Ladies and gentlemen, please welcome the Senior Vice President and General Manager of Intel's Data Center Group, Diane Bryant.
Good morning. Thanks much to you in the room and those of you out on the webcast for joining us for our launch of the new Xeon processor, E7 version two. This is a dramatic refresh of our product line, the product line that delivers the highest performance, the highest reliability, and the greatest scalability in support of those true mission-critical workloads. With the rise of the digital service economy-- line results. Intel did a survey of CIOs last year, and of the CIOs surveyed, 82% agreed that big data solutions would improve their business results. Yet just 5% of them said they had actually deployed big data advanced analytic solutions that were actually driving business decisions. That gap creates a wonderful opportunity.
In fact, it's predicted to be a 6x growth in investments in infrastructure for real-time in-memory analytics between this year and 2018. Growing from $2 billion this year to nearly $13 billion by 2018. Significant opportunity for growth in enterprise IT. Of course, corporations have always had large quantities of data to inform their businesses, whether it's customer data or financial data or employee data, but much has truly changed. First thing that's changed is the quantity of data. Vast amounts of data being generated from both social media solutions, the Twitters and Facebooks and YouTube videos, as well as the Internet of Things. Billions of devices that are becoming more intelligent and are connected and driving data.
The second thing that's changed is the industry is collectively investing in tools that create an easier-to-use, self-serve means for the line of business to actually access and consume the data. Historically, or even today, the line of business are using static, simple reporting of the data. The move now with all the investments in the tools is to give the line of business access to the data to actually do the analytics and the visualization of the data self-serve. The investment in tools doubled over the past two years and is on a path to $17 billion industry-wide by 2015. The third thing is just this ever-increasing delivery of performance and ever-improving price performance of both compute and storage.
Thanks to Intel, thanks to Moore's Law, thanks to the continued growth of standard high-volume servers, the cost of compute and the cost of storage continues to come down. Over the past 10 years, the cost of compute has reduced by over 40%, and the cost of storage in dollars per gigabyte has reduced over 90%. When you have this kind of scale, you're able to process, analyze, compute on massive amounts of data that just would've been cost-prohibitive prior. Today we're joined by one of our premier partners, HP, and with them is an innovative solutions provider, VelociData. VelociData leverages all the attributes of the new Xeon E7 v2 processor, as well as HP's ProLiant DL580 Gen8 server. HP and Intel have obviously been collaborators and partners for decades, and HP consistently delivers great innovation on top of the Intel architecture.
We're very proud to have Jim Ganthier, who is the Vice President of Global Marketing for their server division, and with him is Chris O'Malley, who is the CEO of VelociData. VelociData performs data transformations and data quality and data sorting for real-time analytics from multiple data sources. Welcome, gentlemen. Thanks, Jim.
Hi.
How are you?
Nice to see you again.
Nice to see you, Chris.
Nice to see you.
Thank you very much, Diane, thanks for having us here. As Diane mentioned, HP and Intel have been partners for a very long time. As a matter of fact, that partnership goes all the way back to the creation of the x86 server market. Both of our companies have led it since its inception. It's not just about what we're able to do from a partnership perspective. What we have done consistently over that time as the leaders of this industry is focus on innovation. Not innovation for industry's sake or innovation's sake, but how we're able to use that innovation to drive real compelling business results. Chris and I are going to talk a little bit about that. With that, I'm very proud to announce today the availability of the HP ProLiant DL580 server. That 580 server is basically a Gen8 product.
Like everything else Gen8 that you know and have heard about, it's got all the intelligence. It's got all the automation built in. As a result, it can deliver industry-leading capabilities such as a three-month ROI, and also the ability to nearly double the compute per watt of your data center. It's not just about that innovation. What we wanted to do with the 580 is actually focus on some of our highest-end, most demanding customers. That's our scale-up customers. This is a space that we've been the market share leaders in revenue for the last three quarters. As a result, what we did here is we focused on their most demanding workloads. We focused on data analytics, we focused on business processing, and we also focused on decision support systems.
