Snowflake Data Cloud Summit 2024 | Josh Klahr | Snowflake

Clips
News
More from Snowflake Data Cloud Summit 2024

Josh Klahr

Head of Data Warehouse

Snowflake

Cost management and visibility play central role in this week’s Snowflake announcements

While Snowflake Inc. has orchestrated its platform to run with ease and simplicity for users, there is a subset of customers with an interest in cost management and performance tracking on a regular basis.At Data Cloud Summit in San Francisco this week, the company unveiled a set of tools designed to provide those insights.Snowflake’s Josh Klahr talks about cost management with theCUBE.“The way we think about platform performance is first: How do we make it drop dead simple to just run your workload and not even think about it working?” said Josh Klahr (pictured), head of data warehouse at Snowflake. “But there are some cases where customers need extra visibility

play_circle_outline Snowsight Introduces Innovative Cost Management Experience and Budgeting Feature, Alongside Enhanced Platform Visibility and Performance

play_circle_outline Use of AI and machine learning to generate recommendations for cost reduction and performance improvement

play_circle_outline Importance of security and governance in customer data

Info

Josh Klahr | Snowflake

Josh Klahr

Head of Data Warehouse Snowflake

search

>> Welcome back everyone to theCUBE's live coverage of the Snowflake Data Cloud Summit here at the Moscone Center in San Francisco, California. I'm your host, Rebecca Knight. Dave Vellante will be joining me shortly. I'd like to introduce our next guest to the segment. He is Josh Klahr, the head of data warehouse at Snowflake. Thank you so much for coming on the show, Josh.

>> Thank you for having me.

>> So very exciting to be talking to you about really what is new with Snowflake's platform. Why don't you walk it through for our viewers.

>> Sure. I'm happy to. Just as set context, I'm on the product management team and I really look after into our core query platform focused on how do we get out of the box performance and how do we provide visibility to customers in terms of how the platform is performing and ideally making it as easy as possible for them to get their work done as efficiently as possible. The way we think about platform performance is first, how do we make it drop dead simple to just run your workload and not even think about it working? So there's a bunch of stuff we do around automated performance that we focus on, but there are some cases where customers need extra visibility. So they want to know, am I getting optimal performance? Which queries are running slow? Which warehouses are consuming a lot of credits? At Summit this week, we announced a bunch of improvements around really visibility and performance and cost. And then finally, we've just started to introduce a set of things around proactive insights for customers to help them identify, "Hey, we saw this pattern in your workload. Maybe you should consider making configuration change X or Y to really boost performance."

>> Okay. So let's dig into all of those. So talking about the visibility in terms of making it faster and easier for customers to really see what is going on. Talk a little bit about some of the enhancements you've introduced.

>> The big canvas that we announced this week at Summit was the cost management experience in Snowsight. Snowsight is our end user facing UI for people who are managing Snowflake. And the cost management console starts with, at my organization level, how much money or how many credits am I spending and what do those trends look like? So simple stuff like am I within budget for my next quarter or my next month in terms of spend? So we start with that high level visibility and then we break that down into a set of tiles that let customers know which warehouses are the most expensive, which queries may be the most expensive queries, who is that user that ends up running that job that is really, really expensive, and how do I go talk to that user and help them improve performance? And in that cost management console, we've announced a budgeting feature. And what budgeting allows you to do is set a budget on any resource in Snowflake that consumes credits. That might be a warehouse, that might be a clustering service. Once you've set that budget, we'll monitor it over time, we'll send you alerts if we think you maybe... If you think about the app you have on your phone that says, "Hey, you're about to spend too much money on your credit card." We're doing the same thing for Snowflake to help customers get that visibility.

>> Well, at a time where there is so much pressure on companies to reduce spending in departments and in these inflationary times where there's a lot of uncertainty in terms of where the economy is going. Is this something that you heard directly from customers, "Hey, we need to have a little better understanding of where our money is"?

>> This is kind of like a bit of a double-edged sword with Snowflake. Because Snowflake performs so well because it's so easy to use when we have customers, for example, customers in the financial services industry who are often our customer in those large investment or banking institutions or often a central group. They're the COE for Snowflake and they're supporting a bunch of different business units. And so what they really need to be mindful of is as we start to roll out these capabilities to the different business units, how do we make those business units feel comfortable that the costs and controls are in place to make sure that spending doesn't get out of control? And it's a mix. You want to make sure that you're enabling the business units to do the analysis they want to do, but you need that visibility and control to give the managing directors comfort with their budgeted spend.

