events.unraveldata.com
Open in
urlscan Pro
2606:4700:10::6816:489b
Public Scan
Submitted URL: https://go.unraveldata.com/OTY3LUpOTi0wOTgAAAGPSTRrJ83ZZ-mzl-m22tk7drnI7reAOIWeLTWdhZ_gKdCfM-KGcd4hhXP5uE5AxPEq0tFmsqg=
Effective URL: https://events.unraveldata.com/transform-analytics-cloud?utm_medium=email&utm_source=marketo.com&utm_campaign=transforming+anal...
Submission: On November 07 via api from US — Scanned from DE
Effective URL: https://events.unraveldata.com/transform-analytics-cloud?utm_medium=email&utm_source=marketo.com&utm_campaign=transforming+anal...
Submission: On November 07 via api from US — Scanned from DE
Form analysis 1 forms found in the DOM
<form novalidate="" aria-atomic="true">
<div data-qa="FormHeadline" font-size="32" color="rgb(25, 25, 25)" role="heading" aria-level="2" class="sc-bBeLUv ipejFM"><span class="sc-kOPcWz eVnBED">
<p>Register Now</p>
</span></div>
<div>
<div data-qa="FormNode__Container" class="SuiContainer sc-gFVvzn frzTou">
<div data-qa="QuestionView" class="SuiContainer SuiContainer--flex-group" style="transition: opacity 125ms ease-in-out 0s; opacity: 1;">
<div class="SuiTextInput SuiTextInput--full-width" data-qa="QuestionNode__text-First Name" data-field-name="First Name"><label id="label-2" for="field-1" class="sc-imWYAI hbymbO SuiLabel"><span class="sc-kOPcWz eVnBED">
<p>FIRST NAME <strong aria-hidden="true" style="color:#DD2C00">*</strong></p>
</span></label>
<div id="described-by-3" class="sc-eqUAAy fPOtPu SuiHelperText"></div>
<div class="sc-fqkvVR fxtGFC SuiField SuiField--full-width SuiField--has-form-el SuiTextInput-field">
<div class="sc-dcJsrY dMTXmo">
<div class="sc-iGgWBj dXAkrz SuiField__current"><span class="sc-gsFSXq gUlrGl SuiField__placeholder">Enter your first name</span><input id="field-1" aria-labelledby="label-2" aria-describedby="described-by-3" class="SuiTextInput-input"
name="19290765" type="text" autocomplete="off" aria-required="true" aria-invalid="false" value=""></div>
</div>
</div>
</div>
</div>
</div>
<div data-qa="FormNode__Container" class="SuiContainer sc-gFVvzn frzTou">
<div data-qa="QuestionView" class="SuiContainer SuiContainer--flex-group" style="transition: opacity 125ms ease-in-out 0s; opacity: 1;">
<div class="SuiTextInput SuiTextInput--full-width" data-qa="QuestionNode__text-Last Name" data-field-name="Last Name"><label id="label-5" for="field-4" class="sc-imWYAI hbymbO SuiLabel"><span class="sc-kOPcWz eVnBED">
<p>LAST NAME <strong aria-hidden="true" style="color:#DD2C00">*</strong></p>
</span></label>
<div id="described-by-6" class="sc-eqUAAy fPOtPu SuiHelperText"></div>
<div class="sc-fqkvVR fxtGFC SuiField SuiField--full-width SuiField--has-form-el SuiTextInput-field">
<div class="sc-dcJsrY dMTXmo">
<div class="sc-iGgWBj dXAkrz SuiField__current"><span class="sc-gsFSXq gUlrGl SuiField__placeholder">Enter your last name</span><input id="field-4" aria-labelledby="label-5" aria-describedby="described-by-6" class="SuiTextInput-input"
name="19290766" type="text" autocomplete="off" aria-required="true" aria-invalid="false" value=""></div>
</div>
</div>
</div>
</div>
</div>
<div data-qa="FormNode__Container" class="SuiContainer sc-gFVvzn frzTou">
<div data-qa="QuestionView" class="SuiContainer SuiContainer--flex-group" style="transition: opacity 125ms ease-in-out 0s; opacity: 1;">
<div class="SuiTextInput SuiTextInput--full-width" data-qa="QuestionNode__email-Email" data-field-name="Email"><label id="label-8" for="field-7" class="sc-imWYAI hbymbO SuiLabel"><span class="sc-kOPcWz eVnBED">
<p>WORK EMAIL <strong aria-hidden="true" style="color:#DD2C00">*</strong></p>
</span></label>
<div id="described-by-9" class="sc-eqUAAy fPOtPu SuiHelperText"></div>
<div class="sc-fqkvVR fxtGFC SuiField SuiField--full-width SuiField--has-form-el SuiTextInput-field">
<div class="sc-dcJsrY dMTXmo">
<div class="sc-iGgWBj dXAkrz SuiField__current"><span class="sc-gsFSXq gUlrGl SuiField__placeholder">you@your.email</span><input id="field-7" aria-labelledby="label-8" aria-describedby="described-by-9" class="SuiTextInput-input"
name="19290767" type="email" autocomplete="off" aria-required="true" aria-invalid="false" value=""></div>
</div>
</div>
</div>
</div>
</div>
<div data-qa="FormNode__Container" class="SuiContainer sc-gFVvzn frzTou">
<div data-qa="QuestionView" class="SuiContainer SuiContainer--flex-group" style="transition: opacity 125ms ease-in-out 0s; opacity: 1;">
<div class="SuiTextInput SuiTextInput--full-width" data-qa="QuestionNode__text-Company Name" data-field-name="Company Name"><label id="label-11" for="field-10" class="sc-imWYAI hbymbO SuiLabel"><span class="sc-kOPcWz eVnBED">
<p>COMPANY NAME <strong aria-hidden="true" style="color:#DD2C00">*</strong></p>
</span></label>
<div id="described-by-12" class="sc-eqUAAy fPOtPu SuiHelperText"></div>
<div class="sc-fqkvVR fxtGFC SuiField SuiField--full-width SuiField--has-form-el SuiTextInput-field">
<div class="sc-dcJsrY dMTXmo">
<div class="sc-iGgWBj dXAkrz SuiField__current"><span class="sc-gsFSXq gUlrGl SuiField__placeholder">Company</span><input id="field-10" aria-labelledby="label-11" aria-describedby="described-by-12" class="SuiTextInput-input"
name="19290768" type="text" autocomplete="off" aria-required="true" aria-invalid="false" value=""></div>
</div>
</div>
</div>
</div>
</div>
<div data-qa="FormNode__Container" class="SuiContainer sc-gFVvzn frzTou">
<div data-qa="QuestionView" class="SuiContainer SuiContainer--flex-group" style="transition: opacity 125ms ease-in-out 0s; opacity: 1;">
<div class="SuiTextInput SuiTextInput--full-width" data-qa="QuestionNode__text-Job Title" data-field-name="Job Title"><label id="label-14" for="field-13" class="sc-imWYAI hbymbO SuiLabel"><span class="sc-kOPcWz eVnBED">
<p>JOB TITLE <strong aria-hidden="true" style="color:#DD2C00">*</strong></p>
</span></label>
<div id="described-by-15" class="sc-eqUAAy fPOtPu SuiHelperText"></div>
<div class="sc-fqkvVR fxtGFC SuiField SuiField--full-width SuiField--has-form-el SuiTextInput-field">
<div class="sc-dcJsrY dMTXmo">
<div class="sc-iGgWBj dXAkrz SuiField__current"><span class="sc-gsFSXq gUlrGl SuiField__placeholder">Job Title</span><input id="field-13" aria-labelledby="label-14" aria-describedby="described-by-15" class="SuiTextInput-input"
name="19290769" type="text" autocomplete="off" aria-required="true" aria-invalid="false" value=""></div>
</div>
</div>
</div>
</div>
</div>
<div data-qa="FormNode__Container" class="SuiContainer sc-gFVvzn frzTou">
<div data-qa="QuestionView" class="SuiContainer SuiContainer--flex-group" style="transition: opacity 125ms ease-in-out 0s; opacity: 1;">
<div class="SuiTextInput SuiTextInput--full-width" data-qa="QuestionNode__DO YOU HAVE A QUESTION YOU'D LIKE ANSWERED DURING THE EVENT? ASK BELOW!" data-field-name="DO YOU HAVE A QUESTION YOU'D LIKE ANSWERED DURING THE EVENT? ASK BELOW!">
<label id="label-17" for="field-16" class="sc-imWYAI hbymbO SuiLabel"><span class="sc-kOPcWz eVnBED">
<p>DO YOU HAVE A QUESTION YOU'D LIKE ANSWERED IN THE Q&A? ASK BELOW!</p>
</span></label>
<div id="described-by-18" class="sc-eqUAAy fPOtPu SuiHelperText"></div>
<div class="sc-fqkvVR fxtGFC SuiField SuiField--full-width SuiField--has-form-el SuiTextInput-field">
<div class="sc-dcJsrY dMTXmo">
<div class="sc-iGgWBj dXAkrz SuiField__current"><span class="sc-gsFSXq gUlrGl SuiField__placeholder"></span>
<div class="SuiContainer SuiContainer--full-width"><textarea id="field-16" aria-labelledby="label-17" aria-describedby="described-by-18" name="19290770" rows="2" aria-invalid="false" class="SuiTextInput-input" tabindex="-1"
readonly="" style="resize: none; overflow: hidden; visibility: hidden; position: absolute; height: initial;"></textarea><textarea id="field-16" aria-labelledby="label-17" aria-describedby="described-by-18" name="19290770" rows="2"
aria-invalid="false" class="SuiTextInput-input" style="height: 64px;"></textarea></div>
</div>
</div>
</div>
</div>
</div>
</div>
<div data-qa="FormNode__Container" class="SuiContainer sc-gFVvzn frzTou">
<div data-qa="QuestionView" class="SuiContainer SuiContainer--flex-group" style="transition: opacity 125ms ease-in-out 0s; opacity: 1;">
<div class="SuiCheckbox QuestionNode__checkbox SuiCheckbox--unchecked" data-qa="QuestionNode__I'd like to learn more about Unravel Data in a 1:1 demo." data-field-name="I'd like to learn more about Unravel Data in a 1:1 demo.">
<div class="SuiCheckbox__main-elements SuiCheckbox__baseline-alignment">
<div class="SuiCheckbox__field "><input class="SuiCheckbox__input" id="SuiCheckbox-19" type="checkbox" name="19290771" aria-describedby="described-by-20" aria-invalid="false">