What Chris and I are going to do is instead of talking to you about feeds and speeds and slots and watts and everything else that we would normally do in a technical presentation, we're going to take a different approach. We're going to talk about how a lot of the innovation that we've put in place that's unique from HP and from Intel, and with a great partner like VelociData, are now able to take all of that innovation and drive dramatically different, compelling, and in some cases, industry-leading business results. Chris, tell us a little about VelociData and what you all are doing there.
Yeah. I'll give you a little background on the company first. We focus in terms of our customers who are driving data to try to gain huge benefits in terms of operational efficiencies in their corporations, as well as trying to get to real-time analytics. They look to VelociData to basically offload bottleneck functions that inevitably happen with the volume increase in terms of data velocity and varieties. We have three products: data transformation, data quality, data sort, all built on a HP Intel platform, taking advantage of all the nuances in terms of what they released today. That appliance, in terms of how it works, it basically takes source data from anything you can think of, structured data, unstructured data, streaming data. It runs as a flow through our appliance.
We do our magic in terms of standardizing data formats, encrypting data where it needs to be, sorting that data, getting it basically prepared fully to the point when it gets to the target.
Gotcha
That something like SAP HANA or Hadoop or Data Warehouse can pick up the work immediately and then begin to do analytics or operations.
Okay. Can you tell us a little about some of the business results or examples of what we've been able to do together for customers?
Yeah, one point I want to make quick, the thing that makes us unique and different in terms of how we do that, because other companies do do that, is what we do 10 x faster, 100 x faster, 1,000 x faster.
1,000 x faster.
Not 1,000%, 1,000 x. To give an example of the benefits in terms of how customers are using that capability, we have a healthcare insurance company-
a very large one that's grown through acquisitions. They wanted to do, in terms of synergies, basically have one function for customer service across the entire enterprise, all the brands.
Right.
That entails basically taking claims data from all these disparate systems, doing all that stuff in the middle that I just described, and putting it in a data warehouse so that customer service people could respond to requests around claims or questions around a claim, regardless of brand. The problem with that.
Right
is that over the years, in terms of the complexity of the data, the transformations, just the volume scale, it was taking 16 hours to actually get.
Wow
that middle part of the work done. That was being done by a leading product within the industry.
We worked with the customer, offloaded those bottleneck issues onto our appliance, which is based on Intel and HP technology, and took 16 hours down to one minute.
One minute.
One minute.
Impressive.
Not only did we fix a burning issue for the customer, helped them to realize the benefits of getting costs down as it relates to customer service, but it unshackled them, if you will, from the conventional wisdom of what's possible. Now they're thinking about all kinds of new insights in terms of, can it be driven to the customer service people, can you improve the experience more, and actually cross-sell in the course of those events.
When you and I talk about fundamentally redefining both the economics and the expectations, especially in the era of data analytics and business processing, you and I talk to hundreds of customers a year, we hear consistently, "I want to have more transactions in less time, at less cost, and it's got to be rock solid reliability, because this really is the data that helps drive my business." We also hear a lot about performance, efficiency, and reliability. How about another example of where we're helping redefine the industry?
I can't think of a better example than financial services.
Okay.
We have a financial services customer that wanted to monetize their credit card data. That entailed taking a bunch of data off the mainframe, credit card data off the mainframe, putting it into Hadoop. Once there, they wanted to do analytics against the data to try to deduce what was the genesis behind your transaction decisions. Are you getting married? Are you going on vacation, remodeling your kitchen?
Right.
Then sell that insight actually to their merchants as a lead so that then, in fact, those merchants could engage with those customers at the point that they're actually making those buying decisions to hopefully influence them.
Got you.
In theory, that sounds great, you hear these examples, practically, that was taking, with conventional tools, eight hours, 10 hours, 12 hours to actually perform. It's not that interesting to know why I was buying something at noon when I'm sleeping at midnight.
