Good morning, and welcome. Thank you very much for joining us here in San Francisco, as well as all of you out on the webcast. We appreciate you being here for the launch of our new product, the Xeon E7 v3, targeted at data analytics and real-time processing. As many of you may know, we just celebrated the 50th anniversary of Moore's Law, and Moore's Law has truly made big data viable. Moore's Law has given us the ability to store massive amounts of data, given us the ability to grow memory capacity that's required to operate on those massive data sets, and Moore's Law has given us the compute capacity to execute real-time complex analytics. The rise of the digital service economy has sparked tremendous innovation in new services and has also enabled the improvement of existing services.
It's made those services more efficient, driving bottom-line growth profitability, and it's made those services more valuable to the end user, to the customers driving top-line value. Data is the new currency. It's the currency of the digital world. It's so amazing, in fact, that data is deemed now the new bacon. These T-shirts, for those of you here in San Francisco, these T-shirts will be available for you. We want every one of you to have one. They come in two form factors, women's and men's, in all sizes, small, medium, and large. Please grab your data's the new bacon T-shirt on your way out. We talk about data as the new bacon because adopting big data analytic solutions allow businesses to have a true competitive advantage. It's also, though, clear that all data is not equal.
There is a time value on data, and the faster you can aggregate and analyze and take action, the greater the results. There isn't an industry or a business that isn't able to benefit from data and data analytics. Every industry including one that was founded in 10,000 BC, can be transformed with big data analytics, including farming. Let's take a look.
Farmers are faced with this incredible challenge of increasing food production by nearly 50% by the year 2050. We have to do all of this without getting any more farm land. In fact, we actually lose farm ground every single year.
Part of the challenge is the complexity, because we're trying to warn people about earthquakes as soon as they happen and before the shaking reaches them, and that gives us just a few seconds to work with. The Pacific Northwest Seismic Network has several hundred seismometers, and it's tied into a number of GPS stations, and we watch the earthquakes and look at the danger of the earthquakes and volcanoes in the states of Oregon and Washington.
FarmWise is all about helping farmers use data sets to make better decisions. We take data from hundreds of different sources and bring it all together and analyze it for the grower. The data center is actually enabling us to do things that just a few short years ago would have never been possible without access to supercomputing-like resources. We've seen the company go from nothing to over 20% of the farms in the country using our technology today in just three short years.
From real-time analytics, we're able to generate the earthquake early warning. We look continuously at the data from hundreds of seismometers, analyzing them to detect the earthquakes, and when earthquakes happen, to estimate their size and the impact that they'll have.
Our customers are able to see everything happening across their farm in real time, such as harvest data flowing in, growing conditions, vegetative health.
Our programs are designed to reduce the damage from earthquakes. We can't eliminate it, but we can provide people with enough warning to take some preventive steps, so it'll be able to save money and lives and protect the economy of the region in the case of a large earthquake.
By helping farmers eliminate waste and produce more food, we're able to make this massive impact on the entire planet.
As exciting as the proof points of big data analytics are, there are still clearly hurdles. The first is around trust. Data management has always been the Achilles' heel of IT, but with this overwhelming amount of data that now is streaming into IT, whether it be increased number of devices per employee, whether it be all the social media solutions or the instrumentation of businesses in support of the Internet of Things, all of this is increasing the challenge of data security, both confidentiality of the data as well as integrity of the data. The second issue is speed. The analysis of enterprise data has historically been batch mode. For instance, you're analyzing your customer bookings every day or so as the quarter comes to an end. But those legacy systems don't hold up in the new era of real-time analysis and real-time action.
The third challenge is scale. Obviously, a challenge compared to the legacy world. Not only are the data sets becoming much larger, but you actually want the data sets to grow and become much, much larger because the larger the amount of data, the greater the insight that you're going to achieve. The unfortunate fact is that those early adopters of big data analytics solutions, only 27% of them will say that those big data analytics solutions have actually delivered success. The good news is that all of these challenges have technology-based solutions. It starts by deploying a solution that supports the complementary architecture of scale-out and scale-up. Meeting the needs of speed with the needs for massive data sets. Scale-out, adding more and more capacity to deliver greater performance using a distributed processing approach.
For instance, Hadoop is a great example of a distributed data storage processing solution that can process large bulk jobs with computing occurring at the data. Scale-up, so adding ever-increasing resources to a single node. Achieving maximum performance through higher core count, through very large memory capacity, and through symmetrical multiprocessing software algorithms. SAP HANA is a great example of a scale-up application. To deliver a step function improvement in the performance delivered in scale-up applications, we are introducing today the new Xeon E7 v3. The performance gain of the new processor is significant, especially for those applications that can take advantage of the new instructions, and I'll talk about those new instructions in a minute. Thanks to Moore's Law, we're able to pack now 18 high-performing cores onto a single die. There's two threads per core.
We support two-socket, four-socket, and eight-socket solutions natively. Thanks to the innovation of our customers through their own memory controllers, the system scales out to 32 sockets. Each processor supports one and a half terabytes of memory. For an eight-socket solution, that gets you to 12 terabytes of DDR3 or DDR4 memory. We continue to deliver higher and higher levels of both security and reliability, consistent with what you would expect for a mission-critical solution. For data confidentiality, our encryption instructions, which are AES-NI, we've improved those in this processor version. Both reducing the latency as well as increasing the bandwidth, delivering a 2X improvement in data encryption. IT now can insist upon data being encrypted both in transit and at rest because the performance of the solution is so high, the impact is transparent to the end user.
