Drill to Detail - Drill to Detail Ep.18 'Oracle's Cloud Analytics and Data Visualization Strategy' With Special Guest Vasu Murthy

Episode Date: February 8, 2017

Mark Rittman is joined by Vasu Murthy, Oracle's Senior Director for Product Management of Oracle Business Analytics to talk about what's new with OBIEE and Oracle Data Visualization and the recently released Oracle Analytics Cloud, a dive into the technical architecture of these new additions to Oracle's BI platform, and Oracle's vision for hybrid on-prem/cloud analytics.

Hello and welcome to this special episode of Drill to Detail recorded in San Francisco during the Oracle Business Intelligence and Warehousing Special Interest Group meeting in Oracle's HQ in Redwood Shores. I'm actually joined in this episode by someone quite a few of you might know, actually, and someone I've known for quite a few years, Vasu Murthy, who's a senior director in charge of Oracle BI, product management at Oracle. And he's going to talk to us on this episode about Oracle BI and where it's going. So, Vasu, welcome to the show. And just introduce yourself to anyone who doesn't know you.

Thank you very much, Mark. It's great to be on your podcast. So my name is Vasu Murthy. I'm Senior Director of Product Management, responsible for business intelligence technology products. It's an interesting journey for me. I've been at Oracle for about six years. I came to Oracle via the acquisition of DataScaler. It's an in-memory database startup company that my friends and I had co-founded. And we were acquired by Oracle, you know, to get some of the distributed in-memory database technologies

for the Exalytics in-memory machine. And Mark, I believe that's when I first met you. Yeah. Possibly in London. Yeah. We were talking about Exitics and memory machine. Exactly. Exactly.

I remember I went to, I was invited to a, I think it was an internal, internal X-Week sort of session you were running there, which is an internal training sort of sessions you run for consultants within Oracle. And it was about the aggregate advisor, I think, at the time. Yeah. And you went on to become quite an expert in exolytics and we saw success all over the world uh you know a few thousand machines deployed around the world um that was a great experience for me i've been in startups before that and to get the scale of oracle and

quickly the speed at which we could actually develop products and get it adopted all around the world was amazing. So that was a great experience. And from then on, I moved on to become the product manager for the BI Server product. And that's the time when we were entering this new self-service world, trying to extend our BI Server technology from just working on IT model data and you know, business user self-service use cases. And then on, I'm where I am. So working and leading Oracle's cloud services, bringing the rest of the stack into the cloud. Okay.

And also working on product management for the entire platform, really, that actually includes our on-prem software and Netspace as well. Fantastic, fantastic. So let's actually start. I want to talk to you in this episode about, I suppose, where Oracle BI is going, and particularly kind of with this move into the cloud and so on. But let's just kind of take a step back a minute and just, with Oracle BI that most people know, the on-premise sort of version,

OBI, 11G, 12C, and so on, and data visualization. So where are we currently with OBI? What's the latest release? What are the kind of features of that? And kind of what's the take-up benefit, really? So if you trace the last several years, right, so when I started, you know, Exalytics was the big in-memory technology,

and analytics was the big thing. And the next wave was really self-service. So we started with getting our server infrastructure enabled so anybody could actually bring in data without IT having to model it. And that went hand-in-hand with providing visualization tools where an analyst could actually do all these, you know, self-service analysis and share it and consume. And this package we called data visualization, right? The package that includes server components that do non-modeled data analysis and the user interface tools that enable analysts to, you know, visually, you know, visually find insights and share.

So this tool set and the backend components together were called Oracle Data Visualization. And this was released to market in 2014. And around the same time is also when we really seriously thought of delivering these products on the cloud. And so 2014 is also this year when we launched BI Cloud Service. So BI Cloud Service is a multi-tenanted

cloud service where multiple customers can just purchase as many users as they want, they can subscribe to it, and they get the full breadth of Oracle BI along with it. Okay, okay, okay. So there's a lot to take in there, really. And so obviously, there's the movie to the cloud, there's Oracle DV, there's Airspace and that sort of thing as well. So let's drill into, I mean, where I think where there's been a lot of innovation,

a lot of very interesting new developments from you guys has been in Oracle Data Visualization. And you talked about there, it was a separate product line and so on there. I mean, tell us a bit of the story about how that came about, really. And I suppose, you know, who is the customer for Oracle DV? Is it different to the one from OBIE? You know, what's the kind of the origin story of that, really? And how are you involved in that? Sure.