<div class="sc-jlZhew leuExa SuiCheckbox__checkbox">
<div class="sc-cwHptR gFWjUs SuiCheckbox__indicator"><svg width="21" height="21" viewBox="0 0 21 21" xmlns="http://www.w3.org/2000/svg" class="SuiCheckbox__icon">
<g fill="currentColor" fill-rule="evenodd">
<rect transform="rotate(-45 11.33 10.67)" x="5.83" y="9.17" width="11" height="3" rx="1"></rect>
<rect transform="rotate(45 7.44 12.44)" x="4.44" y="10.94" width="6" height="3" rx="1"></rect>
</g>
</svg></div>
</div>
</div><label class="sc-jEACwC gajZgb SuiCheckbox__label" for="SuiCheckbox-19"><span class="QuestionNode__checkbox-label"><span class="sc-kOPcWz eVnBED">
<p>I'd like to learn more about Unravel Data in a 1:1 demo.</p>
</span></span></label>
</div>
<div id="described-by-20" class="sc-eqUAAy fPOtPu SuiHelperText"></div>
</div>
</div>
</div>
<div data-qa="FormNode__Container" class="SuiContainer sc-gFVvzn frzTou">
<div data-qa="QuestionView" class="SuiContainer SuiContainer--flex-group" style="transition: opacity 125ms ease-in-out 0s; opacity: 1;">
<p class="sc-gmPhUn iWHCpd"><span class="sc-kOPcWz eVnBED"></span></p>
<p>By submitting, I agree to the processing of my personal data in accordance with the <a target="_blank" href="https://www.unraveldata.com/privacy-policy/">Unravel Data privacy policy.</a> I understand I can update my preferences at any
time.</p>
<p></p>
</div>
</div>
<div data-qa="FormNode__Container" class="SuiContainer"><input name="19290772" type="text" style="display: none;"></div>
<div data-qa="FormNode__Container" class="SuiContainer"><input name="19290773" type="text" style="display: none;"></div>
<div data-qa="FormNode__Container" class="SuiContainer"><input name="19290774" type="text" style="display: none;"></div>
<div data-qa="FormNode__Container" class="SuiContainer"><input name="19290775" type="text" style="display: none;"></div>
<div data-qa="FormNode__Container" class="SuiContainer"><input name="19290776" type="text" style="display: none;"></div>
</div>
<div class="sc-jaXxmE bHlkjS"><button type="submit" class="sc-ePDLzJ cxOTQk SuiButton SuiButton--full-width SuiButton--primary SuiButton--wide">
<div class="SuiButton-content">
<div class="SuiButton-label"><span class="sc-kOPcWz eVnBED">
<p>REGISTER NOW</p>
</span></div>
</div>
</button></div>
</form>Text Content
Transforming analytics on the cloud: Supercharge your data applications with Databricks, AWS and Unravel Nov 9th , 8:00am PST Register Now 7:00 PM Session Name Room Name This session is going inspire you to be great! Register Bullet Point Copy Sebastian Poindexter Sr. Technology Consultant OPENFORM + Text goes here X 04 Resource Title Explore why the most dynamic leaders are building strategies for agile event marketing with a focus on scaling in-person and virtual events that are fast, flexible, and hyper-effective at driving meaningful connections. Download Text goes here X 12:00-1:00 PM Main Event Special guest presenters Bullet Point Copy Jake Hernandez Chief Creative Officer Openform Jake Hernandez is senior director of product marketing at Openform, where his goal is to help people unlock the power of in-person events. Prior to Openform, he led global product marketing teams. Speaker Name Short speaker biography. On what date does the event take place? The event is scheduled for Wednesday, May 31, 2024. Watch Enhance your user experience and build brand equity with your design vernacular. Watch Enhance your user experience and build brand equity with your design vernacular. Transforming analytics on the cloud: Supercharge your data applications with Databricks, AWS and Unravel November 9 , 2023 8am PDT | 11am EDT | 4pm BST REGISTER NOW Text goes here X Transforming analytics on the cloud: Supercharge your data applications with Databricks, AWS and Unravel Relax and relive the moment. WATCH NOW Text goes here X What was covered Organizations are feeling pressure to launch new data applications faster to meet end-user demand. Cloud data platforms help accelerate launch times with on-demand delivery of infrastructure and pay-as-you-go pricing. Last year, 98% of the overall database management system (DBMS) market growth came from cloud database platform as a service (dbPaaS). 80% of organizations have adopted agile practices to increase their pace of innovation. With this in mind, Unravel is hosting a live event to help you leverage AWS Competency Partners, data observability, and CI/CD integration to achieve speed and scale with Databricks. Ask one of the 10 best questions as ranked by our presenters to win Unravel swag. All registrants will receive a link to the recording following the webinar. Join Unravel Data for a live event about transforming analytics on the cloud. 8am PST on November 9, Clinton Ford and special guest speakers Lee Bergs, AWS Global Manager, Partner Enablement at AWS and Prajakta Kalmegh, Sr. Product Manager at Unravel will help you discover how you can: Key Topics: What agile event programs are and why they're essential Strategies and tools for building agile event marketing How to scale agile event programs without losing brand integrity or valuable data Steps you can take right now to introduce agile event marketing into your business Grab your favorite snack and tune in. Pro Tip: First, watch the video to learn how to use this type of theme. Then, replace the video by selecting it and clicking on the "Embed Options" tab on the right. Then, simply paste in the direct link to YouTube or Vimeo. Hide this element when you're finished. If you would rather embed a video or stream from a different service, use the "Livestream - iFrame Block", which can be found hidden in the Layout Tab. Need more help? Here are some great Help Center articles regarding embed best-practices: How do I resize an iFrame? What's the difference between iFrame elements and Video elements? How do I embed webinars or livestreams using an iFrame element? What can I embed using an iframe into Splash? What You’ll Learn Organizations are feeling pressure to launch new data applications faster to meet end-user demand. Cloud data platforms help accelerate launch times with on-demand delivery of infrastructure and pay-as-you-go pricing. Last year, 98% of the overall database management system (DBMS) market growth came from cloud database platform as a service (dbPaaS). 80% of organizations have adopted agile practices to increase their pace of innovation. With this in mind, Unravel is hosting a live event to help you leverage AWS Competency Partners, data observability, and CI/CD integration to achieve speed and scale with Databricks. Ask one of the 10 best questions as ranked by our presenters to win Unravel swag. All registrants will receive a link to the recording following the webinar. Join Unravel Data for a live event about transforming analytics on the cloud. 8am PST on November 9, Clinton Ford and special guest speakers Lee Bergs, AWS Global Manager, Partner Enablement at AWS and Prajakta Kalmegh, Sr. Product Manager at Unravel will help you discover how you can: In this 1 hour virtual event, you will learn how to: • Find AWS Data and Analytics Competency Partners who are trained and certified to help you deliver greater value for your business, increase agility, and lower costs • Learn how you and your team can leverage AWS Skill Builder – an online learning center where you can learn from AWS experts and build cloud skills online • Implement Databricks CI/CD integration to streamline data pipeline development and deployment, accelerating release times and frequency, while improving code quality. • Leverage AI-powered data observability and FinOps as part of your CI/CD process to improve code and SQL query performance and efficiency before going to production Your Hosts Sanjeev Mohan Principal and Founder SAnjMO + Text goes here X Subramanian Iyer Unravel Training and Enablement Leader and Databricks SME speedboat + Text goes here X Don Hilborn Field CTO Unravel Data + Text goes here X Speakers Lee Bergs Global Manager, Partner enanablement AWS Prajakta Kalmegh Senior Product Manager Unravel Data What is driving the demand for faster data life cycles? Chapter 0 Enhance your user experience and build brand equity with your design vernacular. Watch Chapter 1 Enhance your user experience and build brand equity with your design vernacular. What is driving the demand for faster data life cycles? Chapter 2 Enhance your user experience and build brand equity with your design vernacular. Watch Chapter 3 Enhance your