Listen to the podcast. Find it on iTunes/iPod. Read a full transcript or download a copy. Sponsor: VMware.
The latest BriefingsDirect roundtable discussion focuses on two prime examples of organizations that have gleaned huge benefits from high degrees of virtualization and aggressive cloud computing adoption.
Join here executives from Revlon and SAP, who recently participated in a VMware-organized media roundtable event in San Francisco. The event, attended by industry analysts and journalists, demonstrated how mission-critical applications supported by advanced virtualization strategies are transforming businesses.
The discussion examines the full implications of IT virtualization, and how accretive benefits are being realized -- from bringing speed to business requests, to enhancing security, to strategic disaster recovery (DR), and to unprecedented agility in creating and exploiting applications and data delivery value.
Our guests are David Giambruno, Senior Vice President and CIO of Revlon, and Heinz Roggenkemper, Executive Vice President of Development at SAP Labs. The chat is moderated by Dana Gardner, Principal Analyst at Interarbor Solutions. [Disclosure: VMware is a sponsor of BriefingsDirect podcasts.]
Here are some excerpts:
Gardner: What's going on with your internal cloud at SAP, and why is the speed and agility so important for you?
Roggenkemper: If you look at SAP, you find literally thousands of development systems. You find a lot of training systems. You find systems that support sales activities for pre-sales. You find systems that support our consulting organization in developing customer solutions.
From a developer's perspective, the first order of business is to get access to a system fast. Developers, by themselves, don’t care that much about cost. They want the system and they want it now. For development managers and management in general, it’s a different story.
For training, it's important that the systems are reliable and available. Of course again for management, it's the cost perspective. For people in custom development, they need the right system quickly to build up the correct environment for the particular project that they're working on.
Also these requirements are much better supported in the virtualized environment than they were before. We can give them the system quickly. We can give them the systems reliably. We can give them the systems with good performance, and from a corporate perspective, do it at a much better cost than we did before.
Our business agility and ability to respond to market drivers is greatly improved by this.
Gardner: How does the training application demonstrate some of the more productive aspects of cloud?
Roggenkemper: The most interesting part about that is that you don’t need a vanilla system, but a system that is prepared for a particular class, which has the correct set of data. You need a system that can be reset to a controlled stage very quickly after the end of a training class, so that it’s ready for the next training class.
So there are two aspects to it. One is the reliable infrastructure on which the systems run, and second part is to get the correct system for that particular class ready in a short period of time.
Gardner:Are there unintended consequences or unintended benefits that come from this cloud model?
Roggenkemper: The thing that comes to my mind is that it allows us to take advantage of new computing infrastructure more quickly. We reduce the use of power, which is always a good thing.
Gardner: This idea of agility when producing these applications proves this concept of IT as a service. Do you see it that way?
Roggenkemper: Absolutely. And obviously, what we use internally benefits our customers as well. To have these systems available in a much shorter period of time for the customer’s development environment is as important for them as it is for us.
Gardner: And a question about future plans. It sounds as if this works for you. Then the virtual desktop infrastructure (VDI) approach of delivering entire client environments with apps, data, and full configuration would be a natural progression. Is that something that you're looking at or perhaps you're already doing?
Roggenkemper: Some things we're already doing, We have a hefty set of terminal services in our environment, as well, which people, especially if they are on the road or work from home, take full advantage of.
Gardner: David, I was very interested to hear you say that advances in pervasive virtualization and cloud methods are transforming how IT operates, giving you the ability of, as you said, saying "yes" when your business leaders come calling. What have you have been able to say "yes" to that exemplifies this shift in IT?
Giambruno: We've increased our project throughput over the past couple of years by 300. So my job is to say, "yes." I'm just here to help. I'm a service. Services are supposed to deliver. What this cloud ecosystem has delivered for us is our ability to say yes and get more done faster, better, cheaper.
The correlating effect of that is we have seen not only this massive increase in our ability to deliver projects for the business, because that’s really what business alignment is. I do what they want and I give them some counsel along the way.
The second piece is that we've seen a 70 percent reduction in the time it takes us to deliver applications, because we have all of these applications available to us in the task and development site which is part of our DR.
So this ability to move massive amounts of information where everything is just the file, bring that up and let our development teams at it, has added this whole speed, accuracy, and ability to deliver back to the business.