To ensure data integrity, we've increased the highly computationally intensive process of creating hash values for data authentication. We've increased the performance up to 2X, and we've done that through new instructions around AVX2 as well as adding additional, a fourth ALU solution. For reliability, we have numerous reliability solutions that we've added into this processor generation. Enhanced memory check architecture recovery. We've added more information that gets logged, that gives information about what actually generated the error, increasing the probability of recovery. We've added address range memory mirroring, so you have finer granularity of the memory that's going to be mirrored, so you get better utilization out of the total memory capacity. We added a spare rank behind the memory controller. Again, increasing the probability of uptime. We've also added parity checking for parity on address and command lines as well.
Again, increasing system uptime. A new feature that we're very excited about is new instructions called TSX. TSX provides higher performing for concurrent software with a lower software investment required. What it does is it moves the effort of optimizing memory locking, which is a task required by multithreading applications. It moves that complex process from software down into the hardware. To tell you a little about how this works. Without the new instructions, the first thing that an application thread does when it goes out to access data is to issue a lock. That lock will likely protect a large range of data because the specific thread doesn't know which byte it's actually going to require.
That lock then placed on the memory block means that all other threads are going to have to sit idle waiting for that block to be removed. Waiting in a business-critical application is obviously a bad thing to do. If you were to ask any application developer, he or she will tell you that optimizing thread synchronization is the most complex portion of developing an application. We actually worked with one large enterprise software vendor. We worked with their application development team on that locking algorithm, and eventually they gave up. When we're talking about threads, 60 threads and more, trying to manage the memory space becomes very complex. What we've done is added the new TSX instruction.
This now says that the hardware will automatically determine what bytes the threads are actually accessing, and it allows just the blocking of those specific bytes and all other threads can continue without waiting. You have multiple threads now able to access a given memory block region or memory region simultaneously. This capability, we've delivered it into our standard libraries that most application developers are already using today. It's a very simple, seamless way of integrating TSX instructions into the application. As we noted TSX instructions can deliver up to a 6x improvement in performance on multi-threaded applications. It truly is a breakthrough in managing parallel workloads, not only because the performance gains are so outstanding, but because we've greatly simplified the application development path.
We are proud to announce that with the new Xeon E7 v3, we have 20 new world records, and we want to thank, obviously, our customers for delivering the innovation that results in those breakthroughs. They represent a wide variety of applications from database to virtualization to business processing, to name just a few. You have obviously, over time, grown to expect that we have industry-leading performance on four sockets with the Xeon E7 processor. We're also very proud to say that with the new Xeon E7 v3, we have taken the leadership position at eight socket as well, taking that leadership from POWER8. We now have leading performance on two of the most standard industry benchmarks, SPECint_rate and SPECfp_rate, and those come to us through the work of Huawei.
I thought it would be helpful to show you a demonstration of how this is actually working and being applied to the energy industry. I'd like to ask Lisa Spelman to come out and give us a demonstration. Lisa is the general manager of the Data Center Marketing Organization . Lisa. Hey, Lisa.
Hello.
Thank you. Come on out. Tell us about how we've applied E7 v3 to energy.
Sure. Thank you for having me.
Yeah.
I want to start first by showing you what we have here. Over here to my right, we have a new one of our four-socket servers with running our 18-core Xeon E7 v3 processor. Also includes Intel high-performing SSDs, as well as our 40 Gigabit network adapters. Looking over here, I've got the SaaS analytics dashboard running off this mobile workstation here, and I wanted to start showing you three things today. First, I think it's fun to look at a little raw performance here.
This is a SaaS application from a real energy company.
Yep
who won't allow us to tell you their name because of being a regulated industry, so they'll just remain anonymous. This is real data from an energy company.
It is real data. What we're showing here first is this comparison of a five-year-old system compared to what we have with today's newly launched processor. When you see what happens here, what we're showing is 100x the data in the same amount of time. If you start thinking about what you're now capable of and the granularity that you can start doing analytics and prediction when you can, in your same amount of time, get to 100x the data crunching. This is billions of rows being crunched in just that short little time zone. We love the world records, as you said, we love seeing stuff like this, I think it's better if we can translate it into business value and customer value that the utility customer is able to see and then deliver to their end users.
Okay, this is a SaaS analytics dashboard for those of you that aren't familiar and don't spend your day here. This utility company has 38,000 smart meters that they're pinging once a day, and they've been using that to drive a prediction line for the energy that they should produce.
This is historical. On the old system, once a day, they ping all of those meters to decide how much capacity they're going to need.
Exactly.
Okay.
The yellow line shows now that they have this new capability, they can ping every 15 minutes and still combine and crunch all that data and get a much more granular prediction. The blue line that's following behind is the actual. You can see it's a very tight correlation now that they can analyze that exponential amount of data. What does this do for them from a business perspective is this allows them to bring their daily energy production down so they're not producing to the top line, they're producing to what they actually need. As you guys know, energy is not something that's able to be stored. This results in a better earth as well as some better operational efficiency. Every 1% of forecast accuracy for this utility company delivers them $1 million of operational savings, real savings every year.
They're seeing $9 million in operational savings per year thanks to the new system and the fast analytics that runs on top of it.
Yep. They brought the line 9% closer. This is real results.
That's great.
Okay. I want to show you one more example of how they are able to combine not only that data across the 38,000 smart meters, but also starting to look at how they look across the cities they support and the transformers that they run. Transformer failure is bad for multiple reasons. One of the things they did was take their 38,000 meters of data. They added in information about average load on the transformers. They added age and model of the transformers, and were able to combine that, including predicted usage of the transformers, and look and see what the failure rates will be. When they went back and looked at their historical data and ran these new algorithms on the Xeon E7 v3, what they found was that they were able to predict 96% of their transformer failures before they happened.