That's quite an interesting story mark and so around 2012 around 2012 um you know around the time actually i don't know if you know or not we used to oem tableau in in in oracle and at way a hyperion and we used to call it Visual Analyzer. And before that, too, Oracle was involved in, you know, self-service user enablement, right? So from S-based to interactive reporting, you know, empowering business users to get access to data and analyze has always been, you know, in our DNA. What really changed was really the merging of that world self-service world where people can connect to a database connect to an space cube or you know connect to files and be able to analyze it's kind of that world is merging with the it curated world where you know for good reasons

organizations needed need trusted delivery of data. That is, if an organization is reporting their quarterly earnings or they're making some big hiring, firing decisions or they're deciding how to compensate their salespeople, that data needs to be correct. And that data needs to be correct, verifiable, auditable, and needs to go through a very curated and governed data pipeline. And that technology is really what, you know, now deployed in pretty much every big business. It's kind of a core necessity for business, along with databases and data management, to provide trusted analytics. The new thing was we took the technology that are use cases that we already knew addressing self-service needs and married that with our enterprise curated data delivery model. And so we have one platform now that can do both. from my HR system and know, you know, my team members and then the latest HR transactions that have happened, but also pull in, say, from, you know, compensation benchmarking data that could come in a spreadsheet or could come in some XML feed, I can blend them together on my visual tool

and be able to derive some insights from it. So that was really the, you know the key thing that we did that was released in 2014 as Oracle BI 12c on-premises, as well as BI cloud services. Okay. So I suppose one of the things that's interesting about seeing Oracle DV being deployed is maybe the different kind of, I suppose, sales model that's involved in it, the different way it's deployed. It's more of a land and expand thing as well. I mean, how much success are you seeing there in kind of, I suppose, the Oracle DV desktop being the thing that seeds the kind of interest and it then builds out into kind of wider deployments of Oracle BI? Is there much going on in that area? Yeah, good question, Mark. So basically, I think your question is who buys Oracle DV and how do we really sell it? The traditional model for self-service tools, exemplified by Tableau and Click over the years, is individuals, when they have this analytics and visualization needs, purchase these tools by themselves, using their

credit card or purchased for a few people. And over time, there's kind of a prevalence of these tools in multiple pockets in a company. And eventually, those may get consolidated into a larger enterprise purchase of these tools or adding these tools into their standard list of tools to support them. Every company pretty much has more than one tool. They probably have like half a dozen tools. So this gets added to the list of standard tools that the enterprise users can use. Now, do we see that and do we have channels to get there? Yes. So we released data visualization desktop, which can be downloaded from Oracle for evaluation purposes. It does not have any restrictions at this time.

And then we have data visualization cloud service for at least a minimum of five users. They can get it all in the cloud, full functionality at a pretty reasonable price, or they can buy in the context of BI cloud service with the entire enterprise and self-service use cases. Now, do we have to go through the same set of, you know, progression, if you will,

same progression, if you will, like Tableau and Qlik? In many cases, we are seeing the progression is much faster and sometimes, you know, really fast that we can be able to actually get enterprises to adopt this without having to go through, you know, the slow land and expand strategy. And that is because we already are talking to the same customers in the HR departments for our HCMs, you know, SaaS conversation.

We are talking to their, you know, sales, marketing service departments for our CX, you know, products that we sell in SaaS, likewise for ERP and supply chain and so on and and because we're already having multiple conversations with those businesses it's often possible for us to have you know a one-shot adoption of these tools in an enterprise. Okay okay so if somebody was evaluating say DV desktop or Oracle DV against say say, Tableau, what would you say would be the differentiator? What's the differentiator that Oracle have really? Anyway, what have you done differently that solves a different problem or kind of makes it a more appealing product in the long term, do you think? Yeah. So the way I would answer the question is, of course, Tableau.

Take an example, Tableau, right? It's probably the most popular tool out there. And they've done years and years of investment in making that really easy to use and, you know, done a great job of that. So why would someone buy Oracle Data Visualization when Tableau is an option? And there are several things, really. First, to start with, I mean, the data sources. So, of course, we do connect to

the same type of data sources that Tableau connects to, same number, but there are specifically for SaaS use cases. We not only have connectivity to more SaaS applications, including Oracle SaaS applications, we also have predetermined data extracts for these users. So if I'm in a choppers and I'm a recruiter, I know exactly the type of, we know exactly the use cases they are interested in. So we can provide specific data extracts for them. We have pre-built content on top of that. So we have pre-built projects for such users. That's one, right?

