Azure Synapse vs. Databricks vs. Snowflake: A Comprehensive Comparison for Data Workloads

[@knakamo1]

This was extremely helpful! I've struggled with streamlining why you need different solution to non-technical leadership and this helped immensely. Thank you!

[@ChrisSeferlis]

Glad it helped! Let me know what other topics I can help with!

[@svdfxd]

Great video Chris...will look out for latest video

[@revathikanduri]

Thank you, this was very helpful. I was trying to understand where Snowflake fits better and your video was very comprehensive to understand. Would love to see a similar take with AWS and snowflake

[@ChrisSeferlis]

There are some exclusive features with Databricks on Azure like Active Directory and unified billing, however, from a pure compute and function standpoint the experience would be similar. AWS has nothing competitive with Synapse! Choose Azure! ;)

[@dudilevi1]

Just a little correction, 2 of Snowflake founders worked in Oracle not microsoft en.m.wikipedia.org/wiki/Snowflake_Inc.

[@ChrisSeferlis]

Thanks for the distinction

[@pic101]

Fabulous rundown, very well described. Well worth watching.

[@ChrisSeferlis]

Glad you enjoyed it!

[@jpetrides]

Great video! Curious as to why Snowflake is so popular and has so much market when it lacks features. It makes me wonder how necessary all those extra features are. For example, ML workloads are usually done by a very small team - even in large companies. Maybe snowflake itself doesn’t have ETL, but there are plenty of best in breed options out there that snowflake customers use. Something has to explain Snowflakes popularity and growth.

[@ChrisSeferlis]

They certainly have a great marketing team, a clean interface, and some nuanced features some customers like. The biggest challenge is the way they scale by adding compute nodes and dramatically increase costs as a result. Customers get locked into a contract, and there are no outs. Synapse offers a more complete solution for all different types of data workloads. Snowflake is more targeted and specialized.

[@rohitk291]

ask customers who have used Synapse and Snowflake both. I feel Snowflake just works under any circumstances where as Synapse doesnt. Synapse is not yet matured and cant provide the elasiticity and tons of other features provided by Snowflake. if you check for customer review on different community pages , you will see so many success stories with Snowflake ,whereas with Synapse people are still struggling.I will vote for Snowflake anyday as my DWH of choice for OLAP and complex queries with its unique clustered warehouse feature and discard Synapse considering its Another redshift or vantage kind of product. Use both the two product with 100 TB of data with 500 concurrent users, Synapse will struggle whereas Snowflake will fly at ease.

[@ChrisSeferlis]

Everyone is certainly entitled to their opinion, and there are certainly situations where there may be better performance on Snowflake as that’s not really the intent of this video. Snowflake is a very good solution for certain data workloads, and in this video I’m highlighting a variety of data workloads, some where Snowflake might not be best solution. Always compare solutions based on your overall requirements, budget, features, and costs.

[@annfang9095]

Great content in an easy to follow format. Thanks for sharing Chris!

[@ChrisSeferlis]

Thank you

[@krasuri]

Great content. Neatly explained about the similarities and differences

[@ChrisSeferlis]

Glad you enjoyed it!

[@elainee7116]

Love this content, thanks for sharing!

[@ChrisSeferlis]

Glad you enjoy it!

[@touchwithbabu]

Good content Chris. thanks

[@naveenkrishnachittineni5348]

Great explanation...

[@ivani3237]

15:00 Snowflake do have external tables to files witch allow you same as in Synapse

[@danhorus]

In regards to connecting Synapse and Azure Machine Learning, I'm afraid there is still no support for a direct connection in the Datastores. Much like with Snowflake, it seems you still need to extract the data to a Data Lake in order to run experiments on it.

[@ChrisSeferlis]

Thanks for the feedback Daniel. Being able to connect in directly is in preview, you are correct, however, please note that a major difference is that with Snowflake you'd have to pay an egress charge to pull the data out, with Synapse you would not, and that can get very pricey.

[@DintlP]

hi Chris ... nice detail video... do you happen to know the backend language how this is coded ... Synapse is C,C++ , DB is Scala and new Photon will be C++ but what is Snowflake ? just curious if you know that info

[@ChrisSeferlis]

Not sure, but will ask around

[@BenDover-ie1ud]

Very informative, thank you very much.

[@ChrisSeferlis]

Happy to hear!

[@eladjohn491]

5:34 "NOTEBOOKS"

[@ChrisSeferlis]

Did I use the wrong term?

[@eladjohn491]

@@ChrisSeferlis The term was correct, I was just referring to how the word was recorded over something else, which made the sentence sound a bit funny :)

[@freebirdyaoyao]

Great share , very useful ! Many Thanks.

[@ChrisSeferlis]

Glad it was helpful!

[@srbasha74]

Very helpful. Thanks One question, Azure Synapse offers data bricks as a linked service. Would that be the best of both worlds situation or would lead into a messed up solution with limitations kicking in from both sides

[@ChrisSeferlis]

Databricks offers a more robust Data Science platform experience, and they tout some performance over Synapse Spark engines, although I’ve never seen a side by side because they still seem to operate in different spaces. Much of your decision should come from the skill set you’re working with, and the ecosystems you’re used to. I’m yet to see a use case that can’t be done on one platform or the other, as most of the time it comes down to efficiency and cost. The only way to know that about your specific use case would be to test on both platforms.

[@srbasha74]

@@ChrisSeferlis Thank you

[@vladimirryabtsev6184]

What was that at 5:35? :-0

[@bolbol8043]

Great content, thanks!

[@KimmoHintikka]

Out of curiosity would this change a lot in case of doing this video 2022?

[@ChrisSeferlis]

I would not say “a lot” but certainly changes have been made to each of the platforms, and the strengths of each is still in line.

[@leejenny2826]

HI ! Thank you for showing this video. Can you tell me what is the difference between 'Databricks' and 'Sumo logic' ? Honestly for me this is quite hard to understand. I think they are both data analytic system. aren't they? Please tell me this. Thx again!!

[@ChrisSeferlis]

Databricks is a data processing engine for a variety of workloads including data engineering, data analytics, and data science. Sumo Logic is used to collect logging data from various sources to detect anomalies in your network. Databricks could be used for this, but there are many “canned” options that are built specifically for that job.

[@leejenny2826]

@@ChrisSeferlis OMG...Thanks a lot....!!! I totally understand about this. For these days, I am trying to understanding what the 'Databricks' is. Your video was best resource that I've ever seen. Based on your knowledge, Could you tell me that why do you think 'Databricks' is popular with people? Can you summarize Databrick's strengths in one sentence? Thank you very much !!!!

[@ckchouk]

Any pricing comparison between all 3 platforms please?

[@ChrisSeferlis]

Honestly, there are too many variables to each situation. Search for a TPC benchmark report comparing all of the data warehousing engines, and that’s about the best you’ll get for cost comparisons.

[@Prashanth-yj6qx]

Commening..to get more videos

[@ChrisSeferlis]

Thanks, more coming soon!