Top 7 Success Factors for the Cloud

Barbara: Hi, I’m Barbara Call, Senior Director of Content Strategy and Operations at IDG; and I’ll be your moderator for today’s webcast, Top 7 Success Factors in the Cloud. I have three speakers with me today.
 
First up is our moderator, Chris Twogood, Senior Vice President of Marketing at Teradata. Welcome Chris. Also joining us is Noel Yuhanna, Vice President and Principle Analyst at Forrester. Glad to have you Noel. And last but not least, Craig Rowlands, Senior Executive Information Management at Metibank. Nice to have you Craig. And now I’ll hand things over to Chris for our presentation. Chris, over to you.
 
Chris: Right, thank you Barbara and welcome to the Top 7 Success Factors in the Cloud. I’m Chris Twogood, SVP of Marketing at Teradata. And I’m very pleased to be joined with Craig Rowlands from Metibank and Noel Yuhanna from Forrester. Good morning guys, good afternoon and good evening.
 
Craig: Good morning Chris.
 
Noel: Morning.
 
Chris: We’re going to do a very interesting kind of format here today in this webinar. We’re going to introduce some of the things going on within the cloud, from a really positive perspective and why people are moving to the cloud; but we’re also going to talk about some of the core challenges and the gotchas that people have to look out for when moving to the cloud to drive success.
 
So before we started though, Craig, I’d like you just spend a few minutes talking a little bit about Metibank, what you do at Metibank and a little bit about your cloud journey. Craig?
 
Craig: Thanks Chris. Well, Metibank a product health insurance fund. We also got diversified in the health services. We’re the number health insurer in Australia, servicing approximately 3.7 million members. And our diversified products include life, pet and travel. As I said, with a health services side where we’ve diversified in. We do health and well-being programs, health offerings include female _____, cardiac rehab, most recently dialysis and _____ of care.
 
As for the cloud journey, and specific to Teradata, we were on the local datacenter, just as many others were; and we were looking at how we resolve for outcomes, both from an _____ risk or an operational perspective. And for us, really, is to have a look at can we increase our capability; can we resolve for our operation reporting, of which one is very key for us, which is annual tax statement; and the final one is kind of to reduce our risk. So that’s the path that we took into moving Teradata from a local datacenter up into the cloud and _____ to give us all of those things that we were looking for. And I can explain more as we go along.
 
Chris: Great. Hey, Noel, can you introduce yourself and just talk a little bit about what you do at Forrester and kind of a little bit about some of your perspectives on the cloud?
 
Noel: Yeah, absolutely. So I’ve been with Forrester for more than 15 years now; and I focus on data management, data warehouses, databases, data integration, data fabric, looking into the cloud and hybrid cloud and multicloud environments.
 
And I speak to about three to four customers with end-user customers every day on inquiry calls, mostly large billion dollar companies from various vertical industries, such as financial services, healthcare, retailers, who actually always engaging on doing data management, data and analytical strategies and such. So definitely a lot of action is going on today, especially with the cloud, I know. We’ve been getting a lot of inquiries on the cloud.
 
Chris: Good. Well, obviously speaking to two or three customers a day, you’ve got to have a really good perspective and kind of a pulse on what’s going on in the industry. When we look at the cloud, I mean quite frankly, the future of business is in the cloud. I mean we see everybody moving to the cloud. It’s not really a matter of are they going to do it, it’s just when are they going to do it.
 
And some of the key drivers that we see is that they want more speed; they want more agility; they want more flexibility that they can drive innovation. They also want to be able to access this unlimited compute and storage that’s available, rather than having to bring stuff on premises. Craig, what were some of the drivers that Metibank went through in terms of thinking about moving to the cloud and why they were going to move to the cloud? Why didn’t you just stay on premises? Any thoughts there?
 
Craig: Yeah, Chris. There’s a number, I don’t know to start. Maybe we talk about one of the success stories; and I think that will – that’ll give you an indication of why we moved. Back in 2016 we had a fairly catastrophic outcome with regard to completion of our tax latents(?) on behalf of our members. That raised its ugly head to the regulators; we got some very bad press; and we lost a lot of money on the share promise.
 
Since our move to the cloud, and where we were before, we were in a relatively comfortable position, but this year, using Teradata and actually being on the cloud, we went from being one of the also ranks to the first _____ insurer to launch successfully with the annual tax office. And that’s a massive improvement for us. So that’s a success story in itself. There’s a number of other success stories about the reasoning about why we wanted to do it.
 
But essentially, what we were looking for was to be able to resolve for 200 plus user base and make sure they could utilize application; ensure that we’ve got an increased uptime for people, so that those facilities and the systems that they’re available for them 24 by 7, 365 days a year. And for us from an operational point of view, to meet those targets that we were clear to being able to meet, not just from an analytics and reporting perspective.
 