Data, the type of data that we can connect to and how easily we can appeal to business users. Number two is really comes to the kind of visual environment that we have developed. So frankly, I should say, I mean, we have looked at all these tools and we have learned from them and we actually made it much easier to, and with less number of clicks,

to get to insights than than many of these tools for example we don't have to create visualization separately they can drag it onto the same canvas as they drag it in all of them are wired and connected so clicking on any component of a chart automatically highlights the related components and all the other charts even though though they may not be in a shared dimension, right? So that we have one-click access to a lot of advanced analytic tools, including clustering, forecasting, trending, and so on. And we have pretty good extensible visualization architecture. So, you know, just our demo team has put together about 20 different visualizations on top of about 25 visualizations we ship. So, you know, just our demo team has put together about 20 different visualizations

on top of about 25 visualizations we ship. So the breadth of visualization is pretty good. And there's another whole area where, you know, I believe we are in general doing a better job. It's really appealing to business users who are not analysts. So, yes, it's very common for analysts to kind of do all these complex stories and share it with others. And typically, business people, say VP of marketing, VP of sales, director of HR,

these people just consume these analysts for this analysis. What we have tried to do is not to kind of give them more independence in accessing data, even though a tool like Tableau or even DV may be too complex for them. So search-based analytics, right? So you type in keywords or even free-form text search

with some natural language processing. That is something we've already invested in. We have the first version of that already available on DV Desktop. It's been available in our on-premises software as well. Beyond that, we already have self-service data preparation included in the same context of the tool. So we can do prepare, analyze, narrate, sort of prepare, visualize, narrate.

And that's been there since last year. And it's pretty complex, and it can actually fully function ship to data sources the entire data preparation steps, right? And the last I would say is in smart insights with machine learning. So we just released our first version of smart insights. So you just load a data set, and it'll get you a set of rich visuals sorted by information content, right? So you just load a data set. we're going to generate a whole bunch of well-defined histograms and charts ordered by, you know, the amount of variance in the data. And the amount of interestingness is the word we would like to use internally.

How interesting is this particular correlation? And we sort that and you can actually work on drivers. You can say, say hey what am i interested in what what's the variable that i'm interested in and it's always going to um you know redo all the charts based on that factor interesting um actually i can probably go a little on this is really interesting yeah so mobile is another area where there's been a lot of investment so the functionality of our desktop tool is 100% compatible, the same as the functionality we have on the cloud because it's all web-based, even our desktop tools are HTML5 embedded. So that gives full stability between our desktop and

this cloud or server infrastructure. And every one of our user interface is responsive to both tablet and a phone, apart from the desktop phone factor. So the full authoring environment that you see on DV desktops there can actually be consumed from a phone if you just enable mobile web. That's the first step. And the second step

is coming up with specific point-and-mobile applications to let you consume the data. So we just released Synopsys. So Synopsys gives you a pretty standalone analysis in a phone. So if you have a spreadsheet or a data set as an email attachment, you could just click and open that in Synopsys. And in a couple of seconds, without me having to answer any questions, it's going to generate a whole bunch of visuals. It's very easy to change and manipulate data and get the insights you want. Beyond that, Day by Day is another mobile application, but this is for a different use case. This is for a business leader, right? So in the context of where I am, what I am,

what's the information that should be most relevant to me? And we just push that information. For example, I have a weekly review with our sales team, North America sales team on, say, Tuesday morning. And we can configure it so it'll automatically learn that you're actually searching for this Tuesday morning and can actually send you this information on where the latest of where the North America sales stands in terms of pipeline, in terms of forecast, can get that information to me as a card or notification. And this can be set or we'll learn, you know, based on where I am,

what kind of questions I ask wherever I am, what time I ask these questions, what else others can share with me. So we want to bring the kind of, you know, model that's getting popular in consumer applications. You say Google now, I love that service, right? If I buy a ticket in Fandango, it's in movie ticket, it's going to tell me, you know, about an hour before the movie starts that, hey, you better start driving, the traffic is so and so and so, so you got to leave now before, you know, for me to go reach the theater in time, right? So you want to get the same model. And actually, that we're releasing in, I i think very shortly now um and we can actually showcase this for customers right now and we have it running internally so a whole bunch of areas

and the last but not least we are designing our platform to be headless right what does that mean it's not required yeah what does that mean? Headless, yeah. So visual tool, developing a visualization service, but headless. That means there's no UI for it. So what does that mean, right? So every API that we are building, every tool component that we are building, including data ingestion, data preparation, visualization, sharing and narration, and so on, all of these can be also accessed via APIs. So the trend is people want to consume analytics in the context of applications. So we want to enable application developers to be able to connect to a data set, prepare a data set, even visualize the data set just using API and pick the resultant visual and embed it into their applications. You see what I mean? Yeah, I do. There's a lot there really.

Is it the case now then, maybe it's a loaded question, but is it the case that DV is really where your development work is going now? It sounds to me like you've almost kind of rebooted oracle bi with oracle dv and this seems to be where the focus is going now really i mean is that a fair thing to say or or or not it's it's yeah it's very much the fair thing to say given that when we say data visualization internally we are not not just talking about the visual interface right this? Yes. Visual interface gives access to all the capabilities. As much work or more is actually going in our backend platform, right? For example...