user experience and build brand equity with your design vernacular. What is driving the demand for faster data life cycles? Chapter 4 Enhance your user experience and build brand equity with your design vernacular. Watch Chapter 5 Enhance your user experience and build brand equity with your design vernacular. What is driving the demand for faster data life cycles? Chapter 6 Enhance your user experience and build brand equity with your design vernacular. Watch Chapter 7 Enhance your user experience and build brand equity with your design vernacular. What is driving the demand for faster data life cycles? Chapter 8 Enhance your user experience and build brand equity with your design vernacular. Watch Chapter 9 Enhance your user experience and build brand equity with your design vernacular. What is driving the demand for faster data life cycles? Chapter 10 Enhance your user experience and build brand equity with your design vernacular. What is driving the demand for faster data life cycles? Enhance your user experience and build brand equity with your design vernacular. When it comes to generative AI, what is the key differentiator that makes one product better than another? Enhance your user experience and build brand equity with your design vernacular. data: unstructured, semi-structured, streaming Enhance your user experience and build brand equity with your design vernacular. How to shift left, with Sanjeev Enhance your user experience and build brand equity with your design vernacular. Unravel Data demo with Don Hilborn Enhance your user experience and build brand equity with your design vernacular. Subbu's Perspective Enhance your user experience and build brand equity with your design vernacular. Q&A Enhance your user experience and build brand equity with your design vernacular. Q&A: Enhance your user experience and build brand equity with your design vernacular. Q&A Enhance your user experience and build brand equity with your design vernacular. Q&A: Enhance your user experience and build brand equity with your design vernacular. Q&A Enhance your user experience and build brand equity with your design vernacular. Virtual Event Q&A Q: How do you see AI playing a factor in accelerating the data analytics life cycle, and how will Unravel use it effectively? A: 72% of technology leaders agree that data challenges are the most likely factor to jeopardize AI/ML goals. Enterprise organizations need more data than ever before for training, validation, verification, and drift analysis. Enterprises such as one of the largest health insurance providers in the U.S use Unravel to ensure that its business-critical data applications are optimized for performance, reliability, and cost in its development environment—before they go live in production. Q: How Unravel helps Government and Public sector in the area of data analytics? Is there any data protection limitations? A: Unravel is SOC 2 Type II compliant and treats all data with the highest degree of security. If you would like more information, please visit Unravel Data Security. Q: How purpose built AI can help businesses directly? Any use case you have? A: Unravel’s AI-powered Insights Engine has been built to continuously ingest and interpret the millions of ongoing data streams to provide real-time insights into application and system performance, and automatically deliver recommendations to optimize code and resource allocations for performance and financial efficiencies Q: Will there be any IoT application/use cases towards the end? A: We don't plan to dive deep into IoT use cases in this webinar, but that would be a great topic for a future webinar, since it's so applicable to enterprise Unravel customers. And it's a key reason companies come to Unravel. For example, “lack of resources/knowledge to scale” is the leading reason preventing IoT data deployments. Budget and staffing resources constraints pose real risks to launching profitable data and AI projects. For example, Maersk is embracing IoT to revolutionize supply chain (see this Smart Maritime Network Article, 330 Maersk container ships to install Starlink; this CIO article, Maersk embraces edge computing to revolutionize supply chain, and our fireside chat, Enabling Strong Engineering Practices at Maersk. Transcript <<Transcript>> Clinton Ford: Hi, everyone and welcome to this live event, Accelerate the Data Analytics Lifecycle with Unravel. I'm Clinton Ford, VP of Product Marketing here at Unravel, and I'm joined today by Sanjeev Mohan, principal and founder at SanjMo, Subramanian Iyer, the Unravel Training and Enablement Leader and Databricks Subject Matter Expert at Speedboat, and Don Hilborn, the Field CTO at Unravel Data. Great to have you with us here today. Don Hilborn: Thanks, Clinton. Good to be here. Sanjeev Mohan: Thank you. Thanks. Clinton Ford: So as you know, there's been a lot of increase in demand for data and for AI. We're seeing this pickup really across the board and wanted to spend a few minutes, first of all starting off on this topic just to get your perspective on some of the key drivers happening out there in the landscape. So for example, one question that I have for you is, from your perspective, what do you see as driving the demand for faster data life cycles, going from data ingestion and curation, all the way to producing a model and then putting that into production? What's making that such a high pressure, high stakes situation right now in enterprises? Don Hilborn: I think the key to success is consistently making good decisions and the key to good decisions is good data, good information. And so all things being equal, the better information you have, the better decisions you'll make. An analogy I sometimes use for this is, let's say that we had a choice of flying on one of two of the latest and greatest Boeing 787s. They've been maintained identically. The crews have been trained identically. Both planes are flying through the Himalayas at night and the only difference is one has an altimeter that works and the other one doesn't. Of course, we all want to be on the plane with the altimeter that works and the point being that given the same decision-making capabilities, an organization can be much more effective if it has the right data at the right time. And we're recognizing more and more that we can process these huge volumes of data and essentially make the entire data set available to decision makers. And I think they recognize that, that's going to be a significant competitive differentiator for them. Clinton Ford: Excellent. Sanjeev, you had a comment as well. What was your perspective on that? Sanjeev Mohan: I agree with everything Don says. What businesses really want is they want access to all data, like he said, and they want it when they are making decisions, not 24 hours later after the campaign's already over, now you come back and tell me that these are the parameters, it's too late. So the reason why businesses sometimes are unhappy with their data ecosystem or their data layout is because the time to decision is too long for them to stay competitive. Subramanian Iyer: I was going to say at the same point. In fact, over the seven, eight years I've been in the industry in Spark, we've seen this big shift. It's gone from, oh, let's do an end of day batch. And now, there's a shift in real time or near real time and the time to insight has been steadily going down. It used to be the end of day, then it became like every hour. Now they want every 15 minutes. Combine that with models and now you want execs asking real time, like a natural language question and getting an answer that is up to date like five minutes ago. What's happening, right? And the companies who get it right will survive, will thrive, and the ones that don't get it right will fall by the wayside. So that's the pressure that you talked about, Clinton. Sanjeev Mohan: Yeah. And if we translate this even into the cost, see this is how things are so different now. I'll give you a funny example. Many of my colleagues, myself included, we've been guilty of running a Cartesian join in Oracle database in the past where you have a million row table and you just join every column every... So it's just in the past, what would that do to your database? It would bring it down to a crawl. The DBA would go find the errant SQL and kill it and then we would be back to normal. What would happen today if you ran it? We just get a $40,000 bill. So we have to be more real time. We have to know exactly