Chris: And that’s really interesting. I mean going from such that catastrophic failure to such success and being the first one to lodge, is a great, great success. And then just obviously agility and flexibility and making sure all the resources are there is really good. What are you seeing Noel? What do you hear from all those customers? Is unlimited compute and storage, is that one of the core drivers? Or are you seeing other drivers as to why people are moving to the cloud for analytic needs?
 
Noel: Well, Chris, obviously agility and unlimited resource, as you pointed out, have been definitely one of the key reasons. But also cost savings. I mean you save money, especially when you go from a larger environment into the cloud, you look at saving at least 20% or more in the cloud, from on premises. So when you go from on premises to cloud, a 25% up savings is actually there, especially with the compute and storage and what have you.
 
But some things also improve security actually as well. Because there’s a lot stronger security measures being taken in the cloud today. Some organizations don’t have a very strong security footprint on their datacenters. Hence, they prefer the cloud as well. Or then there’s also this new features being rolled out in the cloud very frequently, as well. We see that features, new features coming out in the cloud more often than the on premises as well.
 
But I would say the biggest real key benefit has been the whole fast(?) benefits. I mean it’s also – I mean the cost equation that you can actually get big money is one of the big reasons why organizations are moving to the cloud.
 
Chris: It’s interesting that you say that, that 20% savings in terms of cloud costs over on premises. But there’s also the big savings of just not having all of the people and the infrastructure and driving all the support as things are deployed more as-a-service. Craig, did you guys see cost saving as you kind of moved from the environment? Was that one of your core drivers?
 
Craig: Yeah, Chris. I know lately(?) hit the nail on the head. I mean cost is certainly one of them. So it was a big driver for us. And we had lots of productivity outcomes. If you think about the increased automation and that there’s also data _____ talities that you’re looking for. The cloud and specifically a _____ platform gives us a lot of that uniquely.
 
Just an example for us of when we moved from version 14, which is among the Teradata versions, the version 1610, you’ve got a lot of automation and queries. So you’ve got a lot of improvements in those queries, a lot of efficiency straight out of the box. Now you don’t get that from a local data(?) and you don’t get that from on legacy key. The kind of thing that you get is a productivity outcome completely out of the box.
 
The productivity savings, the cost savings, as Noel said, from a capacity point of view, you can elastically move up and down. So as and when you need it, you’re there. Whereas in a local datacenter, normally you’re provisioned and you normally over provision, and you remain on that for the time period of that licensing agreement. So I completely agree with Noel. That was one of our drives. Increasing capability, as I said, was another one. But that has been a good outcome for Metibank.
 
Chris: So it’s about flexibility, it’s about cost, it’s about more capabilities. So I think that’s really good. One of the other things that we see a lot is all these different cloud architectures. Sometimes people say hey, I’m going to go deploy in a public cloud and I’m going to put 100% of my environment in one cloud, one cloud service provider. Other companies are doing more hybrid cloud architectures, where they have something on premises and something in the cloud. Maybe like a sandbox or more of a test and dev or a DR system.
 
And then others are really looking at more multicloud, where they’re actually spanning multiple different cloud services provider, like maybe something in AWS and something in an Azure. Noel, what are you seeing in terms of different cloud architectures and how are people adopting these different environments?
 
Noel: Yeah, that’s a very good question. You know, based on a recent, very recent survey we did, 65% of large enterprises are doing hybrid cloud strategy, 65%. So it’s a big number. And the large organizations actually have lots of data on premises, especially in the financial services, especially in the healthcare business, also manufacturing companies, plants, they have lots, lots of data.
 
So we only believe very small percentage of data has been moved into the cloud for these companies, just because they have lots and lots of these large datasets. And they have a lot of transactional systems running also on premises as well. Definitively hybrid cloud is still going to be there for a while. And some people are concerned about going to the cloud like, financial services and healthcare, because of sensitive data, PII, PHI data. They don’t want to go to the cloud. In fact, 30% of enterprises are concerned about data in the cloud.
 
And the number one concern is security, actually. I mean compliance and security. But multicloud is also gaining traction. About 25% of enterprises are doing multicloud, because they want to take advantage of the native capabilities of these scenarios. And we expect this 25% to maybe double in the next three to four years. Certainly there’s a lot of movement happening in multicloud. Nobody wants to go on a single platform because of vendor lock-in, because you’re concerned about vendor lock-in.
 
But also, when you have multiple clouds, you actually take the advantage of all the best of these services from various cloud providers, which actually helps you in doing the data and analytics strategy more efficiently as well. So I would say both hybrid cloud and multicloud remains an important strategy, especially when you’re in the larger billion dollar companies who have got a lot of their sensitive data and legacy data platforms.
 
Chris: I think that’s really interesting, like 65% still doing a hybrid kind of cloud architecture. Craig, what did Metibank decide to do? Did they go all in the cloud? Are you hybrid? What’s your kind of architecture?
 
Craig: Yeah, Chris, we are hybrid. And there’s are reasons behind that. And Noel has touched on a few of them with regard to the sensitive data. You’ve got to remember, globally, different regulators have conditions. So some of our core systems, our originating systems can’t go into cloud as this moment in time.
 