So what's in the backend platform? What's in the backend platform? What's in the backend? Very good question, Mark. Thanks. So backend, when you talk about backend, it's really there are two levels, right? So one is our BI server, the core processing engine that actually drives Oracle BI.

And then beyond that lie the data management layer with database, big data, and where the actual lot of compute at scale can happen, right? So if you look at the backend platform and talk about BI server, OK, BI server, for example, we want to give smart insights. That is, we want to automatically generate insights for users when they say, hey, what drives that revenue or explain revenue? And we can find out what the biggest contributors to revenue from the various dimensions that we have in the data. And when we want to compute that, we need an infrastructure in the BI server that actually creates a machine learning model out of this, right? Creates a decision tree that can be then displayed.

So such infrastructure is being built in our BI server. Likewise, if I upload a spreadsheet, right, or connect to a SaaS source, which itself cannot compute, it's been an extract model, right? When I connect to a SaaS source, which itself cannot compute. It's been an extract model, right? When I connect to a, when I say SaaS, I'm talking about application clouds like Salesforce or Oracle Sales Cloud or ERP.

When you connect to these systems, data gets copied over. It gets imported into the BI and analytics layer. And so we have built a pretty robust in-memory computing engine in the BI server. So it can just analyze the data itself. It can actually run SQL queries on top of it. So we can cache data, we can make it easily accessible for visual tools

for responsiveness and so on. So this is the kind of infrastructure, the backend work that goes on to really make data visualization work. Yeah, so all of that's a part of data visualization. So when I say we are investing in data visualization, it's not just the user interface that people relate to, but all these server technology that needs to be, that needs to support this capability. Okay. Okay, yeah. I mean, fascinating, fascinating.

I mean, having written the book on 11G and all of the kind of web logic components in there, I guess that it strikes me that, you know, you've focused a lot on, I suppose, where the value is really to customers. A lot of the kind of infrastructure now, I guess, is what's really being taken care of by the cloud. So, you know, it's i mean did you do you see i mean with with with oracle bi i mean certainly 11g that i used to use there's a lot of focus on dashboards and answers and that sort of

thing do you see really in the future that the main interface people are going to have is going to be mobile devices and and search then really do you think that's the way it's going to go for for bi um yes no i i well i really don't know but but those would be important components um i don't know if ever in the future someone would say hey i don't want a pre-built report right because if you look at you know most of our lives there are things that we just um we just want it easy and simple and the questions do not change. Like I want a weather forecast. The format in which weather gets presented to me, it may not change. What probably would change is that I get informed. I mean, if I have something like I have a calendar

saying I'm going to go take my flying lessons, I want to know if it's going to rain because the class may get canceled. So getting push information in context is definitely where things are going. Anything where I have to ask is the next step. So if I have a question that I don't usually ask, then having a search interface is the best thing. But there are questions that I ask every day. Like I work with this railway company who have this connection metric.

How many of the rail cars, they're a cargo transportation company, right? So how many of the rail cars actually got to their right connections every day? And if the ratio is more than 93 percent everybody gets paid their bonuses but if it's less they don't so such in metrics are somewhere to be shown in displays right they are permanent the format is the same and so the dashboard technology is still very relevant now we so there are there are questions that we want to ask every day and we don't want to do us give a second thought where ports and dashboards make perfect sense there are new questions that i may ask where search or ask or you know natural language all those interface make sense and there are things that i don't think of asking

i want to know in context where definitely a push-based notification makes the most sense okay okay so so i want to get under cloud in a moment because there's a huge amount of stuff you guys are doing there as well but so you mentioned earlier on though that you you're responsible for s for exolytics one point and uh you know i actually i actually had i bought one in the past it was sitting in my garage for a long time and uh and and uh you know it was fantastic but certainly there's a lot of those were sold and there's a lot of value in there as well where did some of the thinking of exolytics come into this really and how did you sort of like incorporate some of the stuff that that you've learned from that and say maybe from

airspace and so on in the model that you're using for dv yeah yeah you you know a lot of background about oracle bi so i think you're asking great questions. So what was really new with Exolytics is given a couple of terabytes of memory, how can we actually provide better value? So the hardware technology is at a stage where it's actually, for a reasonable price, we can actually get a few terabytes of memory. So storing data in memory, processing data in an in-memory format, all of those are the ones that we invested a lot in Exalytics. So there are two use cases that we actually took forward and actually drove the value for Exalytics. One, how can we store or cache data in an in-memory database and accelerate, you know, queries that are most often asked by users. So, suppose somebody has a couple of petabytes of

