Databricks vs Palantir
Вставка
- Опубліковано 7 сер 2022
- I've learned a lot comparing these two platforms. My opinions have changed. You don't want to miss this head-to-head comparison between two industry titans!
docs.databricks.com/
databricks.com/blog/2017/05/1...
docs.databricks.com/clusters/...
docs.databricks.com/clusters/...
delta.io/
Foundry Docs: www.palantir.com/docs/
Foundry Use Cases: www.palantir.com/docs/foundry...
Foundry Developers: www.palantir.com/developers/
www.palantir.com/docs/foundry...
Synthesizer and unit tests: stackoverflow.com/questions/7...
ACID patent: www.freepatentsonline.com/895...
FTC Disclosure: I am working in partnership with Palantir as a Foundry ambassador! I have access to a free Foundry Stack.
Join the discussion: / discord - Наука та технологія
When Palantir haters bring me down over one quarterly release, codestrap brings me back up. Thanks again for your contributions, you are one of one out here.
PS continue to kill it with the intro tunes 😂
One of Palantir's issues with external perception is that because it is so shapeshifting and malleable, people tend to try to bucket it as a "vs" to other companies. What they fail to understand is that the holistic value proposition of Palantir is to provide a company end-to-end every level of the "Maslow's Hierarchy" for business and operational enlightenment, and these other companies all fit into small parts of that enlightenment triangle. E.G. Data lake providers, data warehouse providers, bi tools, ml creation tools, these are all individual components of the whole. Palantir can plug and extend into any of these, create a connective tissue between them all, and wrap it all up with operationalization. This is why AWS, GCP AND MSFT all partner with PLTR.
My thoughts exactly.
Except that databricks is also partnered with aws, gcp and azure. I dont care much for databricks, but this logic is flawed
@@abramspamers6456 Actually no, your comment is irrelevant because the partnership aspect was just to show that they're not competitors as people think they are. Regarding databricks also being partnered with them, that's irrelevant. They are a data platform so obviously it makes sense if they partner with hyperscalers. But that's it, they stop at data platform in the value chain. Palantor goes way beyond data platforming in the value chain of data --> operational excellence
@@rayz6307 " Palantir can plug and extend into any of these, create a connective tissue between them all, and wrap it all up with operationalization. This is why AWS, GCP AND MSFT all partner with PLTR."
So when you said "this is why" you were not referring back to the operationalization sentence right before it? Seems like a dodge
@@codestrap8031does Microsoft fabric, now mean Microsoft has created a competitor to foundry
I am grateful that you are working with Palantir. You know the technical applications, terms, meaning, etc. Also, you tell Palantir what they should do to improve, whether they may like it or not, and that is awesome. I appreciate the hard work!
Thank you!
Thank you, you are the reason I believe in Palantir. As great as Amit and Tom and Co are, your contribution is more important because you show us the product. Look at it like this: Teslas sell themselves because owners blab about them until their friends go for a ride, then they're hooked ... all because the product completely rocks. Well, we can't step into Foundry like we can a Tesla. You allow us to see it all better. That's invaluable to the UA-cam Palantir community
Thank you!
This video is super valuable. Thank you yet again for sharing with the community such an insightful summarised technical review. While literally everyone else is focusing on financials and projections, thank you for keeping the attention on the technical capabilities and product offerings.
My pleasure
Thank You So Much! You have given many of us hope the PLTR will be able to hold its technology edge.
This was amazing! Thank you so much Codestrap
Thanks Codestrap, keep em' coming!
Thank you Codestrap for that outstanding comparison. Best Palantir video I have seen for a long long time 👏🏻
Thank you!
Excellent video. I am a BI/database developer, and was interested in learning Palantir tech in regards to possible BI work. Your video showed that I cannot download and learn the software like I can with Microsoft tools, Tableau, PowerBI, etc. But I could do this with Databricks. Also, my expertise would be useful/needed and appreciated with Databricks, whereas skilled devs don't seem so important in a Foundry environment, except maybe the initial setup of the system. Good to know. And like you said, until they make it easy for developers to access and learn their software, a lot of developer talent will skip them as a career option.