It's got to be on time.
It's got to be in the moment.
Right.
We offloaded those bottleneck functions. We took basically the transformations at wire speeds to allow them to actually realize this service to their merchants so they could engage customers in real time. I think that's a wonderful example of the advantages these large companies have with all this massive amounts of data. If they can synthesize it down, create insights out of it, and engage in transient moments, that's how you win the game in the digital economy.
Yeah. As a matter of fact, you and I have actually told customers that we want to help them take hindsight and turn that into insight or turn information into basically something that's actionable. Again, when you look at the capabilities, when you look at the performance, and you look at some of the things that when we take Intel innovations and we add our industry standard innovations on top of that, then we marry that to some really great partners, we really are helping to not only redefine the economics and expectations, but we're helping our customers deliver business results and deliver basically new fundamental things in terms of how they can help change their business and actually help change an entire industry. Chris, thank you very much for the partnership.
Thank you.
Diane, we'll turn it back to you.
Thank you very much.
Thank you.
Appreciate it. Thank you. Good to see you, Jim. Thank you. Analogies to harnessing the wind to create clean energy. As you hear, businesses are harnessing the available data to transform their business. eWeek did a survey where 93% of the respondents across all industries believe that their organizations were losing revenue by not being able to manage their data effectively. Those that lead are clearly winning. In retail, for instance, online retailers are increasing sales through the integration of recommendations into every aspect of the consumer purchasing process. Amazon actually reported that 29% of their sales in 2013 was generated through their recommendation engine, so dramatic. Nordstrom invests tremendously in big data analytic solutions, integrating volumes of data from their website with their in-store data, combining it with social data to then determine what products they should be recommending to their customers.
The healthcare industry is another one that's obviously finally going digital, and in doing so, creating just an unprecedented spiral of data. The healthcare data in the U.S. has grown at a 34% CAGR over the past four years. The largest amounts of that data are coming from medical imaging data as well as the electronic health records. This is just generating a tremendous investment in data analytics solutions. Kaiser is a leader in this space. They have reduced their annual costs by $1 billion through reducing clinic visits, and they do it by using the vast amounts of data they have and targeting the specific high-risk patients, and then providing them with personalized preventative healthcare. Keeping them out of the clinics and driving down their costs by $1 billion a year. It's pretty impressive results.
In the manufacturing, top airlines are using predictive analytics to identify mechanical fatigue and predict that failure before it actually occurs, so you can schedule the preventive downtime, preventative maintenance, and avoid significant operational loss of having to take a flight off or cancel a flight. Intel's own IT organization, which I used to run, so I'm very proud of, they've now deployed a security solution, a big data analytics security solution, that analyzes 7 billion security events a day from 47 different sources, and then they aggregate this data to predict intrusion and predict malware in the environment and take corrective action far more rapidly than they were able to do prior. Driving down the operational costs and obviously more importantly, protecting Intel's IP. Many, many examples of industries that are grabbing the data that they have at hand and fundamentally changing the way they operate.
I'm excited to have with us today a true leader in the financial sector, so the London Stock Exchange. As you well know, the London Stock Exchange is the largest exchange in Europe. With them is a longtime partner of ours, IBM. Moiz Kohari is the VP of Advanced Platform Technology, and from IBM, we have Adalio Sanchez , who's the leader of the System x business. Together, they'll give you some insights into the innovations that have been delivered to the London Stock Exchange and a combination of IBM technology running on the new Xeon E7 v2. Adalio, hey, how are you?
Hi, Diane.
Thanks for coming.
Thank you. Good morning. Moiz?
Hi.
Good morning. I'm delighted to be here. Thank you, Diane, for inviting us to participate in this event today. The new Intel E7 v2 processor is the foundation of IBM's new X6 line of scalable enterprise-class systems that we launched last month. It really enables these new systems to provide new levels of performance, of agility, and really reliability for today's new enterprise-class applications like ERP, like database, like analytics, that have not been run on Intel architectures in the past. We also expect this new line of servers to spawn a new wave of virtualization of enterprise-class apps as we go forward.
