What is Databricks?

Databricks claims to be a cloud data platform that integrates the whole data ecosystem end to end. Their purpose is to enable organizations to equip their employees, customers, operational workflows, products, and services with timely and trusted data. Originally designed for data engineers, Databricks has expanded to focus on self-service capabilities for Data Scientists.

What is Datagran?

Datagran is an integrated, self-service cloud data platform that enables data teams and business teams to work better together - streamlining the path from data to decisions. Our mission is to speed up the process of data workflows without the need of engineering resources, grow data adoption and company-wide trust in data by empowering business teams to access and analyze data, whenever and however they need it. Our robust modeling layer is also user-friendly enough for data-savvy business users to participate.

TL;DR

A good cloud data platform should be accessible and beneficial to everyone. These factors are vital for business teams to make informed, timely decisions. As such, things like learning curve, lengthy training time, implementation time, slow performance and load time, and limitations for non-technical business users are red flags.

While Databricks is a robust, powerful cloud data solution that equally plays in the self-service space, there are certain downsides that negatively impact the ongoing accessibility of data for all users. In this comparison, we’re considering the perspective of a large tech and data saavy company that’s mature in their data journey, with teams embedded in business units. The main downsides include data engineering resources, slowness, overall learning curve and complexity, and cost.

Datagran, on the other hand, is fast to implement, learn, and load, doesn’t require any engineering resources. Plus, we’re significantly cheaper with free viewer licenses.

At a glance

Feature	Databricks	Datagran
Pricing	Pricing is usage based and meassured via a propietary metric called DUs. Pricing of DUs varie per product. For example Databricks Notebook, Serverless SQL and Jobs are each priced separatly. For example, a customer with 100 jobs at $0.10 per DBU when each job costs aroun 250 DBUs might end up paying around $10k per month only on job related pricing.	Pricing is usage based and meassured via a propietary metric called Data Units. Datagran does not have differential pricing between products. Each Data Unit is $1 dollar and translates into one hoy of machine time. On average our enterprise clients spend around $300 in Data Units per month which translates in 300 hours of machine time.
Other than that all of our plans start at free forever and Enterprise at $5,000 per month and include 150 data units, 5 users and 50 million data rows.
Learning curve	Steep - must have Data Engineers to setup infraestructure. Heavy coding requirements in modeling, exploration, and iteration.	Easy - Datagran provides code and low-code tools. It even provides an AI code assistant that understands plain english. No engineering required.
Average roll-out time	6-12 months.	2 weeks.
Data connectors	Not included	Included
Destinations	Not included	Provides multiples destinations like for example Intercom, Salesforce, Snowflake and many more.
Speed	Databricks Notebooks runs with all the customers data which makes the process slow.	Datagran is optimized for big data offering several options for working with sample data sets.
Iterface	In-browser; no desktop install. Mobile version is limited.	In-browser; no desktop install. Mobile version is limited.
Software updates	Updates are completed per customer and not globally for all customers, which can be a pain when migrating your deployment.	Datagran updates all customer deployments at all times and makes sure nothing is broken and no one is left behind.
User permission management	Yes	Yes
Modeling layer	Jupyter notebook like coding environment running specifically on Spark.	Hosted VSCode IDE that runs with or without Spark. Datagran also provides multiple low code tools as well as an AI code assistant.
Customer Service	Expect long delays in terms of responsiveness and resolution	Highly reactive with SLAs in place; quick problem-solving

Cautions for Databricks:

There’s a super steep learning curve to using Databricks, which will slow down the total time you need to get up and running, and for Databricks to start adding value.
- Engineering Resources: Databricks is highly technical and requires Data Engineers to setup and scale machines to run models. Any business unit wanting to deploy models will need assistant for this type of specialists. Also, Databricks is heavy in code, which will not be appropiate for business profiles.
- Roll-out time: 6-18 months including training and implementation.
- Set-up: Once training is complete, it could take up to 30 minutes to run an analysis.
Databricks is complex, and is more suited for core data teams serving the organization horizontally.
- If your first data hires are data analysts or data scientists (as opposed to data engineers who can code), there will be a barrier to entry.
- Oftentimes, the complexity requires companies leverage a consulting firm or third party to help with Databricks. In this case, you don’t own the tool and you won’t get the ongoing data visibility with the consulting firm in the middle.
- Maintenance requires a lot of time and resources on an ongoing basis
Slow load time. This drags out the amount of time it takes to get data out to people who need it, which means critical business decisions are delayed, or not derived from data.
- Takes a long time to load large data sets
- Delay in data reflection in reporting, sometimes up to 24h
- When the load time is too lengthy, business users try to find their own workarounds like spreadsheets to make data-driven decisions on their own.
Expensive with enterprise-level pricing. This may be suitable for horizontal core data teams, but most business units and mid-market companies would get a better bang for their buck with a more affordable solution, that is more suited to their needs and where they’re at in their data journey.

Datagran benefits

Whaly is much simpler and faster to learn and set up. It doesn’t demand a massive amount of time, resources, or upfront investment to start getting value from your data.
- No language requirements.
- Faster roll-out time by 5 months minimum (only 1-3 weeks)
- Once trained, it takes 1 hour to get explorations set up.
- Implementation takes under an hour
Datagran bridges the gap between data teams and business teams, providing them with the same framework of communication. Datagran enables collaboration and streamlines the path from data to decisions.
Datagran adapts to your current data team size / status on data journey - not the other way around.