What a video, wow man!
Thank you for an awesome and detailed comparison!
You're welcome, and thanks for the kind words.
A lot of negativity out there today on the stock. Thank you for releasing this to gain some level headedness.
We needed some good news today magicman. Thank you.
You bet
Great info! Thanks for sharing, CodeStrap.
CodeStrap, you are the MAN!! Thank you!
You're welcome and thanks for watching!
thanks for the video codestrap. great insights. do you have any more info on the freemium tier? no mention of that in the pipeline.
Nope
Haven't watched yet but will later when I havd time. I wanted to ask if you caught the part about "10,000 developers" on the earnings call?
I did and wondered how they are calculating that. Maybe they meant users?
As a Software Developer / Architect your videos are GREAT! Just learning about the Palantir platform... so is it safe to say all development is done in their tools? How does it integrate with Azure and i can timaging not using VS Code for my coding...? This also sounds similar to a piece of software by Wonderware MMI back in the day.... You got a Subscribe from me... look forward to more videos on Palantir developer tools! thanks
Yea buddy! Thanks for the great analysis. Great to see ya! Hope you and the fam are doing well! Party On my friend!😃✌️😃✌️😃✌️
Thanks man, we are doing good. Same wishes to you and yours.
@@codestrap8031 👍👍
I don't want to talk stocks, and I know you can't, but from a pure business and tech perspective, how far would you say their engineering work, business/government/commercial relationships, and etc lead?
Thanks bunch for this comparison! How long would it take Databricks to catch up with pltr? If they do, do you think they will still have pricing advantage of today?
I don't think they are interested in catching up with them regarding feature parity. DBX's plan is to outsource missing features to third parties who can hook up DBX to their SaaS offerings. A good example of that is Looker, which provides advanced visualization. But that's more of the same systems integration problems we have already. DBX is very focused on tools data engineers and scientists will use. So very different businesses despite the overlap in the feature set.
In my experience, when trying to persuade folks to the benefit of Foundry, I focus on the ontology that I find so unique/beneficial of Foundry that is missing among otherwise similar technologies. Discussion of Foundry's native visualization and low-code functionality distracts from the core data integration/modeling capability that really drives its value. Though developing/deploying apps and visualizations in an environment natively integrated with purposely derived data sets is awesome, I find that the audience often says, "But I can build apps with Appian", or "We like Tableau for dashboards"... *Curls fist.*
You hit the nail on the head when you said volunteer will struggle with SMB’s and start ups.
As a CTO, I don’t want people with theoretical knowledge… I want people with subject matter knowledge to guide and advise when we’re talking about platform development
With access to a tech stack, devs Can go in and “garage level tinker “in creating valuable solutions only because they’re in their tinkering in playing with it, breaking things, innovating things and coming and showing me X is doable better this route.
Closed stacks are worthless to me as a partner at my company and the thought of coming out and paying 6 to 7 figures+ on hopium that it will provide competitive operational advantages is useless to my company.
Opening up access would allow me to direct developers to test bed and create micro solutions to then say this can scale up to kill off X system, or kill off X work, or create Autonation etc and come with actual proof ready case study relative to company and industry need.
Palantir can talk all they want… and heck, maybe even have an amazing product… but putting a giant pay wall in front of it only serves a specific audience.
Great for them if they want/care only for that… but will not lead them to being “What AWS was to developers” ten years ago as mentioned by Mr. Sankar.
We are on the same page for sure
Thanks codestrap
I wish they would give out free bumps to get people hooked. I know Palantir offers sub-modules as entries into the system, but they could do more to get people onboarded and hooked. They need to do a much better job making this all easily understood for the least knowledgeable customers.
