Amazon SageMaker vs Databricks Unified Analytics Platform

Likelihood to Recommend

Well suited scenarios:

For quick POC of ML and DL.

To train a model on a large dataset using multiple servers.

To host a model to be used by multiple applications.

Less appropriate scenarios:

For data analysis tasks.

For a data scientist who has less of a programming background.

Read full review

Verified User

Employee in Research & Development

Computer Software Company, 501-1000 employees

View all 5 answers on this topic

DB generally fits 95% of what you need to do
Primarily the ability to transform data and or do ad-hoc DS work

Read full review

Verified User

Director in Information Technology

Financial Services Company, 201-500 employees

View all 5 answers on this topic

Feature Rating Comparison

Amazon SageMaker

—

Databricks Unified Analytics Platform

8.3

Connect to Multiple Data Sources

—

9.0

Extend Existing Data Sources

—

9.0

Automatic Data Format Detection

—

7.0

Amazon SageMaker

—

Databricks Unified Analytics Platform

6.0

Visualization

—

6.0

Interactive Data Analysis

—

6.0

Amazon SageMaker

—

Databricks Unified Analytics Platform

8.0

Interactive Data Cleaning and Enrichment

—

8.0

Data Transformations

—

9.0

Data Encryption

—

7.0

Built-in Processors

—

8.0

Amazon SageMaker

—

Databricks Unified Analytics Platform

8.3

Multiple Model Development Languages and Tools

—

9.0

Automated Machine Learning

—

8.0

Single platform for multiple model development

—

9.0

Self-Service Model Delivery

—

7.0

Amazon SageMaker

—

Databricks Unified Analytics Platform

7.5

Flexible Model Publishing Options

—

7.0

Security, Governance, and Cost Controls

—

8.0

Pros

SageMaker is useful as a managed Jupyter notebook server. Using the notebook instances' IAM roles to grant access to private S3 buckets and other AWS resources is great. Using SageMaker's lifecycle scripts and AWS Secrets Manager to inject connection strings and other secrets is great.
SageMaker is good at serving models. The interface it provides is often clunky, but a managed, auto-scaling model server is powerful.
SageMaker is opinionated about versioning machine learning models and useful if you agree with its opinions.

Read full review

Gavin Hackeling

Data Scientist

Wonder (AskWonder.com)Research, 11-50 employees

View all 5 answers on this topic

Extremely Flexible in Data Scenarios
Fantastic Performance
DB is always updating the system so we can have latest features.

Read full review

Verified User

Director in Information Technology

Financial Services Company, 201-500 employees

View all 5 answers on this topic

Cons

SageMaker does not allow you to schedule training jobs.
SageMaker does not provide a mechanism for easily tracking metrics logged during training.
We often fit feature extraction and model pipelines. We can inject the model artifacts into AWS-provided containers, but we cannot inject the feature extractors. We could provide our own container to SageMaker instead, but this is tantamount to serving the model ourselves.

Read full review

Gavin Hackeling

Data Scientist

Wonder (AskWonder.com)Research, 11-50 employees

View all 5 answers on this topic

The navigation through which one would create a workspace is a bit confusing at first. It takes a couple minutes to figure out how to create a folder and upload files since it is not the same as traditional file systems such as box.com
Also, when you create a table, if you forgot to copy the link where the table is stored, it is hard to relocate it. Most of the time I would have to delete the table and re-created.

Read full review

Ann Le

Freelance Translator

ZOO Digital Group plcEntertainment, 501-1000 employees

View all 5 answers on this topic

Usability

No score

No answers yet

No answers on this topic

Ask the community to sound off

Databricks Unified Analytics Platform 9.0

Based on 1 answer

This has been very useful in my organization for shared notebooks, integrated data pipeline automation and data sources integrations. Integration with AWS is seamless. Non tech users can easily learn how to use Databricks. You can have your company LDAP connect to it for login based access controls to some extent

Read full review

Verified User

Team Lead in Engineering

Financial Services Company, 10,001+ employees

View all 1 answers on this topic

Alternatives Considered

Amazon SageMaker comes with other supportive services like S3, SQS, and a vast variety of servers on EC2. It's very comfortable to manage the process and also support the end application by one click hosting option. Also, it charges on the base of what you use and how long you use it, so it becomes less costly compared to others.

Read full review

Verified User

Employee in Research & Development

Computer Software Company, 501-1000 employees

View all 5 answers on this topic

I also use Microsoft Azure Machine Learning in parallel with Databricks. They use different file formats which teach me to be flexible and able to write different programs. They are equally useful to me and I would like to master both platforms for any future usage. I do prefer Databricks because it could be free if I decided to go with the Databricks Community Edition only.

Read full review

Ann Le

Freelance Translator

ZOO Digital Group plcEntertainment, 501-1000 employees

View all 5 answers on this topic

Return on Investment

We have been able to deliver data products more rapidly because we spend less time building data pipelines and model servers.
We can prototype more rapidly because it is easy to configure notebooks to access AWS resources.
For our use-cases, serving models is less expensive with SageMaker than bespoke servers.

Read full review

Gavin Hackeling

Data Scientist

Wonder (AskWonder.com)Research, 11-50 employees

View all 5 answers on this topic

Quick adoption of cloud services by end users
Cost is high

Read full review

Verified User

Director in Engineering

Financial Services Company, 10,001+ employees

View all 5 answers on this topic

Likelihood to Recommend