>> Exactly. And the more you spend, the more you save. We know that. But in terms of those patterns and the trends you were talking about, I mean, I'm imagining that that is so empowering to companies to understand, okay, this is what is happening here. Here's where our data is, this is what's going on. Can you talk a little bit about the feedback you were hearing from customers in terms of specifically what they wanted to know and how the enhancements to the platform really answer their questions?

>> Yeah, well definitely I would say the number one thing is visibility. It's funny, I was just walking by one of the booths here and there's a quote about dashboards being dead, but I have to laugh because at the end of the day, we all want that basic visibility. And so the first thing we heard from customers is just give me visibility into those credit consuming areas of spend. But what happened immediately after that as we started in previews to roll out these capabilities is customers wanted to know, okay, well now you've identified that this is maybe an expensive query or maybe this table is generating a bunch of credits when it gets reclustered, help me identify actions I can take to maintain performance maybe while reducing the cost. So one of the things we just launched this week is a new capability we call Cost Insights. What Cost Insights does is it proactively goes, looks within the customer's environments, identifies high sources of spend or high areas of bad performance, and it makes recommendations to customers. A very simple example might be, here's a table that you've been storing that's over 100 terabytes and it hasn't been queried in the past month, and that happens over time. As the deployment grows, there's a bunch of objects that get created. And so just like that budgeting app on your phone that tells you you're spending a lot, we now give recommendations on how you might spend less or how you might improve performance.

>> And how are you generating those recommendations? I mean, I'm imagining it's AI powered.

>> It is definitely lots of AI and machine learning in the background. And this is something we've done since the beginning of Snowflake is, and you heard it at the keynote, this idea of simplicity. We want to make sure that we use the data we have at our disposal. We want to make sure we're providing recommendations and automation for customers using ML and using AI so that they don't have to do the heavy lifting and we can do the thinking for them

>> Over the course of this Data Cloud Summit, we've had a lot of great Snowflake customers on the show talking about the challenges that they're grappling with and how Snowflake solutions have really helped them get a handle on their data and wrangle their data. I love wrangled data. I think about a cowboy with a lasso, but can you talk a little bit about why having this built in security and governance is so critical to customers in all kinds of different industries?

>> Well, I think I would say on the cost and performance management side, it's kind of a level of comfort and confidence that you're doing the best. One of the breakout sessions we have at every Snowflake Summit I've been to is called Optimizing Cost and Performance with Snowflake, and there is always a line out the door. In fact this year, we had to do three of those sessions because it was so oversubscribed. And so customers just want the confidence that they're following best practices, they're doing things right on the performance side. And then when it comes to governance and security, the data that is in Snowflake are the crown jewels of our customers. It helps them power innovation, it helps them stay ahead of the competition and making sure that data is governed and secure, that the right folks have the right access to it is super critical. And one of the, I would say, anti-patterns that we want to help customers avoid is the proliferation of data into different systems. If you don't have a single system and you have to start implementing governance and controls in multiple data systems, if you're moving data around because you need multiple platforms for multiple workloads, that's a real challenge. And so being able to support those multiple workloads in Snowflake and then have a unified governance posture on top of that ends up unlocking a lot of capabilities and access that would be hard to do if you had a dispersed siloed system.

>> Take the temperature for us, Josh, here. So this is such a special moment in time where we know according to Sridhar, the moment of enterprise AI has arrived, the era for it, and yet customers, there is still a lot of, maybe not fear, but caution certainly around AI. They've moved from the experimental phase, they're trying to figure out exactly how it's working in their organizations, and yet, as you've already described, they're also under tremendous pressure to save money and cut costs and understand what's going on. How do you approach this moment with customers that you're working with and sort of hold their hand through all of these challenges that they're dealing with?

>> That's a great question, and I think it is a special moment in time. It's exciting to be living through it as someone who has been working in data for a long time. I think enabling our customers to take the small steps to incrementally start deploying that first app that is maybe not just doing BI or analytics, but is actually doing AI, a co-pilot experience on top of data that you are already comfortable with and is already in Snowflake. That's part of the journey we're helping our customers go on is how do you take that next step, but do it within the confines of an environment you're comfortable with, do it within the security confines, do it within data that's already been curated and loaded into Snowflake. If you attended our keynote on Tuesday, you saw Christian pull someone out of the audience who had never used Snowflake before. I think she said she had logged in three times.

>> Couple times, yeah, yeah.

>> And in five minutes she walked through a click button to build a chatbot. And to me, while that feels like a small use case, I think it's a really good example of how we can help customers go on those small first steps to generate that first successful POC and essentially go on the journey into building more and more powerful AI and LLM-powered apps.

>> I was sitting next to my friend who's an engineer, and she said, "I'm going to be out of a job soon. I mean, this is too easy."