Agreed.
Although Databricks is more affordable, when you take into account the specialized staffing needs, development schedule, ease of making future changes, and time to market, does Palantir offer a better value proposition?
Also, considering you have mentioned that 85% of these home grown initiatives fail, how does an integrated tool like Foundry compare to Databricks?
It certainly does, but thats less tangible than hard money. Also, a developer can't try the platform or get certified in it without working for a company that cann afford Foundry. That problem isn't going to be solved by a better value prop.
Databricks wins on all accounts easily
"In a dark place we find ourselves and a little more knowledge lights our way.” - Yoda
Thanks
great vid
Wait what are your thoughts on databricks and snowflake integrating live stream data to their platforms?
Dbx has had the best live streaming platform for a long time IMO
What do you think of C3 AI? Is it a good buy?
Don’t you think part of the reason for PLTRs SPACs was to develop products for smaller cos and solve the affordability and accessibility issues?
No. Companies have product teams and engineers to do this. No one buys SPACs to do product development.
@@codestrap8031 why do you suppose they purchased them? I doubt that it was for investment purposes only.
If I felt I had a revolutionary technology that could give a competitive edge to anyone that uses it I also would look to launch startups on it to disrupt legacy ways of doing business. Thats been one of the reasons I wanted to get my hands on Foundry. Alex from Ticker Symbol You also was like, "never mind investing in PLTR, can u launch a business on Foundry". I dont want to speak for PLTR, its just how I see it.
Affordability and accessibility is a big deal for companies. I work at a moderate size multi-hospital system and affordability and accessibility is the break point for them purchasing software. They want to see ROI before the purchase. Is it possible for PLTR to allow companies to integrate their software for free for several months and show cost savings before purchase? We looked at Lumedx for analytics and it would take 5 years and the hiring of multiple IT personnel to integrate their software with difficult to determine ROI so never purchased
I'm sure its possible, just not something they appear willing to do today
Brian: A great business model for small companies would be for a reseller to offer them access to Palantir modules as if they were departments of the reseller. Resellers could specialize in industries such as hospitals, oil & gas services, restaurants, hotels, architecture shops.
Alteryx also started with a high price model; $40,000 per seat as I remember. Once they dropped the cost to a fifth or even a tenth of that, their market penetration deepened.
CodeStrap: Thanks for not looking down your nose at low-code no-code software. The New Luddites can’t see past the Command Line Interface and are in for a shock when subject matter experts no longer have to interact with programmers day to day.
By the way, the label “non-technical” has always been a misnomer. Programmers use it to pejoratively tag people who just don’t use SQL, C, or whatever other computer language is current, but who might be masters of the most technical subjects in mechanical engineering, geology, or rocket science. No-code low-code apps are the great equalizer now.
@@shandor2522 interesting
We started to consider to combine existing palantir with databricks due to below reasons
1- Palantir doesn't do well with BI tools. It is good to dump parquet but disasterly slow over jdbc connections
2- Palantir pricing is not transparent and it makes it harder during finops era
3- Lack of community support
4- Lack of notebooks and speed of databricks seems faster
They really need to address these issues to not lose customers to rivals especially a big one partnered with the biggest on azure.
Did you check to make sure the bottleneck isn't the database? Reading millions of rows of data from a slow RDS source is always slow unless the DB is tuned for this. Also did you explore dynamic resource allocation to speed up your jobs? Most of the time the complaint is cost not performance. Like I said in the video I see DBX as a complimentary tool and not either or. There are use cases where it makes sense. But I wouldn't for a second think about DBX as any kind of replacement for ontological models, applications, workflows, and simulations. They also are behind Palantir in terms of operationalizing models and change management, two areas that often lead to project collapse.
Also Foundry has code workbooks that are managed via the version control system. These are essentially notebooks but better because they are natively versioned and can be modeled as a DAG.