Again, going back to operational efficiency, but in this case, even more importantly, is user experience. Nobody's happier about those transformers going down.
They're able to predict 96% based on the aggregation of all those different data sources.
Yep.
Amazing.
Yep. They can put this into production now and change their own maintenance costs, their own workforce loading, as well as again, drive up that user satisfaction for having always-on power delivery. Now, that's a mission-critical type of usage.
Expectations are high.
Expectations are very high. This is what I wanted to show you today about how we have a utility customer using the new system.
That's great. Thank you so much, Lisa.
Thank you, [inaudible]
Wonderful real-world use case that we can all relate to the value proposition delivered. After 16 years of our investment in mission-critical performance in our Xeon E7 product line, we have achieved a significant industry leadership position versus RISC alternatives. The remaining RISC footprint is small and declining, as less than 2% of all server ships last year were RISC-based solutions. The reason is the incredible value proposition that we deliver through the Intel architecture and standards-based computing. When IT is going to make a decision on deploying an infrastructure solution, the decision is based on how well does that infrastructure run their given application, and what is the total cost of operation they're going to incur for that given solution.
When you compare the Xeon E7 v3 against IBM's POWER8 at an eight-socket solution, we deliver 10x better performance per dollar, and we deliver a staggering 85% lower total cost of operation. As we've seen over time, this is the true value of an open standards-based architecture. You can see that value then across a full range of real-world applications with best, as Lisa just showed, a 72% performance improvement with the new Xeon E7 v3 so customers can run more complex analysis across a much broader data set. In healthcare, Vital Images, a solution for 2D and 3D medical image viewing, allows the concurrency of the solution to go from 12 users to 20 with a 66% performance improvement. Telco is an area of rapid adoption of Intel architecture, starting with the big back-end enterprise applications.
HiTech is actually the number one software vendor in China for BSS solutions, billing, order, and support systems. With the Xeon E7 v3, they can support more customer transactions per minute. They can deliver quicker customer service and deliver that with tailored offerings. Then SAP HANA, as I mentioned, customers of SAP HANA will see a 6x increase in performance of their solution with the new Xeon, thanks to the TSX instructions that HANA has optimized against, as well as the general performance that we deliver with the new Xeon E7 v3. When I say that big data and real-time analytics can benefit all businesses and all industries, I do mean it, including a product that may sound as simple as paint.
Those that adopt real data analytics solutions have a competitive advantage, and Nippon, which is the largest paint manufacturer in Asia Pacific, has demonstrated that competitive advantage through analytics. Here to tell us about their adoption of real-time data analytics solutions and the impact that it's had on Nippon is their Chief Business Strategy Officer, [Tony Xu]. Tony, come on out. Thank you for joining us.
Hi.
Welcome.
Thank you.
Yeah, thank you. You have to start by just telling us about Nippon and the fundamental scale of the corporation that you folks run.
Certainly. Nippon is the largest paint manufacturer in Asia. In China alone, for example, we have 28 factories, 1,700 distributors, 3,500 retail outlets, plus 8,000 suppliers and 250 million customers to serve. We are also a company who invest heavily in IT infrastructure because we believe that is the backbone of our business operation. On top of that, we also believe very much in eco-friendly, so we use eco-friendly materials to produce our paint to make sure our product is environmental friendly as well as user friendly.
Yeah, it is a very impressive operation. 250 million customers and 8,000 suppliers. I cannot even imagine. Everything in China is big, right? You were a very early adopter of SAP HANA over three years ago. You must have had some real specific problems you were looking to solve or opportunities you were looking to seize. Maybe you could tell us about what drove you to deploy a HANA solution.
We did. The first challenge we had was we were trying to reduce our report generation time, because back in the old days, it would take about five days at the end of the month for financial reports, sales reports to be generated. We just sit there and watch computer turn and turn and turn. With SAP HANA and Intel's Xeon E7 processor, we are able to reduce that processing time from five days down to just four hours. That means our management team can make their business decisions on the first day of the month. That means a 16% improvement of efficiency and productivity for our entire company.
Yeah, real business results, top-line, and bottom-line value. That's a great example, and I think you had another compelling example for deploying a solution you have to tell us about.
The other one is about using big data to improve customer experience. In China, e-commerce is growing very quickly, and consumers in China nowadays, I think they are somewhat spoiled, because when they place their order online, they will expect to receive their delivery in three days. China being such a geographically vast country, so our challenge is how can we deliver the right product at the right time, at the right place. On top of that, in Taobao, this single day sale event, where last year, Taobao sold RMB 57.1 billion worth of product in just 24 hours period.
$9.3 billion US in one day.
Yes, that's a huge amount.
Huge amount. Yes. You had to process all those.
Yes, exactly. For us, that's a 1,000-fold increase in the sales orders that come in, which we have to fulfill quickly. How do we meet that challenge? Again, by using SAP HANA's technology based on Intel's E7 processor plus the HP DL580 server, we're able to crunch four big database that we have. The first one is our sales database, which we collect from our 3,500 retail outlets, plus our historical sales data. The second is our e-commerce sales data. The third is our supplier data, because we have to make sure that our 8,000 suppliers have the right material to supply to us so we can make the product. The fourth one would be the data which we purchased from Taobao.
Because we realized that in the decorated process, consumers are more likely to buy plumbing or electrical products in the beginning, and paint tends to be the last product that they purchase. For example, if we identify someone who's buying a faucet, that person is likely to buy some paint sometime down the road.