what's happening at the time and if something is an anomaly, then we won't act on it instantly and shut it down before the CFO gets a bill and all hell breaks loose. Clinton Ford: Yeah, efficiency really is the new SLA. Subramanian Iyer: I cannot emphasize on that enough because once upon a time in the old system when it's on-prem, there was a natural constraint that you can't run a bill bigger than your system is. In the cloud, unfortunately, that constraint is not there, right? So you've got to be extra vigilant to the cloud. Don, over to you. Don Hilborn: I was just going to say, I think one of the biggest things that has changed is, and I think this was touched on by both Subu and Sanjeev, is the amount of data. A lot of these models that we're using have been around since the fifties. And in fact, LLMs were first proposed in the sixties. What's changed is our ability to process these huge volumes of data. One of the things Google proved to the world was that if you could process all of the data in their case, the internet, you could be the best search engine period. And so Google actually maintained multiple copies of the internet and one of the things that's been able to do from the onset is to go through and index that data using machine learning. And therefore when you do a search, it comes back with a document or something that meets that search. What we're seeing with LLMs and generative AI is now even better than that. You can go ask it a question and it can come back to you with an answer, what appears to be a human-like answer. And I think a lot of that's driven, and Subu touched on this a little bit, is when we had data centers, we were kind of confined with whatever was in that data center. In fact, one of the ways I got into the big data space was I was faced with a problem where we had to index six and a half petabytes of data and search it. And I went to my data team and I said, "Hey, how long will it take you to get this available to me?" And they laughed me out of the data center. It wasn't something that they could make available even if they had the room. But in the cloud, of course in a matter of minutes I could stand up a super environment that was capable of processing all of that data. And that was kind of the eye opener for me that data centers were going to fall by the wayside. The economics are driving it, but also this ability to stand up a supercomputer that you never would've had in your data center and process these tremendous amounts of data is a big differentiator that companies are seeing they're able to do today and it results in better decisions for them. So I think that's another thing driving the growing need for data and more and more data. Sanjeev Mohan: Small, medium size companies today on an average, according to many studies have over 100 SaaS products on an average, every company. If you're talking about a large company, you're talking about on an average of 450 SaaS products. Imagine the amount of data that is being generated from these SaaS products. And the funny thing is a lot of these SaaS, so I'm talking about SASS products like Marketo for marketing or HubSpot or Salesforce, obviously all kinds of products have now come up and the problem is that each one of them has its own data model and that can change when the new release comes, they can change it because they all experimenting. So many of these are new companies. So this is the explosion of data that we are ingesting. Now we have to analyze all of this data and find some nuggets of what are the correlations between these different elements. And by the way, the schema's all different. The semantic meanings are all different. And this is the reason why it is so expensive to build these so-called modern data stacks and hence cost is really high. I see this all the time. The CFO goes to the CDO and says, "You invested $20 million, what did I get?" A few reports and dashboards? It's very hard to determine because there's so much overhead trying to integrate all these data assets. Subramanian Iyer: I think you touched on some really good points. I mean this is what we see out there, right? People are drowning in data, starving for insights. I'm sure you've heard that before, right? And the fundamental reason why this is happening is because like you said, there is data... There's so many versions of the same data, processing it, accounting for versions, massaging them into one canonical form and then doing insights on it. That's the heavy lift. That's where the bulk of the compute and the bulk of the developer effort goes in, and that's just in batch. And now compound that with streaming for real time and you get a sense of how hairy the whole thing gets essentially. Clinton Ford: Yeah. Now Don, to your point, you talked a little bit earlier about everyone wanting the very latest and greatest product. So earlier this year, McKinsey published results from a survey that said 75% of respondents expect gen AI to cause significant or disruptive change in the nature of their industry's competition in the next three years. So when it comes to generative AI, what is the key differentiator? What really helps one product stand out against another one? What is going to be that factor that makes one product better than the other? Don Hilborn: I think it's ultimately the ability to access the relevant data and then give the most accurate answer. I know that's somewhat oversimplified, but it does come down to that. And I think what all these industries are concerned with is they don't want to be disrupted, right? They don't want to be the next company that falls to Uber or to Airbnb or to LinkedIn or to Facebook. These companies have disrupted the way things are done in their industries and they've all used data to do it. And you're right, now they see with generative AI, the threat is even more powerful. Not only are they using data, but now the AI itself is being used to help these companies make better decisions. And again, whoever makes the best decisions most consistently will ultimately come out ahead and companies know that. So I think that very simplistically, they'll be looking to generative AI that makes sure that it's considering all the relevant data and it's giving them the best answers possible. And then of course, all the subtleties around ease of use and cost and other things will come into play. But I think those will be the biggest drivers that will ultimately drive the more successful generative AI platform. Sanjeev Mohan: I'm an analyst. I like to talk, so you have to shut me up. Otherwise, I'm trained to do, I'm paid to do this stuff. But generative AI is a beast by itself. It has introduced some new kinds of problems that we've never really had to deal with in the past, mainly because an LLM or a foundation model is probabilistic as opposed to SQL, which is deterministic. If I run a SQL statement, I will get the same response every single time my data doesn't change. But if I ask a question to an LLM, it may give me a different answer. So the biggest problem that businesses face is hallucination. The second problem they face is security. How do we know that sensitive data is protected? You didn't train on data that wasn't masked. GDPR, data, privacy, protection, sovereignty, all of those things come in the picture. But the third thing that I want to talk about on this call, because this is... We talk about cost FinOps, data observability, and that is a cost piece. So the cost of training the model is out of reach of pretty much any company in the world, unless you happen to be one of the hyperscalers or you've been hugely bankrolled by VCs is hundreds of millions of dollars to keep these models trained. In fact, OpenAI, even with its billions of investment from Microsoft, is bleeding money. And I saw some report a week ago that they may run out of money by next year because that is the cost of training. And even the cost of inference is very high. When I run a Google query, it costs literally pennies, fractions of pennies to run that. But when you run a LLM query, it costs a lot of money. So businesses that want to use generative AI have to rethink the cost factor. A, what model should I use? Not every model is meant for every area. You cannot use GPT4 for every single thing. It's trained on public data. Maybe these are smaller domain specific models. A model that's trained on text to generate Python code or SQL could be used for that purpose. A model that is meant for translating a hundred plus