases for cross-functional enterprise-scale analytics, BI and machine learning projects that can unlock massive business value. Data analysts can harvest rich insights by querying the data lake using SQL, data scientists can join and enrich data sets to generate ML models with ever greater accuracy, data engineers can build automated ETL pipelines, and business intelligence analysts can create visual dashboards and reporting tools faster and easier than before. These use cases can all be performed on the data lake simultaneously, without lifting and shifting the data, even while new data is streaming in.

Building a lakehouse with Delta Lake

To build a successful lakehouse, organizations have turned to Delta Lake, an open format data management and governance layer that combines the best of both data lakes and data warehouses. Across industries, enterprises are leveraging Delta Lake to power collaboration by providing a reliable, single source of truth. By delivering quality, reliability, security and performance on your data lake — for both streaming and batch operations — Delta Lake eliminates data silos and makes analytics accessible across the enterprise. With Delta Lake, customers can build a cost-efficient, highly scalable lakehouse that eliminates data silos and provides self-serving analytics to end users.

Learn more about Delta Lake

marketecture

Data lakes vs. data lakehouses vs. data warehouses

 Data lakeData lakehouseData warehouse
Types of dataAll types: Structured data, semi-structured data, unstructured (raw) dataAll types: Structured data, semi-structured data, unstructured (raw) dataStructured data only
Cost$$$$$
FormatOpen formatOpen formatClosed, proprietary format
ScalabilityScales to hold any amount of data at low cost, regardless of typeScales to hold any amount of data at low cost, regardless of typeScaling up becomes exponentially more expensive due to vendor costs
Intended usersLimited: Data scientistsUnified: Data analysts, data scientists, machine learning engineersLimited: Data analysts
ReliabilityLow quality, data swampHigh quality, reliable dataHigh quality, reliable data
Ease of useDifficult: Exploring large amounts of raw data can be difficult without tools to organize and catalog the dataSimple: Provides simplicity and structure of a data warehouse with the broader use cases of a data lakeSimple: Structure of a data warehouse enables users to quickly and easily access data for reporting and analytics
PerformancePoorHighHigh

Learn more about common data lake challenges

Lakehouse best practices

landing-zone-icon

Use the data lake as a landing zone for all of your data

Save all of your data into your data lake without transforming or aggregating it to preserve it for machine learning and data lineage purposes.

mask-data-icon

Mask data containing private information before it enters your data lake

Personally identifiable information (PII) must be pseudonymized in order to comply with GDPR and to ensure that it can be saved indefinitely.

secure-data-icon

Secure your data lake with role- and view-based access controls

Adding view-based ACLs (access control levels) enables more precise tuning and control over the security of your data lake than role-based controls alone.

build-reliability-icon

Build reliability and performance into your data lake by using Delta Lake

The nature of big data has made it difficult to offer the same level of reliability and performance available with databases until now. Delta Lake brings these important features to data lakes.

catalog-data-icon

Catalog the data in your data lake

Use data catalog and metadata management tools at the point of ingestion to enable self-service data science and analytics.

Shell has been undergoing a digital transformation as part of our ambition to deliver more and cleaner energy solutions. As part of this, we have been investing heavily in our data lake architecture. Our ambition has been to enable our data teams to rapidly query our massive data sets in the simplest possible way. The ability to execute rapid queries on petabyte scale data sets using standard BI tools is a game changer for us.

— Dan Jeavons, GM Data Science, Shell