Wow. Isn't that clever? You're buying data from Taobao about people that are buying electrical equipment or something as simple as faucets.
Right.
You then assume they must be on the move for purchasing paint sometime soon.
Yes. Exactly.
Really great predictive analytics in the paint manufacturing industry.
Yes. We're able to generate what we call a predictive demand map.
Wow
That tell us where and where what consumer may buy sometime down the road. We can pre-produce our product, pre-ship to our 1,700 distributors' warehouse. The end result is that we're able to deliver 30% of our order on the first day, 60% of order on the second day, and 98% of the order on the third day.
Wow. Who wouldn't want their e-commerce purchase to be delivered in a single day? I can't believe 30% of your purchases. Wow. That's impressive.
For that kind of achievement, we were rewarded by Taobao as the best supplier of the year.
Yes. Congratulations.
We're very proud of that.
Very impressive. Yeah. Very impressive. You even are using predictive analytics to predict what color schemes are most popular in which region, you forward produce the right color of paint even.
Yeah.
I mean, the complexity is very, very impressive. You guys have done an amazing job in applying predictive analytics and certainly big data analytics to manufacturing and to customer service. That's very impressive. Thank you so much.
Great. Thank you.
What a wonderful story.
Happy to share. Thank you.
As I mentioned, it's the combination of the complementary architectures of scale out and scale up that's needed, and we see IT organizations implementing both architectures and increasingly connecting those architectures to provide both the scale and the speed needed as the datasets grow, as we saw with Nippon Paint. You have Hadoop running on the Xeon E5 to deliver the data scale, so creating a data hub where high volumes of varying types of data can be managed. The enterprise analytics software like SAP HANA or SAS or Oracle running on Xeon E7 v3, so the scale-up solution then pulling the hot or working dataset from Hadoop into main memory for real-time analytics. It is the power of the integrated solution that compelled us to partner with Cloudera, who's the industry leader in Hadoop, a partnership that is now one year old.
The objectives of that partnership were threefold. Number one, we wanted to work with Cloudera to provide enterprise-ready, enterprise-class solutions that are taking advantage of all of the unique features that we embed in our processors and products. For instance, we have already demonstrated a 2.5x increase in performance of data encryption by taking Hadoop and optimizing it to use our encryption instructions, AES-NI. The second objective is to accelerate the innovation in the open source community around big data analytics, so doing things like providing tools and libraries, as well as directly contributing to open source with the Intel engineers and the Cloudera engineers.
The third is to drive development of a big, healthy ecosystem around big data analytics, so making it much easier for enterprise IT to deploy analytic solutions without having to employ an army of data scientists, as I'm sure Nippon Paint has done. I'm really happy to have with us here today the CEO of Cloudera, Tom Reilly, and he'll talk about some of the momentum that we've seen in the past year since we began our engagement. Tom, come on out. Thank you, sir. There you are.
Thank you, Diane.
Yeah. Great to see you. It's been a year. The year has flown by.
Yes. It has flown fast.
It has been a year of working together. You have to tell us what all we've accomplished over that year.
All right. When we started this partnership a year ago, I was really, really excited. A year in, I'm even more enthralled about what this partnership's going to deliver. It's amazing when you bring together hardware engineers and software developers to design systems that are optimized to improve performance. We did a lot of work in security, performance, and systems management, which lowers the total cost of ownership. This past year, we worked on four releases of Cloudera Enterprise, taking the best of Intel's distribution for Hadoop and combining it with Cloudera's. We now have the number one adopted distribution of Hadoop on the planet with those capabilities.
Four releases in one year. That's pretty darn good.
Yeah.
The performance delivered in those releases is impressive.
The performance is very impressive. One of the things I'm most excited about, and you mentioned, is around encryption.
Yeah.
Intel, within the first six months, Intel taught our engineers how to take advantage of encryption in the silicon.
Now when data lands into Hadoop, we actually encrypt it in the hardware, which has tremendous performance advantages. In fact, the performance impact is really negligible. Now our customers don't encrypt sensitive data, they encrypt all data. We are now advocating that our enterprise data hub is the safest place for data in an enterprise. A year ago, we could not make that claim.
Yeah.
In fact, Mastercard has not only certified our enterprise data hub as PCI compliant, they're now taking it to market to their retailers and financial partners as the place where they want credit card data retained.
That is impressive. Most secure data landing in Cloudera.
Very powerful. Finally, we've done some great work with your help, with your customers, the OEM manufacturers, building pre-engineered systems optimized for specific analytic workloads. Now we have appliances in market that are designed and optimized for doing Spark workloads in memory analytics.
Yeah. Exactly goes to our objective of building out an ecosystem to make it fundamentally easier for enterprise IT to deploy their solutions. Very nice work.
With your help on the product side of things, it's really accelerated our business results.
Good.
I'd love just to share a few of the things that have happened to our company. We're just six years old this past year, going into our seventh year. We added 85% more customers to our customer base since we joined the Intel partnership. We saw our customer adoption in large enterprises grow very fast, and now we have more than 550 large enterprises with big data projects.
Okay.
Tremendous growth there.
Good.
That's helped our business. In our sixth year of operations, we achieved $100 million in revenue. Now Intel will say $100 million is not very big.
$100 million is huge.
Well, it turns out in enterprise software and subscription enterprise software, and open source, we're one of the fastest growing companies in the history of enterprise software. Our software revenue grew 100% year-over-year, we're expecting that to actually accelerate when it years out as these projects go into production.
Great. Yeah, the results are great, you guys have grown both organically, you've also made some exciting acquisitions to help build that capacity faster.