languages, use that for translation. So there's going to be a number of these LLMs and that's a first step. We haven't even- Subramanian Iyer: It's the first step. Sanjeev Mohan: Yeah, I'm going to stop here and let- Subramanian Iyer: No, I think you're going exactly where I was going to go, which is great because you're in my head, literally, right? Is that as a practitioner you got to make a bunch of choices. Right? The success, the accuracy of your model depends on A, the base model you choose. So you have your mosaic, you have DALL-E, you have, HuggingFace list of them, right? So pick a base model, then you've got to build your chain, you've got to install the right agents and line chain, and then ultimately the data that you're feeding into the model matters because we feed junk, your model's going to give you junk, right? Garbage in, garbage out. And that's where a huge part of the effort that goes in for developers like me to make sure that my data is clean, requires me to put all sorts of processing constraints. We touched on this earlier, right? We have all these three different schemas, different versions, copies of data, reconciling it, cleaning, all that stuff is work. And that is going to be the other frontier of cost overruns and blowouts. Sanjeev Mohan: I think data observability vendors, like most of us, don't even know how and where this technology is going at this point. We just don't know. I mean at such early stages. But data observability vendors are going to have a field day trying to figure out the cost overruns, the quality, the latency going from different parts of this generative AI pipeline. Clinton Ford: Now is all this data really structured data or are you seeing a shift as well towards streaming or semi-structured or unstructured for these use cases? Let's start with Sanjeev. Sanjeev Mohan: Thank you for asking that because like I mentioned, choosing your model is only one of many cost containment steps you can do. The second one is what is your use case? In my opinion, and I'm making this prediction here on this webinar, is that in 2024, we will see the use cases that get funded and are deployed will be the ones where there is a very clear cut ROI. For example, to answer your question, will it be structured semi-structured data, streaming data? I think unstructured data is a clear cut case because what happens is that many organizations get PDF documents from Bloomberg, from Morgan Stanley, from Goldman Sachs, from all these places. Every one is in a different format. Somebody, a team is sitting there trying to decipher what's in each of these documents and then putting it manually inside a database. That is a ROI waiting to happen. It's like if you can put an LLM that can intelligently do extraction of entities and then classify or tag them to have a common meaning, it's going to be highly adopted. But if you have a use case where you are putting the LLM because you're making your developers more productive, the question will come up, how do you measure productivity? It's so hard. So how do you know that LLM is useful? So this is going to be the second important cost related consideration for gen AI. Clinton Ford: And Subu, what's your take on that, on the different types of data that you see demand for in terms of these generative AI projects and other things like LLMs? Subramanian Iyer: I think it's going to be all... I think every kind of data is going to get scraped up in this thing, right? It's going to get vacuumed up. With Lakehouse and Databricks, for example, we have opened the door to unstructured and semi-structured data. So JSON data, PDF, audio, video, binary, anything can be processed in Lakehouse, in Databricks very effectively. So I think you will see more and more of that for sure. Of course, structured is always there. The workout of the industry is structured data, right? So long story short, it's going to be all kinds of data. Sanjeev Mohan: It's going to be a combination. We are going to see a combination of operational use cases and analytical. We are already doing that. We call it HTAP, so OLTP and OLAP, predictive, and generative. And let me give you an example of what I mean by that. So let's say I want to find out which customers of mine have a propensity to churn and not renew their contract next year. Every major company has this problem. Today I can either write a SQL statement and do a "where clause" where this is this and I can generate a list. Or if I'm super smart in my company, I can do a machine learning model, and I can do some sort of decision, random forecast or some sort of way to predict which customers. So now I've done predictive and I've done my normal query. Where gen AI comes in the picture is that I'm going to do this kind of machine learning, find out, tell me the list of customers that may churn, and then I'm going to send that list to my LLM and say for each one of these, create generate, because it's generative AI, generate a action plan based on everything that you see about them in my database. And now I may have a separate plan for Don, separate for Subu, separate for Clinton, separate for me, and the salesperson can now have a more precise personalized plan of how to go and try to convince this customer to renew. So this generator piece never existed. That is a brand new, we plug it on top of everything that we are doing today. Clinton Ford: Great. And Don, what are you seeing? You'd meet with customers, are you seeing the variety of data proliferate as well? Don Hilborn: Yeah, I think that's really where big data came into play, right? The three V's, the volume, velocity or variety. And that was really how I got into the big data space. I had primarily worked in structured data up to a point, but I happened to be working for an oil and gas company when one of our rigs, or at least a percentage of the rig we own, blew up in the Gulf of Mexico and attracted a lot of attention. And what came out of that was a subpoena from Congress for us to go through about six and a half petabytes of data and produce to them the results. And I mentioned that earlier where I went to my data team and said, "Hey, what can you do for me?" And they kind of laughed me out of the room, but I was able to take a bunch of small servers and wire them together using this new fangled tool called Hadoop. And I was able to look at the data regardless of whether it was structured, semi-structured unstructured images. And in fact the issue turned out to be an issue with a PDF and just an old PDF being used on the rig, which caused them to use a lighter cement than they were supposed to down the hole. But without that capability to process it, regardless of the volume, velocity and variety, we would've never come to that answer. And so I think big data has been about the three V's for a while, but what we're seeing is of course AI more and more and ML wants all of the data. They want larger and larger amounts of data. We've actually had data scientists for quite some time. They were called actuaries initially, and I think there's one that dates all the way back to the 1700's where they simply looked at a set of data and then they used that to make a prediction. What we're seeing though is that data scientists are becoming very passionate about getting more and more data into their models because they recognize that that is what improves the accuracy of their model more than anything else. So you have to be able to go after all the data regardless of its structure, regardless of how quickly it's being generated, you have to be able to process it. And then of course the size, the volume of it, big data is a little bit of an overloaded term, but it really does mean the three V's, volume, velocity and variety of data. So yeah, I think that more and more we need to get to all of the data and in fact, I think you'll find that a very small percentage of data is actually structured, especially with the internet. So we need to be able to get to it all and incorporate it into our decision-making process. So yeah, definitely seeing that and more and more from data scientists wanting larger and larger amounts of data. Clinton Ford: Great. Now shifting the conversation towards efficiency. So Sanjeev, you had hinted at this. I saw it in a data science article that 2023 marks a shift in focus to efficiency and the broad adoption of generative AI. What are you hearing any recent news that you've heard from any analysts or contacts that you have about the importance of the role