The X6 line of systems is our sixth generation of our Enterprise X-Architecture systems, which is a culmination of many years of R&D development and collaboration with Intel, which quite frankly, has generated a lot of industry firsts, from the first scalable x86 systems, to the first machines that hit 1 million tpmC, to really most recently, what we've been able to do with X6, which is the whole notion of integrating flash into the memory channel with our eXFlash memory channel storage capability. Quite frankly, as a result of this, according to IDC, that is why IBM System x today has the number one position in four-socket and above systems in the x86 world. Today, we're really happy to be here as part of this and also have Moiz Kohari here from London Stock Exchange.
London Stock Exchange is a great customer of ours, and I know that being able to having to operate the London Stock Exchange and having to deal with the demands of performance and reliability really poses a pretty interesting challenge for you and your team. Could you tell us a little bit about how you leverage today System x and the Intel E7 line of processors, and what's your perspective on that?
Sure. We're very excited about the E7. Usually, when you look at new processor lines, you're looking at speed. We're getting that. You're looking at additional memory exchanges, London, Milan, Oslo. On the back end, we also run clearing and settlement services for many of the European markets. From that perspective, our needs are such this lowest latency environments on the front end, from an exchange perspective. On the back end, we have to provide analytics and all the risk management and collateral management services that we have to and fully integrate that back in with the front end so that we can put all kinds of checks in due to new regulations and things that are arriving.
When you think about all of that, what ends up happening is that it's not just we're looking for a large distributed service environment because your data centers, everyone's data centers are growing exponentially. We need denser platforms in order to really consolidate our data centers and still be able to deliver much higher level of service, especially when it comes to latency jitter and things like that. If you think about it, latency jitter gets added into the markets when you have your entire exchange or other components that are distributed in a large environment, and you have networking components. You have that TCP/IP stack, which is non-reliable to begin with. From those perspectives, when you look at denser platforms and allowing us to be able to leverage memory-based communications across some of these adjacent components. It becomes really interesting.
This whole notion of system design end-to-end, given your varied requirements, is really fundamental, isn't it?
Well, absolutely. Even when you do a distributed environment, from our perspective, it is an entire system that you need to deploy. We don't look at it as, oh, that's the database, or that's the processor. The whole thing needs to work together because our clients measure us in our ability to deliver a low latency platform. Of course, reliability, serviceability, availability, that doesn't go away.
as you look at E7 v2 and the new X6 display in your enterprise.
We're certainly, as you know, we're doing all kinds of work in the lab right now. We're very excited because as we look at denser platforms and our ability to consolidate multiple processes into single system images so that we can try to take some of that jitter away, that becomes really interesting. There's all kinds of innovations that are happening.
This whole notion of performance, of agility, and reliability is fundamental to you, huh?
Absolutely. I mean we can't allow that.
That's what X6 is all about with our Intel Xeon E7 processor, we look forward to continuing the collaboration between our two companies as we drive this thing to another level.
Thank you very much
Thank you for being here.
Absolutely.
Thank you for your support and look forward to working with you.
Yeah.
Diane, up to you.
Thank you. Thank you. Very nice. Thank you, Lance. Nice. Thanks, Adalio. It's good to see you. In-memory computing is transformational, as we've said. The move of the large data sets from disk-based data warehousing to main system memory directly connected to compute creates all new usage models. For instance, there's obviously, clearly, a technology that enables these kinds of usage models, and these kinds of results. Real-time data analytics and mission-critical computing requires some fundamental things. They require maximum performance for very high-speed processing that can provide these rapid real-time decision-making. It requires a very large memory footprint. We're talking terabytes of capacity that can support these ever-increasing data sets. It requires very high reliability and availability, as we just heard. When you're treating memory-like storage, it means the system can never go down, that data integrity is critical.