Yes. We did three acquisitions this past year. We acquired a company called Gazzang. When we saw the work with, we were doing Intel around encryption, it gave us the opportunity to go out and buy a key management company called Gazzang. Now we natively offer key management in our platform. We acquired a company called DataPad, which has tools for developers to make the platform more approachable, and recently acquired a company called Xplain.io. This is helping us understand SQL workloads and whether they work best in your traditional databases or they'll work best in this new scale-out environment.
Very nice. Yeah, very nice. Obviously, tremendous business growth, tremendous growth, great partnership. What's most exciting for both of us is when you actually see the deployment by a customer, and you see them benefiting from the solution. I know you have lots of great proof points and compelling customer testimonials, and you have someone that you are going to bring out.
Yeah. We're very fortunate to have one of our customers here to share their use case with you. Hopefully, one of the things you learn today is we're really focused on the use cases around big data. This next customer is probably one of our customers we're most proud of, and the use cases that are having a tremendous impact every single day. Our customer who's here with us today is Cerner in the medical industry. Our presenter is actually a Fellow with Cerner. I just learned Cerner has 20,000 employees and only five Fellows. The VP of Engineering and Fellow for Cerner, Mr. David Edwards, is here to share the Cerner use case.
David. There we go.
Thank you, David.
Thank you.
Nice to see you.
Nice to be here.
Thanks for coming. Yeah. If you have to tell us a little bit about what you've been doing inside of Cerner.
Sure. Well, it turns out that Cerner is in the midst of an evolution, in which we're expanding our boundaries beyond the historical focus on electronic medical records. We're on a mission to bring together and analyze all the world's healthcare data with the goal of making systemic improvements, not only to the delivery of healthcare, but also to the health of communities. In order to accomplish that goal, we needed to create an enterprise data hub that could provide the scale and the data management capabilities necessary to handle what is already several petabytes of data. Our cloud environment performs a significant amount of processing to both normalize the structure and to standardize the content of the raw data sets we're ingesting, then bulk transfers that data into a variety of data warehouses for analysis by not only our data scientists, but also our clients.
A unique feature of our data hub is its ability to aggregate data from an unlimited number of data sources. With the petabytes of data we've already ingested, we're able to construct a much more complete picture of a person's health, as well as the health of populations. One very relevant example here is our ability to more accurately predict the presence of a potentially fatal bloodstream infection called sepsis, then alert the care providers to take immediate action. In our opinion, the combination of Intel and Cloudera, both being leaders in their respective areas of hardware and software, has been a very positive force in strengthening the Hadoop platform. The ongoing advancements in performance and scalability and management and security have been vitally important in helping Cerner realize our mission of improving the health of communities.
Thank you.
It's an amazing story, David. Maybe you can just bring it home a little closer to how these new use cases are impacting your clients and their patients.
Absolutely. Well, it turns out that the capabilities of our enterprise data hub have allowed us to develop and deploy predictive models that are helping care providers make much more informed and timely decisions. In fact, our clients are telling us that the sepsis management system has already saved hundreds of lives. On a daily basis, we actively monitor more than 1 million lives across our entire U.S. client base. That's a very powerful and compelling story, and one that we strongly feel this particular case only represents the beginning of what can actually be accomplished.
It is extremely compelling. You think about the work of Cerner taking technology and applying it to such a time-critical and incredibly serious health issue such as sepsis. Amazing work by Cerner. Very impressive.
We're just honored as software developers to be working with a partner that is helping deliver solutions that are saving lives. It's our pleasure to be here as well.
Thank you.
Thank you.
Pleasure. Thanks for coming. Thanks, Tom. Yeah, we'll see you. Open seems to be the newest of new buzzwords, but delivering a true open architecture means that the architecture is a standard, and that there are many technology innovators that can build differentiated platforms against that standard. Intel has forever been the standard building block provider to the industry, and that's what we continue to do. We have a broad range of customers and partners that have announced their Xeon E7 v3 solutions, announcing starting today. That's sad. There we go. There are a total of 17 OEMs and TEMs delivering 45 unique systems, as well as you can see, a wide range of software vendors covering an impressive array of analytic solutions. It is a very exciting time to be a part of the information and communication technology industry.
The transformation that businesses, both old and new, are seeing through the use of big data and real-time analytics is a reality. Everything from farming to paint manufacturing, from earthquake prediction to saving lives, to precision medicine, to energy conservation, the full gamut. With our new Xeon E7 v3 processor and the system and software innovation that comes from our partners and customers, we are advancing the state of analytics, and we're doing it through increased data security, through increased speed to insight, and by supporting the massive scale required. Thank you. With that, oh, look, you have your data is the new bacon shirt. I am so proud of you.
Free shirt.
A free shirt. Yes. PR always
Ladies and gentlemen, please welcome the CEO of Cloudera, Tom Reilly.
Good morning again. Thank you for providing me the opportunity to come back on stage. I thought I'd use this moment to take another look at big data, but look at it from an application perspective. I do first want to touch, though, about this partnership we have with Intel. I am fascinated what can happen when you take hardware engineers working in unison with software developers. Roughly 70 of Intel's engineers are collaborating continually with 40 of ours. I talked earlier about the first very impactful application around encryption, where our software is able to take advantage of the hardware to do encryption and change how people think about securing data. It used to be that encryption was a huge performance tax, both human performance, even tax, and system performance. It was very costly. Therefore, we would isolate sensitive data into specific tables.