of financial cost in data observability? Sanjeev Mohan: Clinton, when we started this year, I did a webinar for the cube, which I do every year with five or six of my fellow analysts and we do some predictions for this year. My prediction was how important cost optimization is going to be in 2023. Low and behold, generative AI just dominated all conversations and temporarily we shoved it under the carpet, but now it's bubbled back up because you cannot ignore cost optimization. It is so important. So I had done a chapter in data observability, the reality book where Unravel and a number of other data observability companies had all contributed a chapter. So I wrote the last chapter which said the newest kid on the block for data observability, actually the two data FinOps and data BizOps. Data BizOps is kind of a weird thing that I thought of, which is observing the productivity of your data, but let's not go there, let's talk about the FinOps. Now, Gartner has in the meantime also come out with its definition of data observability and now they're onto a second version or iteration of data observability and they've added FinOps into that. So I'm really happy to see that my ex-colleagues at Gartner and I are completely aligned in our definition of data observability, including FinOps. My take on this is that when we do data DataOps ability, actually data governance in general, we are working at the metadata layer. So we are taking the metadata of our data objects, which could be technical like schema and distribution or we are taking business metadata. Finops is actually kind of a usage metadata. How effective is that environment where you are running? Are you under utilizing your resources? So maybe you should use a cheaper resource. How quickly are you running your jobs either bottlenecked and do you have the right SQL queries in place? So this is the financial aspect of data observability. So basically now we have conversed with Gartner and the industry definition, there are three major components of data observability, data quality, data readability and FinOps. So that's what I'm seeing in the industry. Clinton Ford: Fascinating. And Subu would be interested to get your take on this. Have you seen organizations in your past succeed with this efficiency aspect or struggle with it? Subramanian Iyer: No, I think generally everybody struggles. And this was a lot of what I used to do in my previous role was essentially helping RSI partners get to the root of the problem. Right? And unfortunately when it comes to tuning spark jobs, there is no easy answer. You have to get in there and it's a very manual, very expert oriented effort. So that's something where there is a lot of room for us to improve and I think Unravel fits in perfectly. We definitely go there in the demo. I'll make one more appeal. I think as of course companies need to do cost optimization to save dollars. I'll say this, I think humanity needs to start consuming less resources. We see all this craziness about climate change stuff happening and with LLMs it's just getting worse. The energy demands of LLMs are intense, right? It was an article I was reading that said Microsoft wants to install nuclear reactors to power data centers. Unless we blanket the Earth with the solar panels, there's not going to be enough power. Subramanian Iyer: So yeah, so optimize your jobs, use every tool necessary to cut that and then A, save dollars in the short term. And of course there's an altruistic payoff in the end which is you're saving the earth. Sanjeev Mohan: I think Subu, you can get me started. We can turn this Unravel webinar into a TED talk because you've really touched a raw nerve. I think we should do another talk with Clinton on sustainability because sustainability is a really important topic. Clinton Ford: Now Sanjeev, if you have an LLM or generative AI, is that a guarantee of profitability? We're hearing about some big name companies out there creating these amazing models. For example, we see things like copilot. Is that guaranteed to be a success now that they've built it? Sanjeev Mohan: Not yet, but I'm a firm believer we will get there. I feel we are at the same stage when the printing press came out. So imagine what would've happened if feed transports back to the 1800's whenever printing presses came out, Gutenberg, Bible times. Do you think a printing press was a success when it came out? I don't think so. How could it be? Who would be the consumer of a printing press product when the majority of the people had never been taught to read or write because there was nothing to read or write. So it would've taken a few decades, maybe a whole generation before people started reading printed material because now they were being taught in school. Generative AI, like here, is it over hyped or not? I think it may be under hyped, and maybe I'm a very strong believer in the force of technology's progress and who knows? But the fact is that right now people are not productive. I learned something phenomenal last week. I was at an event, I met some people from a big pharma company, one of the top five pharma companies in the world. Everybody knows them thanks to COVID. So the CEO has mandated every business unit head must have production AI use cases in a short amount of time. And these AI production level use cases cannot be Mickey Mouse. Once like frequently asked questions, summarize it. What this company has done is they've bought 10 million medical journals and they basically want to have LLM go through this and try to find are there any commonalities between these different documents from WHO, CDC, New England Journal of Medicine and so on and help find diagnosis, new remedies, new cures, new medicines, reduce the FDA approval time by half. These are all use cases that will generate an ROI. Are we there yet? No, but are we going to get there? Absolutely. It's just a matter of time and we are moving faster than we ever expected. Like last year we won't even be having this conversation. This concept didn't even exist, but today it does and it's moving at a rapid pace. Clinton Ford: And Don, from your perspective, what do you think organizations need in order to scale? So as we start to think about implementing at that level that Sanjeev described, what is necessary to put in place in order for an organization to have that capability to be able to start producing AI models and revenue at that scale? Don Hilborn: Well, I think to some extent these observability platforms because what you can't measure, you can't ultimately control because you can't determine a baseline and a delta. And so you need to be able to take that holistic view of your ecosystem and I know that's an overused term, maybe landscape and take that data and use it to baseline and then determine are you able to scale and if you aren't, where are the problems that you aren't able to scale? The great thing about the cloud is it does build in this ability to scale, but it's not always, like you were saying earlier, it's not a guaranteed win just because the clouds have this ability to scale. You've got to be able to step back and look at things holistically and look at some key KPIs and determine how you're doing and where you're scaling well and where you're not, and then being able to go back and adjust those areas where you're not. It could be things like resources, you don't have enough data scientists, you don't have enough data engineers or it could be things like processes. Perhaps your processes are overly burdensome, overly bureaucratic and that's why you're struggling with scaling. But it all starts with the ability to see what's going on, make improvements and monitor and measure whether or not those improvements are making a difference. And that's really at the heart of these observability platforms. Clinton Ford: Excellent. Subu, from your perspective, what have you seen organizations who do this successfully at scale do in order to make this repeatable, to make this success happen over and over again? Subramanian Iyer: Great question. I think it is a classic process for people and tools, right? So to me it is having the right people. Number one. You need to amplify that effect and that's where tools come in. And then training the masses comes in. So like a train, the trainer model or some sort of enablement mechanism is going to make a difference. I think as a programmer, I spent many, many decades programming. I'm a lazy programmer. Okay? My first thing is if a tool can