The biggest limiter today in deploying big data analytics solutions is complexity, having a broad ecosystem of analytics solutions to match what you hear as the broad usage models is very important. That brings us then to our new product line. This is the second generation of the Xeon E7 processor. The E7 v2, which was code-named Ivy Bridge, for those that you follow our code names. This delivers against all of those vectors. It supports two-socket to eight-socket with our on-die integrated QPI interconnect, so seamless, glueless up to eight socket. It obviously supports beyond eight socket, 16 socket, and beyond through the OEM innovation that's occurring around the node controller solutions.
With the launch, we're launching 20 different product SKUs, varying across cache size, core count, and frequency so that we can deliver a targeted solution optimized for the variety of workloads. With 4.3 billion transistors on a die, I'm sure you can count that and see it's evident here with this wafer. 4.3 billion transistors per die utilizing our 22 nanometer Tri-Gate 3D transistor technology. It integrates up to 15 cores per socket. Each of those cores is hyper-threaded, so that's a total of 30 threads per socket. It runs up to 3.4 GHz . On top of the 3.4 GHz , we also have the Turbo Boost technology. The result of all of that, as you can see, is a doubling of the performance from the prior generation E7.
The new E7 supports 3 x the memory capacity, so 1.5 TB per socket or 6 TB for a four-socket server, obviously, and 2.5x increase in the memory bandwidth. Truly optimized for in-memory analytics. As the core count and the memory capacity and the CPU frequency all increase, it's important that the I/O scale with that so that we maintain a balanced system. We have a 4x increase in the I/O capacity, so both total number of PCI Express lanes going from 72 to 128, as well as a move from PCI Gen 2 to PCI Gen 3. Finally, we have added to this product line over 20 memory I/O and CPU RAS features. We collectively call these features the Run Sure technology.
These are features that identify errors and either self-heal, like with our QPI interconnect, or capture and propagate the failure up to the software stack, whether it be the OS or the VMM, the virtual machine monitor, or to the application so that the application can take action to maintain system operation. The result of all that is five nines of reliability, true mission-critical class reliability, delivering the uptime that is offered with the RISC systems today. Truly bringing us into mission-critical class computing. An example of a RISC migration is the Yunnan Telco Company, which is in the southwest of China. They did a RISC migration off of RISC onto the new Xeon E7 v2.
The E7 met all of their network operator needs for mission-critical operation, delivering 24 by seven against their billing solution, and the company reported a 60% reduction in the total cost of operation from the move off of RISC onto Xeon. Dramatic improvement. With the new E7 v2, we are announcing with our partners, 20 world records. A small sample here is shown. 20 world records across the full span of workloads, enterprise workloads, technical computing, virtualization, database, Java applications, literally across all workloads, we have a leadership position. As you well know, we have long been the leader in a price-performance environment on our Xeon portfolio. Now with the launch of this product, we take a clear performance leadership against all RISC architectures. For the first time, we have performance leadership against IBM as well as the other RISC solutions.
This really creates a game changer in the economics that we can deliver to IT organizations. I said that in-memory computing was transformational, here's a data point to reinforce that. The combination of IBM and Intel technology delivers 140x performance improvement in database query time. If you start with the prior generation of Xeon E7 running on IBM Db2 10.1, you're pulling data off the disk out of the database, that is the reference point, that's 1x performance. If you move now to IBM's new Db2 10.5 with BLU Acceleration, which is their in-memory database analytic solution, and you run that on the new Xeon E7 v2, you deliver a speed up of 148x. Truly dramatic. At 148x improvement, you're able to unlock trends and data and patterns in the data that you just realistically couldn't have done before. Dramatic, dramatic change in capability.
I'm excited to have with us an industry leader in the use of analytics to both create new business models as well as to improve the efficiency of existing models. I think you'd agree that eBay is a perfect example of the digital service economy, and they have a great proof point for the benefits of SAP's HANA solution in-memory computing, running on Cisco UCS. With us is the Vice President of Corporate Enablement Functions, Dan Morales. eBay runs on Cisco UCS systems, benefiting from the incredible engineering innovation that we have all counted on from Cisco. I'm very happy to have with us as well, the leader of the UCS product line and Senior Vice President and General Manager of the Data Center Group, Dave Yen . Dave? Hey, nice to see you. Thank you. Hi. Here you go.
