As if managing your current IT operations wasn't challenging enough, along comes cloud computing to plaster on another layer of complexity. How does your current management strategy need to evolve to support private cloud? How do you seamlessly control a hybrid private-public cloud? No one has all the answers, but BMC Software -- a champion of the concept of business service management -- is moving aggressively toward making cloud just another service option to be managed, like networking, storage and applications.
You're managing and building 150 cloud-based data centres for clients. Are these mostly private clouds?
Behnia: We're actually quite active on the public clouds that some of the service providers are developing. But the majority of them are private clouds.
Beauchamp: We've done some work with Amazon, for instance, around enterprise service-request management. So if someone requests a service, our engine can determine whether or not the most economic place to provision is in EC2 or Google or elsewhere. And then if they say that's where we want to go, we can go provision it and do the chargeback. It goes through their procurement processes, so you've got auditability and control.
How do you define the cloud? There are so many different definitions out there.
Behnia: We actually just use the "new" NIST National Institute of Standards and Technology definition because we find that to be the most pragmatic and most vendor-neutral. It's fundamentally that the cloud focuses on delivering services. I think this sometimes gets lost in a lot of the discussion around cloud computing. Everybody's talking about infrastructure and hypervisors and virtualisation, all of the components. At the end of the day, what customers really care about is getting secure, reliable, trusted services, whether that's from their internal IT department or from the external broker to their IT department, or from an external provider directly.
How do you differentiate cloud from virtualisation?
Behnia: There are three critical elements that we're seeing across the various areas of cloud computing, whether that's infrastructure as a service, platform as a service or software as a service. First of all, the majority of the environments are virtualised. Most enterprises are not 100 per cent virtualised. Many of the workloads and applications operate in non-virtual environments. That's very important from a management standpoint. The second characteristic is that they're highly automated, because you can argue that without automation you don't have a cloud. Amazon doesn't have hundreds of sys admins provisioning these servers. The third piece is that they're service-oriented - in other words, typically there is a set of service offerings that are defined. You can think of it as a catalogue or a menu of items where you can define and differentiate different tiers of service - gold, silver or bronze - and bring forward things such as pricing.
Do customers want to manage their cloud environments with the same tools they use to manage their non-cloud environments?
Beauchamp: We've heard very loud and clear that they do not want to buy a new set of widgets to manage a new set of widgets. Customers have built workflows that they've adopted through the years. They need the tools to plug into those workflows. The fact that it's a cloud is really irrelevant. The customers just need to request a service, have the service provision, have some engine determine where is the least expensive and best-suited platform for the request, and then maintain the service level around the request with transparency and compliance. The fact it's on a cloud needs to be abstracted out of the discussion. That is an infrastructure discussion around a cost model and a delivery model, not an end-user decision.
So can you talk about the technology behind the scenes?
Behnia: The front end of this -- analogous to a restaurant or an online book store -- is a portal that would define a set of available offerings. Users can request the standard Linux desktop, regardless of whether that's physical or virtual, whether it's VMware or Zen, whether it's hosted in Amazon, or whether it's on premise. The second portion is really around orchestration and placement, so that you can actually fulfil those requests and automate that. And this is where we have the broadest and deepest capability, being able to automate provisioning of not just servers, but also network configurations, applications and storage associated with that service.
What about the integration in this cloud environment - if an IT shop is using your tools, but the cloud provider is not, or vice versa? How will you handle that integration to create a seamless management experience?
Behnia: A lot of customers want to set up an internal project cloud, and for a certain set of workloads be able to place those workloads into an Amazon environment. We've built an abstracted data model so that whether that server is placed internally or externally, from a configuration standpoint it is absolutely identical. The approval cycles are absolutely identical. Except when it comes to run time, we actually invoke the correct set of APIs [application programming interfaces] to be able to leverage Amazon's APIs. So our value proposition is rather than tying all the management to each of the different technology silos, we've abstracted that out so enterprises can build their policies and workflows around the processes as opposed to around the technologies.
Do customers need to change those workflows?
Behnia: I think the answer depends. Much of the work that was done around best practices frameworks and compliance doesn't change. It's technology-independent. But we've seen at least a couple of areas where, to leverage the power of cloud and virtualisation, those processes can be optimised. One of the critical areas is around auto-provisioning and configuration management. The whole premise of the cloud is that there's going to be so many changes to the environment that you need to have an up-to-date and scalable CMDB [configuration management database] and configuration management model that doesn't rely on human input. This entire lifecycle can be done in an automated fashion.
Also, the cloud forces some issues around what is a service, what is my class of offerings, how do I optimise my processes around that? Certainly, most enterprises see automation as a clear pathway towards cloud. Many of our clients start by automating server provisioning and application deployment, then move into adding the service catalogue.
Automation tools (not specific to cloud)
There was a recent Forester study that talked about the adoption of automation tools within enterprises of all sizes, and the survey said only about half of the respondents had implemented an IT automation tool of some sort. How does that map to what you're seeing?
Beauchamp: The ones that are adopting automation have a tendency to be adopting the tool for a specific purpose. Once they've adopted it for a specific purpose, they realise what they've got. Our salespeople are not going to try to evangelise how you can completely redesign the way you do IT because it's too long a sales cycle. So they say, if what you're focused on Rackspace [Hosting], as an example, is to be able to quickly provision and bring up a new server for your online subscribers, we'll do that for you. After they've put it that in they say, why couldn't I use that for re mediation? So then they begin to spread out. It continues to evolve and move up.