It’s probably easier to quantify it this way. We have 531 applications running on our internal cloud. Our internal cloud makes roughly 15,000 automated application moves a month. Our transaction rate is roughly 14,000 transactions a second. Our data change rate is between 17 and 30 terabytes a week. Over 90 percent of our corporate workload sits on our internal cloud, and it runs most of our footprint globally.
Gardner: We're talking about mission-critical apps here -- ERP, manufacturing, warehousing, business intelligence. Did you start with mission-critical apps or did you end up there? How did you progress?
Trust, but verify
Giambruno: I have a couple of "isms" that I live by. The first one is “Crawl, Walk, Run” and the second one is “Trust, but Verify.” When we started our journey roughly five years ago, we started with "Crawl" -- very much "Crawl" and “Trust – but Verify.” At Revlon, we didn’t spend any more to put this in. We changed how we spent our money.
We were going through a server refresh, and instead of buying all the servers, we only bought roughly 20 percent. With the balance of that money, we bought the VMware licenses. We started putting in our storage area network (SAN), and although core component pieces, and we took some of our low-hanging fruit file systems and started moving all that.
As we did that, we started sharing with the business. We showed them what we're doing and that it still worked. Then, we started the walk phase of putting applications on it. We actually ran north of six nines.
System availability went up. Performance went up. And after this "Crawl Walk Run," "Trust and Verify," it became "Just keep Going." We accelerated the whole process and we have these things that we call "fuzzies," things that we can do for the business that they weren't expecting. Every couple of months, we would start delivering new capabilities.
One of the big things that we did was that we internalized all our DR. We kept taking external money that we were spending and were able to give it back to the business and essentially invest in ourselves, because at Revlon I'm not going to be a profit center.
For Revlon, the more money R&D has to develop new products to get to our consumers and for marketing to tell that product story and get it out to our channels and use the media to talk about our glamorous products, that really drives growth in Revlon.
What we've done is focused on those things, taking the complexity out, but delivering capability to the business while either avoiding or saving money that that the business can now use to grow.
Gardner: You've been able to keep your costs at or below the previous levels. Do you credit that to virtualization, to cloud, to the entire modernization?
Giambruno: To me it’s the interaction of the entire ecosystem. It is a system. Virtualization is a huge part of that. That’s where all it started. As you look through the transition, it's really been interesting. I'm going to segue back to the saying yes pieces and what it’s allowed us to be.
We have this thing called Oneness. I always talk about being the Southwest [Airlines] of computing, and I live inside of very simple triangle. The triangle has three sides, obviously. One side is our application inventory, the other side is our infrastructure capabilities, and the other side is my skill-sets.
If you're inside that space I can say yes, very quickly. What’s happened inside that space helped us contain cost . When we first started work, our ratio was one physical to seven virtual. A couple years later, we're at 1:35. It’s roughly a 500 percent increase in capacity without any commensurate cost. I give credit to my team for owning the technology and for wielding the technology for the benefit of the business and to get the most out of it.
The frame of reference to keep ourselves grounded is that we make lipstick, and it’s really how much money we can save and how well we can wield that technology to deliver value and do more with less. That’ll enable our company to grow.
We love simplicity and we have this Southwest computing model of taking a very complex ecosystem and making it simple to use. To a large degree it's kind of like an iPad, where the business wants to touch it, but they don’t care what’s going on underneath.
It's our job to deliver that, to deliver that experience and capability back to the business, without them having to think about it. I just want them to ask that we’re here to help and that we can figure a way to deliver it and keep exercising our technical capabilities to wield the technology to do more.
Gardner: What are some of the upsides on the data when it comes to this ecosystem approach?
Giambruno: One of the things that we have is a big gestalt after our cloud was live. We literally had all of our data in one place.
One of the big challenges historically was that we had all these applications geographically dispersed. The ability to touch them, feel them, get access, access controls, all of these things were monumentally challenging. In Revlon, as we went to the Southwest or Oneness model, we organized globally our access controls and those little things.
So when we had all this data and all these applications now sitting at one place, with our ability to look at them and understand them, we started a fairly big effort for our master data model. We’re structuring our data on the way in So when we're trying to query the data, we already know where it is and what it does in its relationships, instead of trying to mine through unstructured data and make reasoning out of it. It’s been this big data structure.