Well, thanks, Diane, for inviting us here. You listen to the description of London Stock Exchange. It is obvious in today's high-performance IT environment, it requires intimate integration of compute and networking. With its networking core competency, a fabric computing-based architecture design, and innovative in doing provisioning and the ease of operation management software, Cisco entered the x86 server business less than five years ago. The reception by the customer and the industry was incredibly successful. We've been gaining market share every quarter since introduction. Today, we are the number two x86 blade server worldwide market share leader and heading for number one. Diane talk about the 20 Xeon E7 v2 performance records. Cisco UCS actually possesses seven of them.
You listen to the wonderful description from the CEO of VelociData, and I couldn't help thinking, about six months ago, actually, the CEO of one of our competitors referred to the rapid growth of UCS, Cisco UCS, as that of Velociraptor. With that, Dan, I'm so glad you can join me here today. Cisco is very proud in partnering with Intel and SAP to support your initiative at eBay. You manage what is clearly one of the largest internet platforms in the world. Can you give us a sense of what's the scale of the eBay business?
Thanks, David. Yeah, the eBay business, it's an incredible story, as most of you guys are probably users of one of our products, so let me cover it briefly. You guys are familiar with the eBay marketplaces, typically where you go on, buy products through auctioning or buy it now and buying new products. We also have what we call eBay Enterprise, which is a division of eBay that helps brick-and-mortar type companies enter into the e-commerce businesses where we can help them with online e-commerce and pick, pack, and ship. Of course, the third major business that we have is the favorite way to pay in a secure way to execute a transaction via PayPal.
Those three businesses make up roughly $212 billion worth of commerce a year on behalf of eBay Inc., which is tremendous amount of data flowing through the actual franchise and through our platforms. That is roughly 143 million active customers buying and selling through these platforms in virtually every country around the world. What's exciting about that is that generates roughly 50 PB worth of information that eBay can leverage for analytics purposes to help these buyers and sellers become more effective storefronts online.
What are the opportunities for eBay to take advantage of the real-time analytics to increase the performance of eBay marketplace?
Yeah, the opportunities are endless, and we're on the forefront of predictive analytics, whether it's to help secure our platforms more because security is on the forefront of eBay and PayPal. I'm really excited about the latest innovations that we've been working on, which essentially is we're looking for signals within the data to help eBay run its businesses better. When I mean signals, it's we employ roughly 300 data scientists across the eBay franchise. They're looking through anomalies within the data to help, whether it's spikes in gold or issues with revenue because of changes in platforms that we don't control. It could be an operating system upgrade on a mobile device or an operating system change in a PC or a Mac that affects commerce.
Rather than being reactive, we're looking to try to find that as real-time as possible so we can make modifications within our platform to help keep the commerce engine running. We've ventured down a path to find signal detection in an automated way. These 300 scientists, they probably do a couple of hundred thousand in a month. Now we're looking to ultimately get to how can we do 1 million signals in a month and give those scientists better data so that they can drill in and better run our platform. Ultimately driving better revenue, more sales for our buyers and sellers on the platform.
This is very exciting. Dan, I know you are currently evaluating the next generation technology from Intel and Cisco. What early results are you seeing in your SAP HANA environment?
Yeah, we're very fortunate to have great partners in Cisco and Intel and SAP and NetApp, that enables us to stay bleeding edge on some of those use cases that I just described. Through that partnership, you guys were able to provide me with four C460 M4s in which we're running HANA version 70 in that platform. When we compared that last week to the earlier signal detection platform that we're doing, we're seeing anywhere from 50%-100% changes, gain improvements in that signal detection platform. That's exciting for eBay because we can run more transactions through that engine, ultimately driving and helping our sellers.