But certainly from a data perspective, just because of I think the enhanced security and the way that we’ve been able to successfully demonstrate to the regulators about how we look after that sensitive data that Noel mentioned, PII and PHI data. So that was part of the reason for us to migrate all of our data, as in our data warehouse and our data landscape. And we have a cloud-first strategy at Metibank as well, Chris. So just to be clear on that one.
 
For us, we saw many advantages in moving over for the data side. We do have, obviously, a few issues when you talk about origination systems being in the local datacenter. So there’s a small amount of latency, but it’s not a great deal of latency. And in in the insurance side, particularly the health insurance side, from that point of view, a lot of them are remained _____ and yes we have a lot of real-time stuff, but there is still a tremendous amount of overnight _____. So that’s our decision for hybrid.
 
I also like Noel’s comment about multicloud strategy. I think that’s an important factor, where you want to be looking at horses for courses. So which workloads are with which cloud service providers also? And that’s an important one. Noel’s point about lock-in I think is incredibly valuable one.
 
Chris: You know, it’s interesting. Teradata, we also have a cloud-first strategy and, in fact, we support the Big 3. So we’re in Google Cloud, we’re in AWS; we’re in Azure. And one of the things we’ve done, Craig, that I know you’re familiar with, is we’ve make our software available with portability and free licensing. So as you start – if you’re on one cloud and you’re like hey, I feel like I’m getting locked in and you want to move that software over to another cloud, or you want to move it on premises into a software infrastructure, you have that flexibility with that license.
 
So these are obviously great things about the cloud. I mean unlimited resources, different architectures, getting value with security and capabilities and flexibility. But there are also some gotchas that we have to look out for. So let’s look at those 7 top success factors. The first one is estimate future budgets based upon your workloads. Don’t trust, let’s say, a vendor that comes in and says hey, I’ve got a small demo or a small benchmark, there’s a TPCH kind of benchmark, which is 22 kind of decision support queries, and say oh, look how fast it runs. And that means mine’s going to run fast.
 
You’ve got to make sure you’re thinking about sizing based upon your concurrency. Noel, how do you think about really advising some of your customers about cloud sizing? Any thoughts about that?
 
Noel: Yeah, yeah. I mean it’s a very important area of focus, especially budget. I mean everybody’s not got any kind of like a budget issues running into some of those challenges for companies. But I would say you have to look at the current workload. I mean what is the most data you have? What do you plan to store? How many users are going to run on the environment. What kind of concurrent connections are you going to have? What kind of high availability, disaster recovery are you going to have?
 
And so you go and size the compute, you go and size the storage, you do the configuration of these platforms to really determine what kind of sized environment do you need in the cloud? But you also have to look at the new requirements, not just existing. When you move from the on premises, you just don’t have to move your existing environment as is, but also have to look at this newer things you’ve got to do with this platform, such as doing maybe new types of queries, new types of workloads you’ll be able to run as well. And so you have to provision those are part of the whole sizing exercise as well, not just about existing workloads you’re going to be running.
 
So the storage matters, the computer matters, in terms of short term. But the good thing with the cloud is that you can configure it, you can scale up, scale down on an on-demand basis, which I think is very powerful in the cloud. So you don’t have to go like an on premises provision it completely. So that’s the advantage in the cloud to be able to be more flexible.
 
But from a sizing perspective, suddenly there’s a flexibility here in terms of what we want to do, in terms of deployment. But you start as a basis as to what you already have and look at that as a foundation, in terms of use case-wise, and then obviously keep on adding additional requirements, in terms of queries and workloads running in those platforms.
 
Chris: So Craig, how did you think about some of these challenges when you were scoping the size of putting Metibank in AWS?
 
Craig: To be fair Chris, it wasn’t just me. It was actually Teradata as well. So before we migrated over, we sat down with a project team and we looked at what do we want to send over there, first of all. So when we talk about gotchas, the first thing, and the great work that was done between the two teams from a collaboration point of view, was let’s clear the deadwood. So let’s not migrate over what we don’t need to, because clearly that’s going to have a cost. Cloud is a consumption model.
 
So that was the first thing. So cleared up what we don’t need and only migrate over what we do. So that was the important factor. And Noel will say, that will take down costs straight away, because keep doing that.
 
I think a second part around the processing was incredibly important around how do you use elastic so you can scale up and scale down? Other elements behind that will be, for the test and dev environments, they’re not there all the time. And so to use our word, blow them away, as in when you don’t need them. Keep those costs down.
 
Chris, the other things, the kind of gotchas, and what we were thinking about when we’re having a look at the analytics and reporting community, how often are you actually on the system? Are you on 24 by 7, 7 days a week? Or are you on 9 ‘til 5? If you’re only on 9 ‘til 5, turn things off in the evening. Turn things off at the weekend. Turn things off over public holidays, etcetera. So there were really good thought processes around it.
 