I’d say we "chewed glass." We spent a couple of years chewing glass, structuring all this data, because the change rate is so big, but there's value in information to the business. I joke, if you've missed at this, we’re in the information age. So how well we can wield our information and give our leadership team information to act on is a differentiator. The ability to do this big data and this master data model has been really what we see as the golden egg going forward, the thing that can really make a difference with the business.
Gardner: How does disaster recovery (DR) play into this larger set of values?
Giambruno: We’ve actually done this. No one was hurt, but last year, our factory in Venezuela burned. It was on a Sunday afternoon and they had what we call a drib. If you look at VMware architecture, they have data center in a box. I always joke that we’re years ahead of them in that. We use dribs, strategically placed throughout the world where we push capacity to for our cloud. They largely run dark.
So our drib "phoned home" that it was getting hot. We were notified that the building was on fire. It took us an hour and 45 minutes, and most of that time was finding one of my global storage guys who was at the beach. We found Ben, and got him to do his part, which was to tell the cloud to move from Venezuela to our disaster site in New Jersey.
So we joke that our model in DR is that we just copy everything. We don’t even think about tiering or anything. It’s this model, sometimes a Casio is just better than a Rolex. Simplicity rules, and not thinking about it ensures that we have all the data available. Again, it goes back to our cloud and virtualization. Everything is just a file. We just copy the deltas all the time. We never stop.
For us it was available in less than 15 minutes. We went in, we broke the synchronization, we made sure everything was up-to-date, and we told our F5s and our info blocks that Venezuela is now New Jersey. Everything swung, we got everything in, we contacted the business units to test everything and verify everything.
Then we brought up all the virtual desktops and we used Riverbed mobile devices. We e-mailed their client to everyone. So people either worked from home or we had some very good partners that gave us some office space where people could use the computers. They loaded the Riverbed mobile devices on those computers. They brought the virtual desktops, people went to work, and the business didn’t go away.
This is a real-world example of how you can do it, and it wasn't a lot of effort. It's this whole idea of simplicity, where you're just not putting the complexity into the system. I always go back to this iPad view of the world, where the business just wants to know what's available and we will do the rest underneath.
This high degree of virtualization lets us move all of this data around the world, and it's for DR, development, and a myriad of capabilities that we keep finding new ways to use this capability.
Redundancy and expense
And some of the other unintended consequences are interesting. You talk about redundancy and expense. Two is one and one is none in a data center. Do you really need to be fully redundant, because if something happens we'll just switch to the other data center?
I only need one core switch or whatever. You start to challenge all these old precepts of up-time, because it's almost cheaper for me or less-expensive. I can just roll the computer over here for a little while. I get that fixed, if I have a four-hour service-level agreement (SLA) with my vendors for repairs.
You can start to question a lot of the “old ways of doing things” or what was the standard in figuring out new ways to operate. One of the interesting things I love about my job is you can question yourself and figure out what you can do next.
Gardner: Tell us about this extended business-process value that you're starting to explore.
Giambruno:One of the things we realized is that we could start extending our cloud. We spend a lot of time managing security and VPNs, and the audits that have to go around that.
If I could just push out a piece of my application or make that available to them, they could update their data, reduce the number of APIs, the number of connections, all of that complexity that goes out there, and extend our MDM.
Then we can interface our MDM through our cloud to do some of this translation for us that they can enter data, or we can take it from their systems, from our cloud edge securely and in context and bring that back into our systems.
We think there are huge possibilities around automating and simplifying. But at the end of the day, it's about collaboration with our community of vendors and suppliers, and enabling them to interact with us easily.
So you're always trying to foster those relationships and get whatever synergies you can. If we make it easier on them to interact with us from a system’s perspective, it just makes everybody happier. We've got some projects slated for deployment this year. Maybe in a year, if you come back, I can tell you how well we’ve done or what we’ve done. But one of the things that we are looking is we can think really change how we operate as a company.
- Case Study: Strategic Approach to Disaster Recovery and Data Lifecycle Management Pays Off for Australia's SAI Global
- Virtualization Simplifies Disaster Recovery for Insurance Broker Myron Steves While Delivering Efficiency and Agility Gains Too
- SAP Runs VMware to Provision Virtual Machines to Support Complex Training Courses
- Case Study: How SEGA Europe Uses VMware to Standardize Cloud Environment for Globally Distributed Game Development
- Germany's Largest Travel Agency Starts a Virtual Journey to Get Branch Office IT Under Control