We'd have to decide what data is sensitive, isolate it, and then encrypt it. With the work we've done with Intel, we no longer just encrypt sensitive data. We advocate encrypting all data as if it were sensitive. That's one example of software engineers working with hardware engineers. Today, our data scientists who are working on the applications are actually collaborating with Intel's engineers, looking at the algorithms that are driving these applications and finding common lower-level algorithms that they're operating and writing instructions in the hardware to accelerate the performance yet again. We have a very exciting roadmap going forward. This work is very important because as we know, data is exploding. The data volumes are, as the world gets interconnected, growing at astronomical rates, and the numbers are tremendous.
These use cases are helping us get very local within our cities, understanding where people are, what their buying habits are locally, and it's also connecting the global world. As we go into Mother's Day, my favorite retailer starting two years ago is marketing to me. That retailer is in Australia. I purchased a bag for my wife two years ago on her birthday, and now I turn to that retailer all the time. They're building a relationship with me from Australia. Every single industry is going through a revolution. What I'd like to share with you, though, is what are the common use cases we're seeing going into production today? How is industry capitalizing on big data? There's 3 categories of use cases that we're seeing emerge and where the emphasis is. The first is around customer 360. Do you know your customer?
Not what they have bought from you, but do you know how they feel socially or their sentiment? Do you know more about their family? Do you know their demographics? Do you know where they live? Do you know where they work? Can you anticipate their needs and create specific products or specific campaigns to address their needs? The next category of use cases are around new data-driven products or services. We heard from Cerner earlier that they've augmented their electronic medical record software with the ability now to remotely monitor patients and anticipate toxic reactions like sepsis. That's a new data-driven service that they're offering to their clients, a new revenue stream. The third area is around risk. Protecting data, protecting clients, managing your risk.
I'm going to step through a few of the verticals that we see traction and then dive down into some specific use cases so you can get a concrete understanding of how this is really changing industry. The customer 360, very common in the telecommunications industry for them to get their arms around the customers. Those of us here in the U.S., we know that we get one-year or two-year contracts, and every time our contracts are coming up for renewal or we're getting a new phone, we're trying to think of, do we switch carriers? Customer churn is one of the biggest challenges in the telecommunications industry, and they're using big data to reduce churn. We can imagine looking at call data records, looking at dropped calls, looking at spotty cell coverage, that they can anticipate unhappy customers and reach out to you.
We're very familiar. Having our long contracts, I think of churn on an annual or every other year basis. We're working with telecom companies overseas on an even greater challenge with tremendous results. Telkomsel in Indonesia, they don't have annual contracts. Their customers pay with prepaid cards. They might buy three weeks worth of calling time on a card, and then they churn. They have use cases in Indonesia where it's, let's monitor a card, and when it's almost being depleted, let's send a text message to that individual and direct them to the nearest kiosk where they can buy a new card. Now it turns out that kiosks in Indonesia, across 1,100 islands, kiosks are actually young boys and girls on bikes who can now ride near to you and replenish your card.
That's a great big data application about how to retain a customer that you don't even know. In data-driven products, we're seeing lots of great success in the manufacturing industry. For instance, in heavy machinery, tractors, and big heavy machinery equipment is getting fully instrumented, so that now if you manufacture heavy equipment, you don't say, "Hey, I've got the best equipment for the job." You say, "I offer a service where I will keep that equipment in production with minimal downtime by remotely monitoring it and providing predictive and proactive maintenance controls." Basically outsourcing the maintenance of those devices just by monitoring them. Cloudera ourselves, we're a manufacturer of software. Our software is deployed in hundreds and hundreds of data centers around the world with thousands of clusters and tens of thousands of nodes and servers.
Our customers opt in and allow us to monitor the health of those clusters across the globe. Now with this amount of data, we're actually able to predict and proactively reach out to customers who may have a wrong configuration or may have downtime. That is saving us 20% customer support deflection of cases. Very powerful. In risk, the Financial Industry Regulatory Authority, called FINRA, who manages all the exchanges, is actually collecting data every day from 4,000 security firms. All the exchanges would amount to 50 billion events every day. They're using our software for trade surveillance, monitoring for bad trades, insider trading. They used to be able to have to sample data sets and guess where they think insider trading may be occurring.
Now they're able to look across 50 billion events every day and use technology to look for trade surveillance. Caesars. The first thing we think of when we think of Caesars is the table on the right. It's where you go to gamble. Now it turns out, in Vegas, gambling no longer is the largest source of revenue. These resorts have turned into retail malls, high-end restaurants, entertainment venues, and a lot of resort services, whether it's spas, going golfing, playing tennis. In fact, the demographic of the customer of Caesars has changed from the heavy gambler that they want to understand to families, that they need to really understand who they are. Now their old systems couldn't handle this new type of data, non-gambling data.
They used to run what were called shelf campaigns, where they identified their demographics of customers who were gamblers, and of course, went for the whales and had to understand how they'd recruit more gamblers from around the world. They've now completely changed. They have a new approach of understanding how they can make money across all of their properties. They have completely changed what it means to understand the customer in the traditional casino industry. This is an interesting one with eharmony. eharmony is obviously a dating site. How eharmony differentiates themselves from all the other dating sites is they really pride themselves on making great matches. These are not simple matches. What they're really trying to understand is each of their clients very intimately and look for attributes about those clients that'll make them have a very good match.
If you look at eharmony's advertising or marketing campaigns, they talk about the success of their matching. eharmony is using big data and Hadoop in a cloud application, and they're leveraging Spark, the new in-memory analytics engine. Why is this important? Every morning, they have to create 10 million matches. They have 10 million matches of people waking up in the morning hoping to be introduced to their future partner. The success rate of that is very important to them. They're adding more and more data and running different algorithms all the time. One thing doesn't change. They have to deliver the matches every morning, and that morning changes with time. That's when they find their clients are most likely look to see if they've had a match.