And as we said when we moving workloads over with Teradata, it was how do we finetune? Noel touched upon it earlier. How do you optimize he queries? So if you optimize the queries, you get better outcomes. We did that. That work was done. So there’s a lot of planning went in to make sure it went very smoothly from a migration point of view. But actually the adoption post has been a success.
 
Chris: Well, and I think that’s key, Craig, and by ____ to hear you say that Teradata helped in front(?) of that process. Because we absolutely can help companies understand what is your actual usage; what kind of queries are you running; what kind of SLAs do you expect to get? And some of the other things too, is some cloud technologies don’t scale in a linear kind of fashion. I mean we’ve run many benchmarks; and as you start to scale up to 50, 100, 200, 300 and beyond users, some of the other technologies just start to flatten out.
 
And the other thing that’s important is a lot of the queries that you run are sub-second and sub type of queries. And so they’re not always evident when you just run something as basic as say like a TPCH type of demo. 
 
So let’s do the second one. And this ties in a little bit to what you were talking about, Craig. Get expert guidance when migrating to the cloud. Don’t listen to claims that it’s really simple. I mean it doesn’t have to be difficult; but there are different ways to do it. You can modernize all the way up front. You can do a lift-and-shift and then modernize from there. Craig, what was your approach in terms of your migration to the cloud? Did you lift-and-shift or kind of modernize?
 
Craig: Well, Chris, we lift-and-shift in this case. I will say before the lift-and-shift, as I said, from _____ earlier, we removed a lot of the data, so clean that down and then shift over what remains and what’s relevant. So that’s how we moved. And it took us all told, it took about so three, four months in the planning stage. But actually, for the moving itself, we’re talking weeks, a number of weeks. And it’s insane, but the ability to be able to finetune once we’re on the cloud has given us a unique outcome.
 
So some of the things I probably want to share with you are, since going on the cloud and the changes that have been made. Application uptime has increased by 17% for us; performing(?) outcomes, as in their efficiencies, 80%; uptime of a system, 100%; incidents that we’ve got from problem management, incident management, etcetera, it’s down 70%; user query speeds are up 50%.  
 
Because the things that we did when we moved, well, we continue to have those conversations, return(?) _____ harder, to make sure that we can resolve some of, I’m just going to call them teething issues. But actually for us it’s just a continual improvement direction that we’re looking to move in.
 
Chris: Yeah, those are some, I mean those are some amazing results, and not just around reliability, but also increased performance. So that’s great, great news and results and moving Vantage and Teradata on AWS. Noel, what are some of the things you’re seeing and what are some of the advice – what’s some of the advice that you give to customers about migrating to the cloud?
 
Noel: Well, I mean migrating to the cloud is sometimes not straightforward. I mean as I mentioned, you have to be first of all, be watchful for this whole security policies. In fact, what addressing last year, we saw a 40, 4-0, some large enterprises moving back from the cloud to on premises. And that’s obviously scary. Why you guys moving back from the cloud? And basically it was the auditors told them to move back, because largely because of security and compliance.
 
So I think we saw the same little(?) trend around offshore development(?), outsourcing for the healthcare, almost like a decade ago. So I mean this is definitely a concern. I would say definitely you’ve got to do some planning. You’ve got to understand what data moves; how you’re going to move that; what kind of cost savings do you achieve. There has also been some concerns by customers around cost savings.
 
In fact, some people were expecting 20% savings, but they are now paying more than 20% over on premises. Why? Because the services, the number of services have been growing in the cloud. So it may not have been as attractive anymore. So you’ve got to be watchful for your _____. But you also have to be watchful for security.
 
And also you got to look at what are the approaches to migrating to the cloud? And tools definitely help in the lift-and-shift. I think tools, when you’re migrating, you really look at the tools. It can help move data. It can help creating the schema structures. It can help creating the indexes. It can help creating some of the data pipelines. And really help you to really establish a more easily in terms of doing all this manually.
 
But Chris, what are some of the guidance you guys are providing for Teradata clients or customers wanting to migrate to the cloud?
 
Chris: You know, Noel, it’s really interesting because we actually, with Teradata and with Vantage, I mean you can do like Craig talk about, you can do a lift-and-shift. You can also modernize right away. You can integrate with a lot of those third-party cloud services, like Object Storage(?), like SageMaker and just kind of integrate it and modernize Vantage as your engine and your platform for data analytics, as you go.
 
So we really work with both type of approaches. But we actually recommend like lift-and-shift first. Get some of the advantages that you’re seeing here in the cloud, the agility, the flexibility, the new capabilities, and then start to modernize and take advantage of some of those new first-party cloud service integrations.
 
The one thing that we’re really cognizant of, we hear vendors out there just talking about oh, it’s low cost, it’s easy, it’s cheap. [right, right] Like there was one vendor that actually offered one of our customers $2 million in migration credits. And they tried to migrate to stuff and some of the queries didn’t work; some of the workloads didn’t work; they couldn’t manage the technical queries. And so you just have to watch out for that.
 