@@codestrap8031 I thin you misread it. Problem is getting the data out via jdbc using Microstrategy or Power-BI where this is not an issue
They really need to give a faster way of consuming data out of palantir if palantir is going to be the final source of truth. If alternatives are faster than it is a battle lost.
@@codestrap8031 Code workbooks are good but not great as you can't replace it with a notebook where you can switch between runtimes (SQL/Pyspark) in the same notebook. SQL lovers don't really enjoy code workbooks.
I feel notebooks are coming to palantir but kinda late.
I would love to hear your thoughts around my other comment under the faster than rest video. Items 5-6-7
Don't get me wrong I'm in love with the product I just don't like to lose, normally very easy to win debates, if these stuff was addressed on time.
What's that song in the begining?
Fingerprint by the Mini Vandals
I like it! The video was good too 😆 Super informative man thanks.
How much money does PLTR need to acquire databricks? That move will be a deadly chess move for the industry, preaty much game over
Affordable and easy should be easy to fix, but it seems like they aren't that interested in it? So adversarial to the customer as a starting point.
Datadog vs PLTR or snow ❄️
Codestrap!!!
Emanuele was in DA HOUSE!✌️😃
You ask "who else does that" SAS Viya does all of these with code AND low/no code.
I wish PLTR offers free trial and pay as you go
The data sharing in both platforms is obsolete. Both could just use Snowflake’s data sharing which obviates downloads. In-database processing is the key, and patented, I believe.
First
ne meeeee
actually no.
Adam B was in DA HOUSE!✌️😃
databricks have ci/cd and integration to ci/cd and online IDE
Not when i made this video.
And their version of an IDE is workbooks and a basic pipeline editor.
I wish I would nominate you as the chief of sale department and pay you 4million cash per year after learning the SBC of the last earning report.
Technically I was first :)
Hahah, true
Kris, has your narrative changed about Databrick now that you've seen side by side comparison?
@@HDave2170 no. Databricks is executing. While Palantir tech is better, management needs a better grasp of software sales and go to market strategy.
Kp in DA HOUSE!✌️😃
databricks provide Multi-Tenant with Enterprise Edition
Cool. When did they roll that out?
Been there for a while, not sure when exactly to be honest
No Ontology means piss weak ML and A.I. operability. Does databricks have natural language processing for operational data analysis like Palantir just launched? The product is inferior and more expensive but perhaps more marketable.
I like your style Kyle, but I have to disagree with the conclusion. Databricks created Spark, ML Flow, and Delta. Their founders are legitimate legends in the software space. The tech is extremely good in the problem space it's meant to operate. They aren't trying to be a thick platform with all the bells and whistles. Their strategy is to build a developer community that can deliver those solutions using their platform, That strategy seems to be working. We'll see if Karp's vision of one platform to rule them all works out in the long run. I certainly hope more developers can take advantage of Foundry and scale it the way the DBX community is scaling.
When moon?🤨
Databricks is open source while palantir is closed source.
Thats just not correct. Check them out on GitHub. Their spark distribution is open source. Their react framework is open source. They also use open source under the hood including Flink, ElasticSearch, HDFS, etc. DBX only recently open sourced the best parts of Delta, but their platform is no more open source than Palantir's. DBX is open to tje public is open to the public though, and Palantir is not. Im sure that will change eventually.
@@codestrap8031 Open source doesn't mean you utilize open source librairies to distribute software. Hence, Windows is closed source, but NT implements a couple of OpenBSD/FreeBSD stack library.
By your own definition DBX is closed source since they do not, to my knowledge open source their multi tenant SaaS platform or the stack that gets deployed in your infrastructure. To my knowledge they open source key libraries and technologies just like Palantir. Checkout both their github pages please and get back to me.
@@codestrap8031 We did.
To be honest, I think C3 AI has the same plaform as Palantir, but more powerful than Palantir
vahahahahahahhaha
Then explain how many customers does c3 ai have