do it, let's bring the tool. So over the decades I have a quiver of tools I will go to every single place with, right? So if you ask me to build a data platform, a data estate, my first tool of course is going to be Databricks. So that's my first preference. Now of course you've got to manage the costs on the Lakehouse and instill confidence and my second tool is going to be Unravel, no surprise. Right? And even within Databricks and Unravel of course, I mean this can be its own webinar, but I have... You pick a use case and I will tell you what features to use and what not to use, what the patterns are, what the antipatterns are. So long story short, pick the right tools, get the right people, you don't need an army with the right tools. And then build on that. Clinton Ford: And then Sanjeev, from your perspective, how do you shift this concept of efficiency left into the developer's mindsets as they're building so that you get ahead of the cost early on? Have you seen companies do this well? Are there any best practices that you think should be repeated? Sanjeev Mohan: Clinton, that's a great question. Shift left is one of my favorite things to do and shift left basically is for everything, data quality. The sooner you find there's a problem, you fix it, the cheaper it is for you downstream. Security, the sooner you fix security, the safer you will be downstream. So same thing for observability. What has really shifted, and this comes down to data culture, is that it used to be... We started this webinar by Subbu and I were sort of converging on this Cartesian joint example I gave. At that time FinOps, it wasn't called FinOps, but the whole financial thing was CFO's responsibility. Today it's everybody's responsibility. Data developer needs to know what is going to be the cost of what they're about to start. And a lot of things I know you support BigQuery for instance. BigQuery can tell you before the query runs that this query is going to cost you X amount of dollars. So that responsibility and accountability for fiscal discipline is now throughout the teams. It's not no longer just a CFO. It'll be too late if the CFO or the finance team is the only team responsible for it. Clinton Ford: I'd like to thank each of you for your participation, Subu, Sanjeev, Don in this amazing event and thank you so much for your insights. So let's jump into a demo with Don of the product. Don Hilborn: Yeah. All right, thanks Clinton. Hello everyone, this is Don Hillborn and we'll be talking about the Unravel data platform today. I'll be walking you through a demo of that platform. If you can imagine what Unravel does for you, I like to draw an analogy between these devices that they have. You can plug into your car and it tells you all the times that you're driving aggressively or you've taken a corner too quickly as well as it makes recommendations on how you can improve your driving. Well that's a little bit of how Unravel works. Unravel installs these very lightweight agents into your data landscape. Those agents send data back to the application layer of Unravel and it organizes that data and passes it through AI and ML algorithms to make recommendations for you ultimately on how you can drive better or operate your data landscape better. And of course one of the areas that it looks at is cost. And that's what we're looking at right now is we're looking at some trends in cost in this fictional environment that I have pointed this to a Databricks environment in this case. And so you'll see that I can look at cost by a given workspace. I also can look at it based on a range of data focusing on particular days, but I also have the ability to look at costs in other ways. I can look at cost from a chargeback perspective where perhaps I'm looking at the users of the system, not just the workspaces, and I can see who's using the system the most. And if need be, I can actually drill into that individual user and look for opportunities to make optimizations where I can look at all the different tasks that they've run within Databricks in this case. And I can focus on one of the jobs that they've run and actually go into that job and look for recommendations, what it's telling me that I should do to make this run better. And in this case it's telling me that I have a bottleneck, which is a contended driver and I also have potentially efficient issues as far as my sizing compared to the job that I'm running. Other ways that Unravel helps you control your environment, manage it, make sure that you're driving as efficiently as you can is through budgeting. So you're able to actually define budgets about or around your landscape and then monitor those on an ongoing basis. And in fact, here is another place where Unravel uses its AI ML capabilities to forecast or try to determine what your budget is ultimately going to be. So hopefully it can tell you before you've exceeded your budget by 987% that you're going to exceed your budget. And of course the hope would be if it tells you enough in advance, you could actually go in and make changes. You could look for particular trends or you could look for opportunities here again to optimize where it takes you into a given set of tests that have been run and a given set of jobs that have been run. And it allows you to drill into them and find recommendations potentially. So here again, we have another bottleneck. You also see that we have some very relevant data about the number of events that we're processing per batch. And in many cases we can actually go look at the resources that are being used. We can look at the configuration that existed at the time that this particular job ran, and we can look at different aspects of that configuration, look for areas again where we might be able to make improvements. We also allow you to look at the actual compute that you're using and we organize that into all of the jobs that have run, all the clusters that have stood up or the jobs that have run those that have finished those that are running, but probably most important view those that we see as being inefficient. And again, we have an opportunity to go in here and it makes some tuning suggestions to us. So we drill down into that. We see we have a contended driver and we also see that we have some opportunities to make our process run more efficiently. Another thing that Unravel does for you is it actually looks at how you are using your existing data stores, your tables, your blob storage and probably most valuable is to determine whether or not your storage could be used or needs to be used as the most expensive hot storage or could you use warm storage or in some cases even cold storage. So again, giving you an opportunity to optimize your overall environment, in this case focusing on storage. There are certain reports that come with the platform and then of course you can actually build out your own reports. One of the most popular ones is the top X report. So what are my greatest offenders? What applications are consuming the most of my resources? And being able to look at those or even schedule or rerun that report. Probably one of our most powerful aspects of the platform is this concept of auto actions. And here we're actually able to go in and based on an event, based on an event from a job or from a cluster, we can take action, we can send an email, we can also make an arrest API call so we can potentially reconfigure a cluster. We can look at something that tells us that this cluster might not be sized properly and we could make an AI call to decommission that cluster, reconfigure it, and then start that cluster up again and then run the job on that cluster. Generally most of our customers start out treading lightly here with emails, but as they grow more and more comfortable with the platform, they may build out some of those more advanced rest API calls. We also give you this ability to look at the insights in your system overall. So in this case we'll go back to the most history that we have, which is in July. And then we'll come up to today to look at the most current. And this really gives us a very high level overview of the cost, the resource efficiency, and then the app usage that we're benefiting from. What are some areas where we as Unravel can make recommendations to you to improve your environment? And here we're seeing some of those where you can go in and you can again drill into certain areas, get into the particular job, and ultimately drill down into some form of