All right, let’s jump onto number 3. Avoid unexpected costs and keep your projects alive. You know, Noel, you were talking a lot about this. It’s great to have elasticity, but if you kind of have elasticity that gets out of control – in fact, I know Forrester has made a comment that 19% of organizations say it’s too difficult to track cloud costs. And I’ve seen some other analyst firms say as much as 75% will see cost overruns in the cloud. Well, how do you manage that? What are your thoughts Noel?
 
Noel: Well, I mean we are hearing that as well. I mean we are hearing a lot of customers are concerned about costs, especially running these queries, where these queries are taking longer. Some of these are runaway queries. Some of them are not optimized and tuned queries. So not if you’re running queries which are not optimized very well, well, they can take away a lot of resources, actually. Your compute resource can go away, but just a single query, if you’re running a table scan, for example. So this also be where things to be mindful for.
 
But also, how many things are you going to control? You really have to control, I would say, the processes to ensure the efficiency, optimize and running. ____ we are seeing companies making rather redundant copies of data. You not just make one copy, but multiple copies to scale. And when you make multiple copies of data to scale, you can scale, but then you’re running the cost, this unexpected costs. That’s because you scale. So you’ve got to be looking out for that perspective as well.
 
But then there’s also additional services, like monitoring, heat management, backup and recovery, integration, streaming data, may or may not be free. You’ve got to be look at, look after those services. Are they free? Are they not free, they are bundled? How are they bundled? All of these are really – could be unexpected costs for organizations.
 
Chris: Yeah, in fact, we’ve seen some of our customers that deployed a competitive solution, they thought their price point was going to be somewhere around $100,00 and then they get a bill for a half a million dollars, just because the brute force hardware continued to scale up.
 
Craig, how did you deal with the costs of cloud and making sure you’d get those costs under control? I know you talked earlier about shutting some things off if it’s test and dev and not using it. But are there any other thoughts, in terms of what Metibank did to control costs?
 
Craig: Yeah, look, Chris, the first thing I’ll say is, have a forecast of what you believe your spend is going to be. That’s the first thing. So you’ve got something that you can aim against. That’s incredibly important. And then once you’ve gotten that forecast, then make sure you can track your billing, to make sure it lines up with your forecast. I think both of you have already articulate the fact that if you don’t keep track of these things, you get a nasty bill show up and that’s not really where you want to be. That’s a difficult conversation to be holding. But the first thing I would say is keep track of your billing. Make sure you’ve got a forecast in there and make sure you’re tracking to your forecast.
 
Noel has mentioned some really good points earlier, where he talked about optimizing queries. We always say, in ours, from a DevOps perspective, optimize your top ten worst running batch processes and your top ten worst running steps within those. So Noel has already mentioned like blowing away your environments that you don’t need, most definitely are there. So I would agree with all of those.
 
For some cloud providers above others, the ability to reserve instances is there, both for an annual basis or for even longer. So if you know that you’re going to be using those workloads for 12 months or longer, reserve some of those instances. Others talk about using ____ instances(?). So get processing reduced cost as well, particularly if it’s not time relevant. So they’re really good outcomes as well. So there’s a number of those.
 
But the tuning of queries, the forecasting of what you’re going to be doing, the constant reviewing to see what’s idle. So what aren’t you using, and why are you using it should be in your monitoring. It should be in your alerting. So you’ve got those.
 
Noel mentioned one other really good point, which I love, which was these backups, these snapshots and all of these other things. People make multiple copies. Why on earth are they making multiple copies? You should be looking at your backup policies, your archiving policies as well, and utilizing those to the maximization benefit for yourself. So in our case, we resolve for what our regulators want from records management, but you’re definitely keep the cost down.
 
Chris: I think those are some really good points. It’s not like you just want to put your workloads in the cloud and then just forget them. You’ve got to manage them; you’ve got to optimize them; you’ve got to make sure that you’re looking at which ones are being used, which ones are not being used.
 
I had a customer say with some other technologies, they don’t have to manage auto scale, but they have to manage credits, because they’re trying to ensure that their financial governance is not going out the window.
 
All right, let’s move on to number 4. And number 4 is meet service level goals with software first. In fact, I heard that the more sophisticated your software is, the less hardware resources you need. Where if your software is very simple, it requires a lot of brute force hardware. Craig, let me go back to you. Are there some things that you saw when you upgraded to Teradata 16.X environment that there were some sophistication in terms of the software that enabled you to drive some of these core performance features? Any comments you want to make there?
 
Craig: Yeah, Chris, two or three. So there’s some that we’ve done and there’s some that we haven’t done yet, that we want to do. The ones that we’ve done, as I mentioned, you’ve got automated optimization of queries, from a legacy perspective when we were on an older version of Teradata, up to 16.10. And that in itself has been a really good outcome for us.
 