data in their, you know, big data or it could be terabytes of data in their data warehouse, and they want to do some reporting on top of that, that is reports and dashboards that people log in and see every day. How can we make this experience easier? And we did that by intelligently scanning and understanding the workload, creating aggregates whenever the entire data would not fit, or getting the entire data into an in-memory database and delivering from there in a seamless fashion. So by understanding that workload, bringing the data in, moving the data, creating aggregates, creating a memory database, you know, table sources, and delivering

the value, right? That's one investment. The second investment was really in S-Base and multi-dimensional database. So how can we do an in-memory representation of, you know, the data that S-Base uses in reporting? I mean, for people who don't know S-Base uses in reporting. For people who don't know S-Base, it's an extended spreadsheet database. That's how it started. So it's to handle spreadsheets that are of size and complexity much larger than, say, Excel can handle, where multiple people need to work on the data at the same time. And we provide a pretty convenient Excel interface to access it, but the data is all stored in this multidimensional database.

And so we came with an in-memory version of storage for S-Base. So those are two innovations. And we are taking these two innovations in the cloud. In the cloud, you do not have to buy specific hardware, right? So we provide scalable infrastructure. So depending on your memory needs, you'll just get the compute instance with that memory. Or in a completely managed service, we'll automatically allocate more memory as required.

And if you want to put more resources, you just pay for more resources. So there are two areas of continuing to translate into our cloud. Not all of them have been released as product features yet. We do plan to have an in-memory database caching layer in the cloud. Our in-memory database, Oracle in-memory database, is actually already out in the cloud. So we could use that. Today you have a customer have to put that together. We will

come up with a way to create that automatically and manage. On the S-Base side, we already have big enough compute instances where S-Base can run in memory. But that's not where it ends. So if you run S-Base

in memory, customers who have S-Base can now take their applications to run in the cloud and get good performance. But the S-Base value goes way beyond that, right? So we could embed S-Base functionality into our data visualization tool and meet the needs of users beyond just, you know, understanding and getting insights from data. In what way? In what way? Because obviously, S-Base has got a very, very long kind of history here. And I'm sure there are lots of S-Base kind of developers and users, you know, listening

in at the moment. So imagine you're talking to a bunch of millennialsilials and you're you're you're kind of like pitching uh dvcs so you know oracle bi in the cloud and we'll get on to the cloud version of things in a second how would you explain to them what s-base does that they wouldn't be able to get from say a tablo cash i mean you know where's the value in there for them that really yeah yeah so what s-base is good at is providing a collaborative spreadsheet environment, kind of close to Google Sheets, really. I think a lot of people love Google Sheets, Michaels included.

When I have to do my checklist, I do my travel plans and all of that, my wife and I can actually collaborate on the same data at the same time. That's good for such use cases, where SBS actually takes it to the next level, is actually you can define pretty complex relationship between data. You can have multiple dimensions, not just the X and Y axis. Plus, you can actually have permissions. So who can see what kind of permissions

so go a step higher a level there are tons of spreadsheet processes in companies that are still pretty uh you know pretty time consuming and really empty calories if you will yeah a phrase that you've used before well it's true isn't it it's it's things that you know it's tasks that you do uh it's tasks that you do that are more about how to implement things really that should be done already yes um and empty calories by empty calories i mean so let's take a specific example a concrete example right so a manager wants to get capital expense forecast from his team for next year. It's like how many laptops they need, how much hardware they need for next year for their development activities. So what the manager does is actually have a secretary or someone, some poor soul, who has to now send out this format of spreadsheet to like 10 reports of the manager, get inputs from each one of them,

and then cut and paste them into a single spreadsheet because we don't want one person's request to be shared with another. This actually is more acute in terms of compensation and bonus allocation and stuff like that, right? So they send this out and somebody has to collate all this together. And until everybody's data is in, there's not a good picture of how the spend is trending.

And now the total adds up to more than what the manager can spend. And now they have to go back another cycle, asking them to adjust their budgets and requests and all of that. So a lot of time gets wasted in just massaging spreadsheets and copying and pasting. And we are introducing a new feature in data visualization driven by S-Base where any data set, we could actually assign rows to specific individuals in the company. Permissions get set automatically. When they update some data, and they can do that right in the context of Excel if they want to, and they can update the data, and all the data gets consolidated in real time and is available for reporting.

So data collection, data collaboration is an area where we believe there's a bunch of unmet need with the same people that actually use um you know data analysis and visualization tools so that's our next uh you know next set of new things that we would roll out using s-based technology okay okay because certainly in the in the 12c release s-based there seemed to be more of a kind of like a data a query acceleration layer you know but you're saying there there's some very distinct kind of use cases separate to query acceleration that are particularly well placed for S-based to sort of to meet them. Yes. So query acceleration is an area where S-based does really well too. And that comes from... Yeah, I mean, I'm not suggesting to interrupt you there, but I mean, that's something that was a really good feature in 12c when it came out and nobody talked about it.