>> Yes. I think all of us were like, "Oh, geez. It's easier than I thought." But yeah, I think we're in a unique position, which is our customers are comfortable with their enterprise data and Snowflake. They already are sitting on this corpus of really valuable data, and if we enable new interfaces and use cases against that same data, it makes it a lot easier than spinning up some separate environment to try the new thing.

>> Exciting times. Josh, thank you so much for coming on theCUBE. A real pleasure having you.

>> Thank you for having me.

>> I'm Rebecca Knight. Stay tuned for more of theCUBE's live coverage of the Snowflake Data Cloud Summit. You're watching theCUBE, the leader in enterprise tech news and analysis.

Josh Klahr | Snowflake

search

>> Welcome back everyone to theCUBE's live coverage of the Snowflake Data Cloud Summit here at the Moscone Center in San Francisco, California. I'm your host, Rebecca Knight. Dave Vellante will be joining me shortly. I'd like to introduce our next guest to the segment. He is Josh Klahr, the head of data warehouse at Snowflake. Thank you so much for coming on the show, Josh.

>> Thank you for having me.

>> So very exciting to be talking to you about really what is new with Snowflake's platform. Why don't you walk it through for our viewers.

>> Sure. I'm happy to. Just as set context, I'm on the product management team and I really look after into our core query platform focused on how do we get out of the box performance and how do we provide visibility to customers in terms of how the platform is performing and ideally making it as easy as possible for them to get their work done as efficiently as possible. The way we think about platform performance is first, how do we make it drop dead simple to just run your workload and not even think about it working? So there's a bunch of stuff we do around automated performance that we focus on, but there are some cases where customers need extra visibility. So they want to know, am I getting optimal performance? Which queries are running slow? Which warehouses are consuming a lot of credits? At Summit this week, we announced a bunch of improvements around really visibility and performance and cost. And then finally, we've just started to introduce a set of things around proactive insights for customers to help them identify, "Hey, we saw this pattern in your workload. Maybe you should consider making configuration change X or Y to really boost performance."

>> Okay. So let's dig into all of those. So talking about the visibility in terms of making it faster and easier for customers to really see what is going on. Talk a little bit about some of the enhancements you've introduced.

>> The big canvas that we announced this week at Summit was the cost management experience in Snowsight. Snowsight is our end user facing UI for people who are managing Snowflake. And the cost management console starts with, at my organization level, how much money or how many credits am I spending and what do those trends look like? So simple stuff like am I within budget for my next quarter or my next month in terms of spend? So we start with that high level visibility and then we break that down into a set of tiles that let customers know which warehouses are the most expensive, which queries may be the most expensive queries, who is that user that ends up running that job that is really, really expensive, and how do I go talk to that user and help them improve performance? And in that cost management console, we've announced a budgeting feature. And what budgeting allows you to do is set a budget on any resource in Snowflake that consumes credits. That might be a warehouse, that might be a clustering service. Once you've set that budget, we'll monitor it over time, we'll send you alerts if we think you maybe... If you think about the app you have on your phone that says, "Hey, you're about to spend too much money on your credit card." We're doing the same thing for Snowflake to help customers get that visibility.

>> Well, at a time where there is so much pressure on companies to reduce spending in departments and in these inflationary times where there's a lot of uncertainty in terms of where the economy is going. Is this something that you heard directly from customers, "Hey, we need to have a little better understanding of where our money is"?

>> This is kind of like a bit of a double-edged sword with Snowflake. Because Snowflake performs so well because it's so easy to use when we have customers, for example, customers in the financial services industry who are often our customer in those large investment or banking institutions or often a central group. They're the COE for Snowflake and they're supporting a bunch of different business units. And so what they really need to be mindful of is as we start to roll out these capabilities to the different business units, how do we make those business units feel comfortable that the costs and controls are in place to make sure that spending doesn't get out of control? And it's a mix. You want to make sure that you're enabling the business units to do the analysis they want to do, but you need that visibility and control to give the managing directors comfort with their budgeted spend.

>> Exactly. And the more you spend, the more you save. We know that. But in terms of those patterns and the trends you were talking about, I mean, I'm imagining that that is so empowering to companies to understand, okay, this is what is happening here. Here's where our data is, this is what's going on. Can you talk a little bit about the feedback you were hearing from customers in terms of specifically what they wanted to know and how the enhancements to the platform really answer their questions?