The native object storage as well, from what we are looking at is really good, from our point of view. And the optimization of queries has been really good from a Teradata perspective.
 
The bit that we haven’t done, and the advice that we’ve got from Teradata, which I think is very relevant, is we have – and I’m guessing a lot of other companies have got – we’ve got a lot of analysts on SaaS. And a lot of those individuals go and grab lots of information from Teradata, or any of our warehouse, and they drag it all back down; and then they utilize it in the local area.
 
Now the advice for someone like Teradata is they can do the grunt force of that and they can do the vast majority of that, actually, in Teradata and bring the results back. And so only work on the results in the analytical engine, which is far more efficient when you think about it. Anyway, so that’s – although we’ve got lots of things that we have done, that one is going to be a really good outcome for us, when you consider we’ve got 140 SaaS users in Metibank, for us to be able to do further optimization.
 
Chris: Yeah, so do a lot of the optimization in the engine, versus moving it around the environment and duplicating data and duplicating processes. Noel, what do you see in terms of the sophistication of different software that is deployed in the cloud and how that drives to performance?
 
Noel: Well, I mean if the platform, the analytics platform does not have the software sophistication, then you will require additional people, like data engineers and developers and architects involved to fill the gap, actually. I mean they would be doing all of the building of the functions to cube analysis and rollup(?). And which may not be there.
 
Maybe if you don’t have the in database analytical function, for example, they may be required to actually build those, which is taking away the time of these folks out there to support them.
 
So it’s all about the maturity level. It’s all about the bells and whistles which may be part of the whole offering. So definitely consider that as the requirement. When you look at these solutions out there, you’ve got to consider what kind of maturity levels do these platforms have; and what kind of things are going to be required in your organization from the analytics perspective? What kind of functions are going to be required so that they are supported and built in?
 
Because at the end of the day, no, after a year, you don’t want to find out that well, you really need to stand like 10 developers on this function, taking away some coffee time of those folks can do something as well. Chris, what do you see from this perspective?
 
Chris: Well, I think there’s a lot of interesting things. I mean like what Craig said around the query optimization. I like the native integration with Object Storage to really drive a low cost foundation. But also tell you what’s really interesting in the cloud is this core element of being able to drive tactical queries, have workload management. Elasticity does not equal workload management. You have to be able to manage your tactical and your strategic workloads against all of your different users that are going into this environment.
 
But I also love, Noel, what you said. In Vantage we built in a bunch of advanced analytic functions into the engine, including some of the SaaS and some of those – our own machine learning, that bring it all together, so you’re not having to move data around these different environments.
 
All right, let’s move on to number 5. Number 5 is demand flexible cloud pricing. Too often what we’ve seen is consumption is a great cloud pricing model. In fact, Teradata just announced ours here in the last couple weeks. But it also can drive out of control. If you’re just a pure consumption model and if you’re getting up into utilization of 90 and 100% of your resources, your costs can go pretty high. So you want to make sure you demand the ability to start zero, and pay for only what you use.
 
But also you want to have the flexibility to change that to a model that’s more of a blended pricing model that has some reserve with some on demand. Craig, what did you guys do? What kind of pricing model did you select from Teradata?
 
Craig: Yeah, we’re flexible. So Chris, this is – that’s where we’re looking. And I’m going to add, as you mentioned, it’s Teradata in conjunction with AWS. So the outcomes to us are twofold. We’ve got double the advantages, I would say, from both of them.
 
We look at, and have looked at, discounted multiyear offerings, ability to lock in reserve instances, as we talked about. There’s also the ability through conversations we did with sales, free upgrades to unlock the ability to realize productivity and cost efficiencies. Typically for us that manifested itself in automation. Clearly part of the programs health(?) as well. So remembering that when you got a strategic partner, work on the partner programs.
 
I would say take advantage of spare compute, capacity and AWS is another one. But I already talked about the on-demand, off-demand between something that you’re going to be using all the time and something you’re not. So that flexible and that flexible outcome is certainly an advantage for us with regard to where we are today and what we’ve got from you.
 
I think the – we call it the Teradata report – but that solution to provide a balance between fixed capacity and variable capacity is exceptionally good for us.
 
Chris: It’s just that flexible pricing, so that you guys are only really paying for what you need. But when you are driving really high consumption, that you’re getting the lowest possible costs. What are some of the challenges that you’re seeing, Noel, with kind of cloud vendor pricing models? Any thoughts?
 
Noel: Well, what the – it’s an interesting topic. It obviously becomes a complex topic as well, because we see a lot of vendors offering different pricing options, ranging from compute base; to storage base; to memory base; to query based; and consumption based. And you look at all these options out there, there’s so many different options.
 
I think it’s at the interest of the consumer here to pick up the best consumption model, or the pricing model which is best suited for their environment, based on what is the priority running the – what’s the workload running. So look at those very carefully. Because at the end of the day, no, you don’t want to be spending millions of dollars unnecessarily, especially in the cloud. Because adding up a million dollars could be very, very easily done, if you’re going to be going on maximum of all those memory and CPU or storage.
 