That is true the thing is we have a lot to talk about uh and and you know it's it won't come in the top three um so but essentially because as base uh you know stores data in a format that is uh you know what we call a block storage um and and to then the next version of the block storage is storing the spreadsheet in all its granularity as blocks. And to make reporting easier, S-Base also automatically creates aggregates on top of these, of the data that is stored.

And because it automatically manages aggregates and is able to run queries on top of that, it's a very good tool for us to do query acceleration as well. And in Exalytics, that's another way that somebody could accelerate the API queries. These are in our roadmap to actually make it automated. So going into cloud, we don't want to expose all this technology to customers where they need somebody who's really an expert to configure these acceleration methods. What we would likely do in the future

is to provide options in our analytics cloud where they could choose an acceleration option using S-Base and we would just configure it automatically. Okay, I'll take your massive hint there to move on to cloud. So we're going to do that now. So, I mean, the other thing, I mean, as you say, there's so much to talk about really but the other thing is this massive amount of

activity and investment you guys have been making around kind of moving behind the cloud so just again to set the scene for people just just kind of outline i guess how you got to where we are now so we've gone from on-premise then there was the bli cloud service uh and now there's our analytics cloud just just summarize i suppose kind kind of that journey there, really, and let's start to talk about the new kind of Analytics Cloud service that you brought out. Cool. Sure. So we knew BI is moving to the cloud,

and Oracle as a company has been really focused on getting everything cloud-centric for the last several years. And we started with deciding to take the most business user-facing aspect of the product, which is data visualization and self-service, to the cloud first. And that required us to develop cloud-native multi-tenant infrastructure. And that we released in 2014 as BI Cloud Service. So with that, someone could say, hey, I have 10 users that need analytics

and you just subscribe 10 users. We actually give you a production instance and a test instance. There's a web-based way to define a data model if they need to, or if they don't want to define a data model, they can just go with data visualization and start rolling out analytics to users. So that's where we started. And we then extended that to support our on-prem data models as well.

So the next step was existing customers that already have Oracle BI now can actually take their data model and reports and put them on the same cloud infrastructure. So it gives them a way to move from on-prem to the cloud. Then the third progression we did, third step there, was to provide hybrid capability. That is, many times customers want to move analytics from on-prem to the cloud,

but their data gravity is still on-prem. Okay, so they still have their ETL processes on-prem. They have data warehouse on-prem. They most likely have an on-prem application as well, like EBS. And so for that to actually take the first step into the cloud, we want to make it easy to do analytics on the cloud. And so we developed a technology called Remote Data Connector that can connect our cloud to on-premise databases. So they can move the BI layer to the cloud without moving their data.

And it's performant only because we actually invested a whole bunch of technology to make that really easy. So why is that a harder problem to solve then than people would think? Because, I mean, yeah, just maybe just a hole in the firewall and a JDBC connection would be enough, surely? Yeah, one would think so. So when we actually ran some tests on it, it turns out JDBC and ODBC connections are quite chatty. So they may amplify the effect of latency. And so we got to make sure,

pretty much every transactions that we do are actually done in one round trip. And so we actually did an abstraction layer on top of that, move the driver too close to the database. So we actually run the driver pretty close to the database and we just can ship the SQL queries on the WAN. And we had to invest in a whole bunch of security

and authentication on top of that. And this is really what enabled this to happen. And we've seen a tremendous uptake for customers to actually do this, to take this technology and adopt it. Yeah, yeah. So tell us about Oracle Analytics Cloud then. So what is this? And how does it differ from BICS?

And, you know, where is this going really? What was the driver for it? So Oracle Analytics Cloud is really for customers who want more flexibility than, you know, a multi-tenant offering would offer. What do I mean by that? So if you look at large manufacturing companies, banks, and in general an enterprise with more than a thousand employees or beyond, they want to define their own high availability strategy. They want to define

their own disaster recovery strategy. Many times they want resource isolation. And they also want to pay for resource and not just users. Because if I have 100,000 users, everybody may not be using. Let me manage resources more efficiently, and I don't want to pay by user. And so for those use cases, you know, Analytics Cloud offers them all the flexibility they need. So they can purchase and deploy our cloud service

per CPU, right? They can decide the memory. And we provide a whole bunch of API instead of completely managing it ourselves. For example, in BI Cloud Service previously in a multi-tenant solution, we take care of upgrades. We just notify the customers and we upgrade it over some weekend.