>> Yeah, well definitely I would say the number one thing is visibility. It's funny, I was just walking by one of the booths here and there's a quote about dashboards being dead, but I have to laugh because at the end of the day, we all want that basic visibility. And so the first thing we heard from customers is just give me visibility into those credit consuming areas of spend. But what happened immediately after that as we started in previews to roll out these capabilities is customers wanted to know, okay, well now you've identified that this is maybe an expensive query or maybe this table is generating a bunch of credits when it gets reclustered, help me identify actions I can take to maintain performance maybe while reducing the cost. So one of the things we just launched this week is a new capability we call Cost Insights. What Cost Insights does is it proactively goes, looks within the customer's environments, identifies high sources of spend or high areas of bad performance, and it makes recommendations to customers. A very simple example might be, here's a table that you've been storing that's over 100 terabytes and it hasn't been queried in the past month, and that happens over time. As the deployment grows, there's a bunch of objects that get created. And so just like that budgeting app on your phone that tells you you're spending a lot, we now give recommendations on how you might spend less or how you might improve performance.

>> And how are you generating those recommendations? I mean, I'm imagining it's AI powered.

>> It is definitely lots of AI and machine learning in the background. And this is something we've done since the beginning of Snowflake is, and you heard it at the keynote, this idea of simplicity. We want to make sure that we use the data we have at our disposal. We want to make sure we're providing recommendations and automation for customers using ML and using AI so that they don't have to do the heavy lifting and we can do the thinking for them

>> Over the course of this Data Cloud Summit, we've had a lot of great Snowflake customers on the show talking about the challenges that they're grappling with and how Snowflake solutions have really helped them get a handle on their data and wrangle their data. I love wrangled data. I think about a cowboy with a lasso, but can you talk a little bit about why having this built in security and governance is so critical to customers in all kinds of different industries?

>> Well, I think I would say on the cost and performance management side, it's kind of a level of comfort and confidence that you're doing the best. One of the breakout sessions we have at every Snowflake Summit I've been to is called Optimizing Cost and Performance with Snowflake, and there is always a line out the door. In fact this year, we had to do three of those sessions because it was so oversubscribed. And so customers just want the confidence that they're following best practices, they're doing things right on the performance side. And then when it comes to governance and security, the data that is in Snowflake are the crown jewels of our customers. It helps them power innovation, it helps them stay ahead of the competition and making sure that data is governed and secure, that the right folks have the right access to it is super critical. And one of the, I would say, anti-patterns that we want to help customers avoid is the proliferation of data into different systems. If you don't have a single system and you have to start implementing governance and controls in multiple data systems, if you're moving data around because you need multiple platforms for multiple workloads, that's a real challenge. And so being able to support those multiple workloads in Snowflake and then have a unified governance posture on top of that ends up unlocking a lot of capabilities and access that would be hard to do if you had a dispersed siloed system.

>> Take the temperature for us, Josh, here. So this is such a special moment in time where we know according to Sridhar, the moment of enterprise AI has arrived, the era for it, and yet customers, there is still a lot of, maybe not fear, but caution certainly around AI. They've moved from the experimental phase, they're trying to figure out exactly how it's working in their organizations, and yet, as you've already described, they're also under tremendous pressure to save money and cut costs and understand what's going on. How do you approach this moment with customers that you're working with and sort of hold their hand through all of these challenges that they're dealing with?

>> That's a great question, and I think it is a special moment in time. It's exciting to be living through it as someone who has been working in data for a long time. I think enabling our customers to take the small steps to incrementally start deploying that first app that is maybe not just doing BI or analytics, but is actually doing AI, a co-pilot experience on top of data that you are already comfortable with and is already in Snowflake. That's part of the journey we're helping our customers go on is how do you take that next step, but do it within the confines of an environment you're comfortable with, do it within the security confines, do it within data that's already been curated and loaded into Snowflake. If you attended our keynote on Tuesday, you saw Christian pull someone out of the audience who had never used Snowflake before. I think she said she had logged in three times.

>> Couple times, yeah, yeah.

>> And in five minutes she walked through a click button to build a chatbot. And to me, while that feels like a small use case, I think it's a really good example of how we can help customers go on those small first steps to generate that first successful POC and essentially go on the journey into building more and more powerful AI and LLM-powered apps.

>> I was sitting next to my friend who's an engineer, and she said, "I'm going to be out of a job soon. I mean, this is too easy."

>> Yes. I think all of us were like, "Oh, geez. It's easier than I thought." But yeah, I think we're in a unique position, which is our customers are comfortable with their enterprise data and Snowflake. They already are sitting on this corpus of really valuable data, and if we enable new interfaces and use cases against that same data, it makes it a lot easier than spinning up some separate environment to try the new thing.

>> Exciting times. Josh, thank you so much for coming on theCUBE. A real pleasure having you.

>> Thank you for having me.

>> I'm Rebecca Knight. Stay tuned for more of theCUBE's live coverage of the Snowflake Data Cloud Summit. You're watching theCUBE, the leader in enterprise tech news and analysis.