So the flexible pricing model is definitely one thing to consider. But also look at short term what is the usage going to be looking like. And then what are the usage which is more beneficial based on the model that would be driven?
 
Chris: Yeah, and so at Teradata we offer flexible pricing models. So we have for consumption, which is pay-as-you, what you need, and pay-as-you-use. Or you can literally start with zero down and no obligation. I mean how many of you guys have heard that from Teradata? And you can just pay for what you consume. And you only pay for successful queries. But we also offer blended pricing. So it’s a blend of enable you to have that fixed capacity like Craig was talking about, but also that on demand. So that if you’re driving very high utilization of workloads, you get a very, very low price.
 
But it’s also interesting, as we start to look at number 6 here, it kind of ties into what you said, Noel. How do you measure success in the cloud? What we say is you measure it by price performance. Price performance is the only metric. In fact, we see that some of our competitors are just measuring it based by how much you consume or how much CPU. But the reality is, how many _____ are you getting for what cost are you paying for?
 
So how do you think, Craig, about cost and measuring the price performance, and how you measure success from a cost perspective in the cloud? Any thoughts?
 
Craig: Yeah, Chris, and I’m going to give you probably a live example from us. So I mentioned already some of the advantages of using someone like yourself and particularly around an _____ Object Store(?). We were looking at what I would class as a really unworkable query for us. Where we were looking at, from Metibank’s _____ and joining a couple datasets. So single Parquet file, 71,000 records; Section 1 journal(?) file, Parquet file, 3.9 million records; Teleslide 1.5 gig, to go into detail.
 
Now that was taking an inordinate amount of time beforehand. Now doing those two _____policy and policy journal, using the Teradata outcomes, we’ve got that down to 20 seconds. So we need to talk about price versus performance. There’s the tremendous use case for us straightway of how do we bring the price down, because we optimized the performance straightaway?
 
So those native toolsets that were available for us are incredibly important. So that’s the first one. Obviously optimizing the environment and we’re looking also about the privacy versus performance. How much of it can we do _____ that SaaS analytics from in database perspective? Clearly, when you get down to them, and you know many providers do this now, that they provide cost per minute, cost per second, rather than they used to cost per hour, etcetera. So that’s another good utilization of them.
 
And I guess the third one for me would be – and this one’s going to be a little bit odd, and I apologize for this – but actually being able to deal with your strategic partner face to face. Some of the cloud providers actually make you go through partners before you can have that conversation to do the optimization. Then _____. That isn’t something that occurs with Teradata, or with AWS, by the way. So both of those conversations that we had where we’ve got that unique blend of both of you. So that’s why we chose to go I think from some of them, and that’s why we _____ performance discussion with some of those individuals.
 
Chris: That’s interesting. I mean having that relationship and being able to work on those core problems together, I mean that’s always been a cornerstone of some of the core value that Teradata brings. You know, it’s interesting, Craig, we did a recent benchmark against one of our cloud native competitors. And in that benchmark, exact same hardware, they ran about 50,000 queries. In that same timeframe we ran 600,000 queries. It’s all about this cost per query and the price performance you’re getting as part of it.
 
Noel, what are you seeing? I mean do you agree with this concept of price performance as the only metric?
 
Noel:               00:43:28 – Well, you know, everyone wants the best performance of their data went out to the cloud, at the lowest cost, right? So that’s the bottom line, right? I mean you really need that performance. So I mean you want to do more with less. So absolutely. The real reason why people go to the cloud is to save money. So suddenly, if you can do more of these queries, more of the data processing in the cloud at a lower cost, absolutely. That’s exactly what customers would want.
 
So price performance matters a lot. I think especially as more of the data grows, the data grows obviously the factors. But then also this need for real-time requirements are growing. So when you have real-time data requirements, real-time analytics requirements, the price performance becomes even more important. Why? Because you are now running many more users, many more on-prem users. You’re running many more of these very fast queries. And that’s the time when you really have to look at how many queries will you run within the next 10 minutes or in an hour? I think those really make a difference.
 
Also we know, people want to really make sure that they can provide you a better TCO and ROI. I mean and the whole notion of cloud is to provide a better model, as such. But again, price performance I think is definitely one of the most important elements of decision-making about who the vendor would be. And also, why go to the cloud I guess, except(?) for that reason.
 
Chris: Yeah, we absolutely at Teradata, we offer the lowest cost I mean at scale. Because you can run massive amounts of queries and get the kind of throughput that I was talking about before. All right, our last one of our top 7 success factors in the cloud. And I know Noel, this is one that near and dear to your heart, ensure end-to-end security. Come on Noel, why should anybody care about security? What do you have to say about security in the cloud?
 
Noel: Yeah, I mean end-to-end security is a big important, right? I mean as I mentioned I think earlier on, we’ve had some incidents where organizations moved back from the cloud to on premises because the auditors told them to move back. And security is very, very important. It’s the number one concern.
 