By leveraging Spark, by leveraging Cloudera's Hadoop offering and running on Intel platform in the cloud, they're able to increasingly make more matches, success rate of those matches, and do it with continually new algorithms trying to improve their match rate. We heard about Cerner from David Edwards this morning. Cerner's monitoring of patients in the hospital just really drives home how big data can change our lives. Rather than kind of repeating the Cerner use case, because David did an amazing job, I'll talk about two other use cases in healthcare that are pretty interesting in how big data is changing how we look at healthcare. Intel partnered with The Michael J. Fox Foundation last year. The Michael J. Fox Foundation is focused on improving Parkinson's disease, both anticipating and the treatment of Parkinson's patients.
It used to be that a patient would go in to a doctor monthly or quarterly, and there was a 15-minute survey done, where the doctor asked, "How are you sleeping? How are your tremors? How is this medication working?" Maybe testimony or what have you. That was the data point where we built patterns of treating Parkinson's. In fact, over the last 100 years, very little has changed about our knowledge of Parkinson's. What Intel did with The Michael J. Fox Foundation was actually to instrument Parkinson's patients, so we can now track how they're sleeping, we can track if they have tremors, we can track blood pressure, heart rate, and in fact, we're collecting 300 events per second per patient. Now you can start modeling the progression of Parkinson's disease. Fascinating.
We worked with Children's Healthcare of Atlanta around their NICU center, the intensive care unit for premature born babies. It used to be these babies be in the glass room, isolated and fully monitored, and nurses down the hall tracking what's happening with these babies. That data was just falling on the floor. If nothing happened, they just lost the data. What they did working with us was actually not only retain a history of that data, but they started taking environmental data. They looked at when the lights were on and the intensity of lights in the room, how often the door opened, how many people entered the room, what the temperature of the room was, and they're able to identify what was causing stress in these little babies from just the room environment.
They completely changed the policies and procedures of who's allowed to enter the room, when they're allowed to enter, what the temperature was, to improve the health rate of these poor babies. Tremendous use cases happening in healthcare. In the supply chain space, just being more efficient are tremendous opportunities. Intel and Mitsubishi Electric partnered on a proof of concept with tremendous results. Intel has, in Malaysia, a big manufacturing facility. What they did is they took the Intel Atom processor and instruments throughout the manufacturing facility and worked with Mitsubishi Electric that has an IoT gateway. What that gateway does is it collects the data at the edge, transmits that data securely back to a data center where Cloudera's analytics are running on top of Intel's platform. In a POC, the results were amazing.
The first thing they were able to do is reduce the misclassification of bad parts. Not only say, "These parts are bad, these parts are good," they actually were able to classify many more parts as good, which improved the productivity and the yield of the plant. They were also able to do predictive maintenance of the whole supply chain operation and proactively keep machinery up and running and place inventory throughout the supply chain faster. During this proof of concept, a proof of concept is generally say, "Could this potentially work? We have a good idea." During the proof of concept, they delivered $9 million of business value to this operation. This is the next generation of factory automation, having better insight into what's happening on the shop floor and improving yield. I talked earlier about Mastercard.
Mastercard is maniacally focused on protecting all of our personally identifiable information, and specifically around our credit cards. Mastercard is one of the leading companies behind the PCI standard, the Payment Card Industry data security standard. They basically enforced the standard on all of their merchants that retain credit card data to protect it. Mastercard not only enforced it on retailers, but they enforced it on themselves at the highest standard. In working with Intel and Cloudera for about a year, took us most of last year through this partnership, we actually not only addressed their security requirements for PCI security, becoming the first Hadoop offering on the planet that's PCI certified, but we delivered to such a security level that Mastercard Advisors, their consulting arm, has formed a partnership with us to take this offering as what they call a secure data vault to market.
Mastercard is advocating to their customers, retailers, and financial institutions to place credit card data, to analyze credit card data into an enterprise data hub built on Hadoop. I never thought a year ago that I would be able to stand in front of customers and say, "Because of the work we've done with Intel, an enterprise data hub built on Hadoop is the safest place for data in your environment." A year ago, security was one of the biggest concerns around this open source platform. Today, we turned it into one of its greatest advantages. What's ahead? There's a lot more work to be done around security, and with security comes privacy. As the world is streaming data in, you don't want that data stream intercepted and then false data replaced with it. You have to authenticate all these devices.
You have to secure the data on transmission as it's coming back into data centers. Not only do you keep it secure and encrypted at that central location, but you have to have very tight controls around who gets to see that data. Under what circumstances are you allowed to access that data? It's not just individuals, but applications. We're doing a lot of work with Intel around securing data and putting tighter controls in place. We have open source project called Project Rhino and Project Sentry, which are all about access controls, protecting privacy, protecting the data. One of the greatest inhibitors to growth, I think, in the future is going to be concerns around privacy. We as vendors have to step up and put in place the controls to deliver privacy.
A lot of times, those controls are going to come back to us as individuals. Lowering the total cost of ownership is very important. The last thing our clients want are to build server farms with tens of thousands of nodes or boxes. We're a scale-out platform. The advantage of our platform is you can incrementally scale it out. The work we're doing with Intel is driving the performance of each box significantly up, therefore requiring fewer boxes for our customers and lowering their total cost of ownership. What we're doing around these new future algorithms and new workloads in designing analytic chips, analytic design chips, will drastically reduce the number of boxes our customers need, lowering their total cost of ownership. We're collaborating on systems management tools.