We automatically take backups and so on. And we manage the service level goals. Now, in Analytics Cloud, we provide APIs for all these primitives. That is an API for customers to backup, an API for customers to scale out this instance, scale up this instance to a larger compute, and then, of course, the opposite, scale down as well. We then provide APIs for them to upgrade or patch these instances. And also make them participate in larger clusters of deployments that includes Oracle Database, Big Data, S-Base, and so on. So you see where I'm going with that?

So more control provides more flexibility. Of course, that will require more knowledge and effort on part of customers, but that's where a lot of our partners come in, or they may already have IT infrastructure with expertise that actually can handle a more flexible, at the same time more powerful platform okay so this strikes me as probably a different type of purchaser um than being for bics so bics what's the different personas you'd have there presumably bics is

more of a departmental kind of user this sounds more like an it cell maybe what do you think on that and is is there another product coming down the line that maybe abstracts some of that away again but lets you have the kind of the customization ability whatever that you get with kind of analytics cloud yeah i think you're on the right track i mean we are thinking pretty the same way so there are three personas that we really build for you know first persona is the analyst so person who spends a little more time with data and they want to get insights for themselves, create insights and share with others. And BIC is a DVCS ideal for such departmental use. The analytics card is really for developer or IT type of use where they have an enterprise platform to roll out and they are dedicated people who knows and understands these.

They'll create data models or not, but they do care about availability and consistency of all this data and reports. And so, like, before an upgrade, they'll actually do a UAT test and, you know, roll it out in production. So, for developers, that's really where Analytics Cloud comes in, or IT, really. We do want to take this service to everybody, right? So even business people who may not even want to do things like analysis. They just want to have sensitive data access

and do queries on top of that. We do plan to come up with an addition that's even more easy for people to use it's definitely be multi-tenant and a single person can subscribe and and track the data over time on all the things that i have access to and answer questions on top of that okay okay so what about i mean so we'll be talking about cloud and isolation here but typically i think you mentioned early on about you know how we work across the firewall and so on. What's the long-term story about, say, hybrid deployments, really,

with Oracle? Do you see this? I notice there's a cloud machine you guys have got and so on, but what's your kind of vision around hybrid analytics and deployments? There are two sides to the story. Internally, we just want to develop one set of code that can be deployed in multiple ways, in a managed cloud, customer-managed cloud, Oracle-managed cloud,

or even in what we call the cloud machine, which I'll explain shortly. On-premises is something that we would probably take the same code line and release periodically. And there's other dimensions in which the cloud is evolving at Oracle. So we just released our bare metal cloud. We already have container-based deployment. So our plan is to kind of make sure we can deliver on top of bare metal cloud and also deliver on top of container-based deployment where people just pay for the amount of resources they use. It could be fractional CPUs or they could pay per call. Oracle Cloud Machine, on the other hand, takes us to a totally different dimension. So there are many cases where customers do not want to put their data in a public cloud.

And a public cloud, I mean, in a cloud data center that hosts multiple customers. And for that, we have one version of cloud called dedicated compute, where we have a completely isolated set of cloud infrastructure for a particular customer. But more often, even that does not work because of data residency and compliance reasons. So for those, what we have done with a cloud machine is to deliver a rack of servers. And you can actually get smaller than a rack, probably a third of a rack or something, a bunch of servers, where we also embed our core cloud infrastructure. So our software that manages our cloud. And Oracle manages this whole system through a remote connection or even a disconnected mode. But what happens is once a customer procures or subscribes to this cloud machine, they have Oracle Cloud running in their data center.

So they're able to then procure, utilize any cloud service that Oracle offers in their data center. So they're able to then procure, utilize any cloud service that Oracle offers in their data center. The APIs remain the same. So if they want an instance of database cloud service with four CPUs and 16 gigabytes of memory, the API call they would make to the public cloud and to this private cloud machine is exactly the same, except that this cloud database is actually getting created in the cloud machine in the data center. So likewise, they could create an analytics cloud instance, create a big data cloud service instance, and they can consume it there. So we're getting the flexibility of the cloud, which is basically API-based management, pay-for-use, and efficiency, you know, density,

all of those benefits are available in a server that actually runs. Interestingly, the price they pay and the way they pay it all remains very similar between the Oracle Public Cloud and Cloud Machine. Yeah, fantastic. Well, I'm conscious of your time. There's one more question I want to ask you, actually, and that's about, so BI applications. So BI applications is a, you know, certainly in the U.S.