When you look at end-to-end security, it’s all about integrated security, with your ingestion layer; the processing layer; the storage layer; the transformation layer, all together working with a common policy, working with an integrated security policies. Definitely this is something to be put on the top of the list right there, that everyone moving to the cloud should at least start from a security perspective from day one. Forget about day two, day one they should actually put emphasis on security, having a security team involved with the analytics team when they move – when you do the lift-and-shift.
 
But also they’re building these new platforms the cloud, you want to make sure that all of these things are taken care of from a compliance perspective, from a security perspective; but also you want to take care of all of the bells and whistles, like the encryption of data, the masking of data, auditing of information as well. So you’ve got to leverage all of those things. Really, especially when you’re dealing with sensitive data, like PII and PHI data, especially those things. So absolutely, it’s important.
 
Chris: It just doesn’t happen magically in the cloud, if you go to the cloud security is taken care of?
 
Noel: Yeah, yeah, maybe I should – a magic kind of thing. Does not.
 
Chris: Hey Craig, what did you experience when you were thinking about security in the cloud?
 
Craig: Yeah, Chris, I mean we got a basic premise, particularly from a data perspective. As we’re designing what we’re looking for – and that will be regardless where it is – but certainly as relevant on the cloud as anywhere else. When we’re designing it for the user, are they clear on the data they require and from what – so only use it for what you need. There’s a time requirement on it, typically for projects use _____, etcetera. As in a time requirement as well on the expiry of that data. So they’re all there from a user perspective.
 
We start with a user and putting it in their mind; and then touching up on the things that Noel has said. You’ve got other things in there. So you’ve got the physical security that you’ve got from a datacenter. So a lot of datacenters don’t even allow you to walk round them anymore. It’s a virtual one. Whereas back in the day, you used to be able to go around them.
 
I remember a few years back going into one interview, kind of you had to go over a moat and it had darkened glass and it was bullet proof. And that was many years back. This is an Experian one back in the day for a reference agency. So you think about those just in historic terms.
 
In our terms, Noel has already touched upon these and these are a service really where by Teradata – but encryption _____ rest and in flight. That’s incredibly important. And have you got the safety behind all of those? Can it comply with a role-based access outcome? So you’ve got proper user access management. Have you got your backups encrypted as well?
 
I think for us, another important element and one that played to us really well was, there was a view that we had around multitenancy versus single tenancy. And we believe a multitenant data and analytics platform exponentially increases the risk. Teradata was kind enough to provide us a single tenancy platform, which actually reduced the risk of exposure. So if you see those examples that we’re looking for, and then you give us really good outcomes.
 
Again, I defer to Noel, so I think he’s right. Masking is incredibly important of information. So don’t show information that isn’t there. So the example that we all provide is, from a credit card, to do credit care analysis, you don’t need all 16 digits of a credit card. You actually need digit number 1 and you need digit number 16. Number 1 gives you the Mastercard or Visa card detail so you can do your analysis. Number 10 gives you the ability to pull in a control group in there. The rest of it can be completely anonymized, so you don’t know. – That’s the beauty of the products that are out there today. And I think cloud services that incredibly well.
 
Chris: Well, yeah, and I’ll tell you from a Teradata perspective, we take security very seriously. I mean not only do we have our security experts that can help you lay out a security strategy, but to your point, Craig, we offer single tenant, which reduces the strategy elements. We also do the full certifications in the cloud.
 
Well, those are our top 7 success factors for the cloud. Estimate your future budgets on your workloads, just don’t take the word of a small TPCH benchmark and 22 queries and expect that to be your workload. Get expert guidance when migrating to the cloud. You don’t want to just take somebody’s word of it that says it’s easy. It can be easy, but you need some advice. Avoid unexpected costs and keep your projects alive. Manage it. We got a lot of good advice from Noel and Craig about how do you think about costs? How do you think about managing what’s going on with that environment?
 
Meet service level goals with software first. Make sure you choose a vendor that has sophisticated software that can optimize performance with software, and then with really hardware and being able to scale in that dimension. Demand flexible cloud pricing so you’re not locked into just one type of pricing model, as your usage changes. Number 6, measure based upon price performance; and look at cost per query as a main factor. And then make sure that you’re enabling end-to-end cloud security.
 
So both Craig and Noel, thank you so much for your expert guidance during this event and helping to our customers and prospects out there really think about how they’re continuing to drive a success in the cloud. So Barbara, I’ll turn it back over to you. Thanks guys.
 
Craig: Thank you.
 
Barbara: Thank you Chris, that was great. That’s all we have time for in today’s presentation. Thank you again to my speakers, Chris Twogood of Teradata; Noel Yuhanna of Forrester, and Craig Rowlands of Metibank. For more information on this topic, please visit the resources section on your screen. And thanks for joining us. For IDG, Teradata, Forrester and Metibank, I’m Barbara Call.
 
END