How do you monitor this farm of servers with lots of different software running on them and keep these boxes up and running and well-tuned and well-optimized? All this work lowers our customers' total cost of ownership. Finally, we're working on time to value. That early stat that Diane shared, I think it was like 23% or 27% of projects are successful in the big data space. What we're doing is actually addressing that and accelerating time to value for our customers. A lot of that has to do with pre-engineered systems, right? Can we design the software with the hardware, prepackaged on appliances, so our customers are not going through that, what configuration do we need? We know and we can test what are the best configurations and make that available on pre-engineered boxes or for services in the cloud.
The next work we're doing is with the broader ecosystem. In the past year, our partnership ecosystem has accelerated in large part to the partnership with Intel. A year ago, we had 800 companies in the Cloudera Partner Program. Today, we have more than 1,450. We work with what we call our partner engineering team that works with this whole ecosystem of partners to focus on time to value. We test. If you start at the bottom with the infrastructure players and talk what we're doing with the OEM manufacturers and the cloud providers to optimize our platform for their infrastructure. We're integrating with the traditional data systems and making sure that you can move data back and forth easily in pre-testing all those integrations, all the way up through the application space.
If you're a SaaS customer or a Tableau customer or an SAP HANA customer, you want your existing tools to operate against this new back end of Hadoop. We're pre-testing all of that. Finally, the large system integrators are really stepping up their practices in the space in building industry-specific use cases to transform industry. If you think of the large systems integrators, they're always looking for game-changing applications to consult to their customers. What they're realizing is that data-driven applications around big data is how to transform industry. They're building the practices, and we, along with Intel, are working with them around the specific use cases and making sure we have the platform in place to address them.
I want to share with you some use cases that moves us from the hype of big data to an understanding of how industry is leveraging big data to transform business. From saving lives, to protecting identities, to improving operations, to selling more to customers. We're seeing big data go into production. We, as an industry, are focused on these use cases to move from the hype and promise of big data to the reality of what's happening. We believe this is one of the biggest growing markets, and every single industry and every single vertical moves to take advantage of big data. That concludes my prepared remarks. I have a few minutes to address any questions if there are any in the audience. Yes, sir. Right back here. I think we have a microphone coming.
You obviously partner extensively with Intel, whose graphics processors or FPGA in some of these implementations, and do you think they have a play in some of these data centers?
I am not prepared. I don't know the answer to that question, I apologize, but I wish we had one of our architects here to probably answer how we're using the FPGA. I do not know the answer to that. Yes. Oh, there's Eli.
Thank you. Cloudera's Chief Technologist. Yeah, we've been working with Intel on a number of the processor technologies. There's a lot that we've been offloading to the core CPU, they've also introduced a number of new co-processors that are in the pipeline that we'll be able to enable. You mentioned FPGA, the Xeon Phi is very interesting, we've got some investigation going on there that I can't really share more than that on. It's something we're actively looking at.
Any other questions?
Tom, I have one for you. You've talked a lot about different use cases today, healthcare, telecom. What do you see as the breakout vertical for 2015? Where do you think the explosion is coming next? Obviously, there have been a lot of really great stories coming out of healthcare, what do you see next?
The most active vertical is financial services. In financial services, the most common use case that we're seeing is around compliance and regulatory pressure. It is amazing the amount of money that's being spent by financial institutions trying to be compliant or to address the concerns of the regulators, and the regulators are continually changing. It's not uncommon that a large financial institution will spend anywhere from $800 million to $2 billion a year on IT and human costs trying to be compliant, and yet they're getting fined. The most recent fine was with HSBC, where HSBC has 28,000 employees working on anti-money laundering, yet they were just fined for money laundering. It turns out that anti-money laundering is a big data problem. The folks that are conducting money laundering are very, very smart.
You have to use big data tools to look at all the different accounts that are being set up in different countries by the same individuals attempting to use your bank for money laundering. An organization like HSBC that acquired across the globe has all these different systems. Large financial institutions are turning towards this platform to address things like anti-money laundering, fraud, KYC, managing risk. Risk aggregation is a tremendous challenge for large financial institutions. How much risk are you taking in the market, and do you have enough capital on reserve? The too big to fail concern. What's happening is the regulators are putting more and more capital into reserves, which therefore the financial institutions can't put to work.
This big data platform is being used for risk aggregation and measuring risk, and then having a tool to say to the regulators, "We have control over our risk." I think to answer the question, financial services we're seeing a lot in a set of use cases that are directly relevant to their platform. If you're a financial services institution, you have to retain all data about all transactions in full fidelity original format for seven years in the U.S., 20 years in China. That volume of data, imagine the volume of data that they have to retain, and then they have to quickly analyze it, they have to quickly retrieve it. This platform is being used in that. It's probably the most common use case we're seeing within that industry. Thank you. Oh, another question.
Competitive advantage versus Pivotal or Hortonworks.
Certainly. What's our competitive advantage? From a technical standpoint today, security is a clear advantage. We've invested very heavily into the security of the platform, the security of data, that is our strongest technical advantage. Our true advantage are the number of customers that Cloudera has in production. We are very focused on high-value production use cases, working very, very closely with our customers, leveraging our relationship with Intel to deliver time to value faster. What's very clear is customers do not want to fail in their big data projects. We deliver the expertise. We're building use cases by industry with blueprints on how to achieve these through best practices. Our strongest differentiator is the success of these deployments. All right. I would like to thank Intel for allowing Cloudera to join this event around the E7 processor.
We are big believers in scale-out technology through our distributed compute platform. We're also leveraging the scale-up work that Intel is doing with E7, delivering, I think, very, very impactful use cases. Thank you for the time.