Well, I'm very glad to hear. You are definitely heading down an innovative path. When your team set out to architect the infrastructure for this platform, what attributes were most important to you?
Absolutely. A couple of years back, we were taking a look at the platforms that run the back office. It happened to be mostly our SAP finance platforms, HR platforms, procurement platforms, they were due for an upgrade. They were running some monolithic RISC-based platforms. We said, "Okay, times have changed." Virtualization was out there, we evaluated Cisco UCS, ultimately decided we were going to migrate to Cisco UCS. I could say that the goal was because we wanted to go from months of rolling out traditional monolithic to days/hours. I can say as of two months ago, we're 100% virtual on those platforms, and we shut down our last monolithic RISC-based platform running SAP. We're literally there.
True story, I was in a meeting last week with my architects, we were talking about spinning up an instance of SAP, because we wanted to do some testing of a new language pack for SAP. I looked at him, I said, "Okay, I know we're 100% virtual. How fast can you get that environment up for me?" They're like, "Today." The days of months are gone. We're excited to see that because of partnership with you guys at Intel and SAP.
Thank you for sharing your story with us.
This is a fantastic real world look at the new breed of analytics that Dan and Intel is talking about here. Thank you very much.
Thank you.
Thank you, Dan.
Thank you. Thank you. Good to see you. Lots of fun. That is great. Amazing results. What eBay's been able to achieve with SAP HANA and with the UCS system is obviously very impressive. I want to demonstrate an example of another transformation from another industry, the oil and gas industry. Typical for modern rigs these days are tens of thousands of sensors that are distributed across the range of equipment in an oil and gas platform, monitoring and predicting potential failures in the rigs. Failures obviously lead to downtime. Downtime is very expensive, particularly in this industry. It's calculated at $16,000 per minute or nearly $1 million per hour when these rigs go down. Using in-memory analytic solutions with the Xeon E7 v2, maintenance decisions can be made real time, having a significant economic impact.
I'd like to invite on stage Shannon Poulin, who runs our Datacenter Marketing Group, to come and give you a demonstration of that. Hey, Shannon. How are you?
How are you? Good.
Come on in. Tell us what you're going to do.
Sure. Really cool demo today. We're going to look on the left side of the screen. You guys will see a previous generation Xeon E7 system. On the right side of the screen, you'll see the new Xeon E7 v2. The difference between these are pretty obvious, but a couple of things we've already heard about are the large memory footprint that's available on the v2 product. What you heard from Diane is, as you look at oil and gas and as they're getting a bunch of the sensor data in, you can generate a very large data set. In this particular demo, we're looking at 4 TB of data. If you're on the left-hand side, you can't pull all that data and put it in memory. You've got to pull some of that off a disk.
On the right-hand side, you can fit that entire 4 TB data set in disk. You can process it a lot faster. This is actually a SAS analytics application running on SAP HANA. It's really cool data. We'll get it started.
Okay.
What you can see, it's analyzed all the data. It's brought all that 4 TB of data in. The demonstration on the right-hand side finishes in about five seconds. On the left-hand side, it takes about 12 minutes. We're not going to wait until that one's done.
We're not going to let you stay on stage for 12 minutes.
We could, but we won't. What you can see is the right-hand side, obviously, the performance of doing something in memory is just incredibly faster. What can they do with this if you're an oil and gas company? What can you do with this? Well, you can see in Poland, in Łódź, we have an issue. It's looked at all the data that it analyzed as the four terabytes, compared it some previous data and said, "Hey, let's go evaluate. We think we might have an issue here, so let's go take a look at it." You go in and you say, "Okay, what specifically is the issue with the data we looked at? We got this massive speed up, this 128 x speed up. We did everything faster. We got a result in five seconds.