we seem to be a really common way that we see Oracle BI being, you know, being used at customers and company sites. So what's the story around BI apps? And is it gradually being kind of phased out? And is it now more about kind of DVCS against SaaS apps? What's the story around kind of, you know, I suppose operational reporting, reporting against, you know, line of business applications, really? Okay, so for those that do not know what Oracle BI apps are, that is, so probably I'll take a

minute to explain that. So over many years, over the last 15 years, because our knowledge working very closely with, you know, various application teams, including EBS, PeopleSoft, Shady AdWords, and so on, and now recently with Oracle Cloud applications, we have developed a standard data warehouse model that any customer could use and then built pre-built analytics for finance, HR, projects, ERP, supply chain, and so on. In fact, there's another layer of industry-specific data models that have also been built by our various business units at Oracle for communications, retail, hospitality, and so on. Specifically for the BI, this is what we call BI apps. And these BI apps are very popular because they have prepackaged the whole analytics stack,

including data models and reports for core use cases. Now, these are pretty mature. And so we are not really investing in adding more content to these, but we are committed to maintaining connections to these data sources. So we'll make sure we update the connectors to our SaaS sources and the various on-prem applications and their versions out there.

But the new strategy at Oracle is that we are okay to actually give this technology to a customer who's willing to deploy this on the cloud, right? So as long as customers are using Oracle Cloud, Analytics Cloud to kind of develop and deploy, you know, this data model, the customers can get access to these data models. So, you know, on a subscription basis, they will be paying for Analytics Cloud, the database cloud that is required to run databases, the, you know the data integration cloud. On the other hand, the content itself, we believe, is a great starting kit or starting point for customers to develop their own customizations. And we have a wealth of partners like Ritman Mean and others who can actually help customers get the best value by customizing and maintaining it.

Yeah, I think certainly in a way, I suppose the way we develop BI projects has changed quite a bit over the time since BI apps came along. And I think that the model you're taking with DV Desktop and maybe with OBI 12c, where maybe content might come in initially through desktop application, users will be very much participating in the definition of that metadata. It's quite the opposite, really, I suppose, of BI apps, where it was a predefined model that you kind of work with and so on. I mean, it strikes me that probably a book that would be written about Oracle BI development now

would be quite different to a book written, say, five years ago by me, for example, saying how it was developed then. Do you think it's quite changed a lot how we do things now on the platform? I think it's kind of in a flux, right? So I do see customers coming in and saying, hey, they do need all these pre-built analytics to get started. But the amount of innovation that happens, the customization happens on the top is actually, you know, evolving rapidly. So in fact, we ourselves have realized this and started developing starter kits for specific roles in specific departments on data visualization.

So without having a data model, we're actually building, if I'm an HR director, what are the various projects that I would like to do, analytics projects that I like to do? And we are building this and providing as a starter kit. So at this time, you know, we believe because there's so much desire to customize and innovate and do it yourself kind of model.

So you're better off giving starter kits, which gets get people started at a much more much higher productivity level than if they started from scratch. Interesting. And last question. Last question. Oracle Synopsys. I saw this announced. Is it Synopsys, the mobile tool that I saw announced just before Christmas? That looked very interesting. What was that? What's that then? Tell people what it's about, really, and how it looks. So Synopsys, I'm really super excited about it. I have it on my phone. Synopsys is a free application.

So anybody can download it. It's already available on Google Play or shortly available in iOS. And Synopsys is a tool. It's a simple app that lets you visualize any data you have on the phone. So if you already have some data on the phone, it could be a spreadsheet in a Google Drive or an email attachment or whatever, just open that in Synopsys. And in two seconds, you're going to be visualizing the data on the phone. It's not going to ask you tough questions like, oh, which is a metric? Which is a dimension?

Or what should I have on the y-axis, what should I have on the x-axis, what chart type? It's not going to ask you any of that questions, right? It's just going to visualize the data and show you a summary of data. You can always go back and customize it, but it gives you a synopsis, visual synopsis of

any data you have on the phone. It's completely offline. You could actually run it offline. Our goal is, you know, this will then synchronize with any data source that you define in your BI cloud service or analytics cloud. In which case, the entire enterprise data that channels through that

is available for you on fingertips. But right now, you can just play with it with any data like i think you probably have a bunch of strava data yeah i do yeah i mean yeah exactly i mean i the latest article of britain foracle magazine is on using strava data and withings uh wi-fi scale data to to kind of plot my path to try and keep my weight off actually this year so last year so yeah i mean it's it's there's i think there's so much data that we generate now and we have available to us it's great to be able to just work with it and play with it really

yeah and your article was great um you know thanks for that you should probably try that on synopsis i will do i will do now i'm conscious you've got to go now so look at fast so it's been fascinating speaking to you really really good um and it's great to talk about oracle bi as well for once now which is good um so look thanks coming on the show um and uh maybe hopefully have you back at some point in the future absolutely mark i really enjoyed uh you know speaking with you here and i listened to your podcast and it's great to be on the other side so thank you very much fantastic okay Thank you.