recommendation with some very detailed information about that particular run. That's a very high level overview of Unravel. There are other aspects of it that perhaps in another session we could drill into. We have this concept of our app store and something we call Unity apps where applications can be incorporated within the platform. And then of course we have the ability to diagnose our platform as well to help you see how the resources that we're using, which is very minimal, less than 1% on any particular node that we're running on, but it also allows you potentially to connect to other workspaces and do analysis on those workspaces as well. So with that, I'll close out the demo and I'll turn it back to you Clinton. Thank you. Clinton Ford: Well Don, thank you so much for that incredible demo. Really appreciate that. All right, so now we're going to transition into some Q and A. And Don if I could ask you the first question that we have from our audience participants here. So this question is about choosing the right architecture. So Don, the question is, if I don't already have an established architecture for my data platform, is there a right answer in terms of the best platform choice? Don Hilborn: Yeah, so I think a question I hear a lot is around architecture. What's the right architecture I should implement in GCP in Azure or in AWS? And I think one of the things you definitely have to consider when looking at an architecture is do you have the ability to manage and monitor that architecture? And if you don't, where are the gaps? Because the right architecture depends on the entity itself, what businesses it's in, what particular use cases it's addressing, and to be able to determine if you have the right architecture, you need an observability tool or something that gives you that holistic view of your environment so that you can determine is this architecture performing optimally for me given what I need to do with it? So I would say when considering architecture, there are certainly lots of things to consider. And one of those is do you have the ability going forward to monitor and manage that architecture? Clinton Ford: All right, it looks like we have a couple of other questions. I'll turn the next one over to you, Subbu, do you mind reading the question before you give the response? Thanks. Subramanian Iyer: We just got a question from the audience. This is the Databricks Lakehouse question, so I'll take it. And the question, and I'm paraphrasing here is can you give an example of where Unravel can make a significant difference in optimizing Databricks and Lakehouse? And I'll give an anecdote here. One of the most difficult situations tuning a Spark job is catching data skew. And it typically happens when the job runs, runs, runs, and then it comes to 99% and then it fails at the very end. To catch data skew requires you to click through a bunch of things and then the symptoms will reveal themselves, right, and then you can diagnose it. Guess what? Unravel, because the two, and then of course before I go to Unravel, imagine hundreds of jobs over tens of workspaces, all potentially having data skew issues and performance challenges. And then you begin to get a sense of the scale of the problem. Now that's where a tool like Unravel will come in handy because it's monitoring all these workspaces, it's monitoring all the jobs, it's pulling out telemetry and it'll give you... As the issues happen, it'll start giving you alerts which will help you. You can have one expert who's looking at Unravel and then basically fixing things as they happen, which is huge, which will massively save cost and improve time to market essentially. Clinton Ford: All right, we have time for one more question. Sanjeev, do you mind taking the last question? Read that out before you provide the answer. Thank you. Sanjeev Mohan: The question that just came in was you gave some examples of how to do cost containment for generative AI. Are there any other examples you can share? So one that comes to my mind that I'd like to share is that we've had this concept of caching for the longest time in our data space. Now the same concept exists even in generative AI. For example, if salespeople are always asking the same question in the Chatbot, then that question does not need to incur a very expensive call to open AI LLM. And we also know that it's slow. It takes 10 seconds for the response to come back. If we build generative AI applications correctly, then when we get these questions and we create a vector embedding, we can compare and see if there's already a response from an LLM, then we can serve it from a vector database instead of making an open AI API call. And these are the things that the data usability products can incorporate, they can discover and intelligently route the queries to wherever it is most cost effective. Thanks for asking that question. Clinton Ford: Well, that concludes today's webinar. Thank you so much to everyone who joined us and attended. We'd like to also thank our speakers, Sanjeev Mohan, principal and founder at SanjMo, Subramanian Iyer, Unravel Training and Enablement Leader and Databricks Subject Matter Expert at Speedboat and Don Hillborn, Field CTO at Unravel Data. Thanks to each of you and we look forward to seeing you at our next event. Meanwhile, be sure to go ahead and sign up for your free Unravel account for Databricks and reach out to us if you have additional questions. Thank you so much. Register Now FIRST NAME * Enter your first name LAST NAME * Enter your last name WORK EMAIL * you@your.email COMPANY NAME * Company JOB TITLE * Job Title DO YOU HAVE A QUESTION YOU'D LIKE ANSWERED IN THE Q&A? ASK BELOW! I'd like to learn more about Unravel Data in a 1:1 demo. By submitting, I agree to the processing of my personal data in accordance with the Unravel Data privacy policy. I understand I can update my preferences at any time. REGISTER NOW REGISTER NOW Text goes here X Your Hosts Jake Hernandez Chief Creative Officer Openform Jake Hernandez is senior director of product marketing at Openform, where his goal is to help people unlock the power of in-person events. Leo Di Salvo Sr. Technology Consultant Openform Leo Di Salvo is the Senior Technology Consultant at Openform, where his goal is to help people unlock the power of in-person events. Run of show 12:00PM Webinar Begins Opening remarks from your hosts 12:15PM 5-Tips 5 Tips for Creating Virtual Events 12:45PM Closing Final remarks from your hosts Grab your favorite snack and tune in. Pro Tip: First, watch the video to learn how to use this type of theme. Then, replace the iFrame embed code of the video below by clicking into it and clicking on the "Embed Options" tab on the right. Hide this element when you're finished. Need more help? Here are some great Help Center articles regarding embed best-practices: How do I resize an iFrame? What's the difference between iFrame elements and Video elements? How do I embed webinars or livestreams using an iFrame element? What can I embed using an iframe into Splash? Resource List 01 A Guide to Agile Event Programs that Can Survive Anything Explore why the most dynamic leaders are building strategies for agile event marketing with a focus on scaling in-person and virtual events that are fast, flexible, and hyper-effective at driving meaningful connections. Download Text goes here X 02 Crisis Mode: Ideas for Managing Events in Challenging Situations Learn strategies, experiences and lessons learned from top event marketing leaders who have walked through the fire more than once and lived to talk about it. Register Text goes here X 03 How to Build a Rock Solid Agile Events Program The old way of doing event marketing—a large, inflexible investment in a small number of tentpole events—doesn't fit into a business world that demands adaptability and speed. Watch now Text goes here X Our Sponsors See you online. © Unravel Data Chapter 0 Text goes here X Chapter 1 Text goes here X Chapter 3 Text goes here X Lucy Hernandez Chief Creative Officer Openform Lucy Hernandez is senior director of product marketing at Openform, where her goal is to help people unlock the power of in-person events. Prior to Openform, she led global product marketing teams. [confirmation_headline] [confirmation_messaging] Watch Now Text goes here X Add to Calendar Google Outlook Apple Yahoo Skip to content Share with Friends Facebook Twitter LinkedIn CONTACT THE ORGANIZER Google Outlook iCal Yahoo RSVP Google Outlook Apple Yahoo