Now, oh look, we're coming up with something in terms of predictive failure. What they can actually go do is they can go in and say, "Okay, I want to replace the pump," or, "I want to gracefully shut down so that I don't have an issue, and I want to get things back up and running very quickly." What that means to them from a cost perspective, you talked about it, massive speed up, but also 12 minutes to them is a pretty big deal. It could be up to $200,000.
Right.
The great new performance of the in-memory capability on the platforms delivers them a great result.
Tremendous. Yeah, very impressive results. Hey, thanks, Shannon.
Great. Thank you.
Super. That's impressive. Yes, a great application, predictive maintenance. We have all watched the RISC market decline over the past 15 years, this past year alone, power system revenue declined 33%. Today, the RISC architecture holds just 6% of the unit share of four-socket and above, yet those units actually comprise 52% of the system revenue in four-socket and above. With the Xeon E7 v2, we just further democratized mission-critical computing. The four-socket Xeon E7 version two offers up to an 80% higher performance compared to a four-socket IBM POWER7+ solution. While doing so, delivering an 80% reduction in a four-year total cost of ownership.
Dell and Intel have had a long-standing partnership. We're delighted to be part of the Intel Xeon E7 v2 launch. Our customers are facing a challenge extracting value from the enormous amounts of data that their organizations produce. To do so, they need more memory, faster cores, and more of them.
Real competitive advantages, as you heard from eBay. QlikView is a great example of a business discovery tool that delivers true self-serve BI, putting the data in the hands of the line of business. Oracle, their 12c database with in-memory application support is perfectly suited for the Xeon E7 v2. Lots and lots of ecosystem partners making these solutions come to life. In-memory computing is transforming the business through insights and faster decision-making. IT is the business. Advancements in IT have enabled the digital service economy. The Intel Xeon E7 v2 family delivers top-of-line performance, the highest reliability, highest memory capacity, and highest scalability, all of which are critical for mission-critical applications, including in-memory analytics and business processing.
The Xeon E7 v2 will give a highly anticipated boost to enterprise IT, giving them the ability to deploy new capabilities that were previously impossible, adding even more value from IT to the businesses they serve. Thank you very much. I invite up on stage Mark Miller, who runs public relations for the Data Center Group, and we'll open it up for questions here and I also believe on the webcast, right?
Yeah. Thank you, Diane. Quick instructions. For those in the room, there should be a couple of mic runners. Just raise your hand. We'll make sure that you get a mic so the folks on the web can hear. For those on the web, there's an Ask Diane button, so go ahead and ask your questions there. That'll be fed here in the room, and I'll take some questions from both. For those who have questions of our partners, of all the great examples that we heard this morning, they will be here immediately following, so please feel free for those of you in the room. One last note, if you didn't get your question answered on the web or in the room, please let myself or Radek know and we'll be happy to answer your questions after that. All right. Thank you. Anybody with a question?
It's hard to see. Anybody in the room? Could we have a mic here in the middle, please? If you could also identify your name and organization, that'd be helpful.
Sure. Diane, I'm Ambrish with BMO. Just going back to the CAGR you've shared with us for the DCG business. At the last earnings call, you kind of stuck with a long-term view.
Where does this fit in within the overall CAGR, does that change the composition of what's going to drive the growth? Thank you.
In our projected growth CAGR, we absolutely have built in already to our numbers the growth in big data analytics solutions. I firmly believe that these kinds of examples you heard from eBay and from London Stock Exchange of using massive amounts of data to benefit the business will just continue to grow. We're really in the early stages of in-memory computing, this is a growth driver for our enterprise IT business. When we talk about enterprise IT growing at about 8% CAGR over the long haul. Big data analytics is part of that growth engine. Thank you.
Excellent. Anyone else? There in the back.
Rik Myslewski with The Register. I understand that the new chips have two different memory channel types. There's a performance memory, and then there's a lockstep. How does that affect in-memory computing?
Wow. That's a very good question. Is there an architect in the room? Is anybody-- I'm looking for a lifeline here. I think we'll have to get back to you on a-
Just get back to me. That's no problem.
Yeah, we'll get to-