There is nothing growing faster in our pipeline than automation. In this last quarter, our licensed bookings for non-mainframe products was 40% year-over-year growth, and the strongest piece within that is automation. And the pipeline's growing faster than that.
Behnia: One of the things you can measure is the time from when somebody in the business requests a new service and when that service is provisioned. We do this as a process audit with enterprises of all sizes. And it's a big eye opener, because if you were to look at this from the business standpoint, that reflects the agility of the IT department in being able to respond to increased demand.
From a revenue perspective, will automation subsume your traditional systems management products at some point?
Beauchamp: No, it won't. It's certainly a faster-growing market than traditional red-light, green-light systems management, which has also migrated into something different now. It's really much more predictive and real-time optimisation, seeing it and being able to reconfigure. Now when our traditional systems management product sees an issue, it can call over to its friend in the automation components to reconfigure the system in order to address what it sees coming, say a capacity-related issue it sees on the horizon.
So we're finally seeing real, live autonomics - systems that can heal themselves based on policy? The industry has been talking about those for 20 years.
Beauchamp: Yeah, we've been talking about this forever. The code actually works now. It's starting to really happen. I think traditional red and green light is not going away, but it's mutated into something that's more modern. But even with that migration, it's not as fast a growth as automation is right now.
Are customers finding new ways to put these tools to work? Are they discovering ways to bring in revenue or at least achieve better ROI?
Behnia: As I said before, one of the leading metrics around how agile the organisation is [is] how quickly it can respond to new capacity requests or requests for new services. On average we're seeing anywhere between 40 days in best-case scenarios, well-run organisations, to as high as 110 days in some of the largest billion-dollar-IT-budget companies. We can reduce that down to literally a day. I was with a CIO that said how this is almost like FedEx. They want to have next-day delivery, or second day, or fourth day, depending on what the service is. So what that shrinkage in time means is that that's more time for developers to write new software, and marketing programs can get run faster. Not only does it impact the bottom line in terms of efficiency, but it directly affects the top line as well.
Are they telling you about ways they're making money?
Beauchamp: Sure. We have a lot of customers in the insurance industry. A year or two ago they all kind of simultaneously panicked about their infrastructure, and they came to the realisation they had to -- I'll use the old term -- Webify their offerings. What they were really talking about was revamping their distribution models. We've got to get the customers to go online, stop talking to people, stop picking up the phone ... just do this all on the Web as much as possible. So they went into this rush to bring all their front-office applications online and to make them really simple. One customer told us that right before they standardised on our software, they had an entire state that was down for two days and didn't know it until the reports came in and they saw they had zero revenue come in from the state.
Now they want to go the next step. They want to be able to deploy new applications very rapidly, push out new systems and new revenue-generating applications. The revenue drivers that they're forecasting have to do with new content going out. Content's got to be technology based, and we've got to do it in a cloud environment, we've got to go quickly. If they don't get it out quickly, or if they're price-inefficient, they're not going to be competitive. They all want the self-service model.
I'll give you another example. One of the world's largest software companies called us up. They traditionally have sold their software with the white-shirt, red-tie, blue-suit kind of approach where somebody walks in the door and you buy the big blob of software. They have become convinced that the road ahead for them is one where customers will go to a website, they will specify, "I need these application modules," and here are different pricing models. This one is number of seats. This one is all you can eat. This one is nine to five, Monday through Friday. Customers basically want to arbitrage price and have the ability to decompose their pricing. Traditionally companies have fought like crazy to not allow that. Now they're basically saying, "We give up." It's going to ultimately become disaggregated. Customers are going to be able to see our pricing models. But then in order for us to not go broke, we have to know what our cost model is for delivering that.
You're talking about retail.
Beauchamp: That's right. Enterprise is going retail. In order to do that they need cloud-based solutions, self-service, customer-requested. They can't have any people involved in this. It's got to be a no-touch environment. This is what I think is fascinating about where we play. When I joined BMC in '88 the job was: Somebody bought something, it was working fine and then it broke one day so they called up somebody else like us to sell them a tool so that wouldn't happen again. Like high-speed backup or red light, green light [management]. Now what they're saying is, cloud is as much about automation and about self-provisioning and service-level management, compliance and transparency as it is about the infrastructure itself. Management actually becomes the delivery mechanism for the application.
Software as a service
Where are you playing on the SaaS front?
Beauchamp: We give our customers the ability to deploy clouds so that they can be providers. We also work with the service providers -- and by the way outsourcers and systems integrators who want to set up multitenancy environments for delivery systems -- to configure, deploy, and charge and manage cloud-based service platforms.
In the last quarter we announced two major SaaS offerings -- our Remedy, which is, you know, the world's largest IT service-management system, as a service on demand. It's hosted. It's delivered. It's secured. You don't have to go in and set it up in your centers, and you can run it yourself and deploy very rapidly. Also, Salesforce.com announced Dream Force a few months ago. So they now have a service desk on their platform called Force.com that their salespeople actually get paid to sell. And this is not their typical partner deal. This is a special relationship between our two companies.
The other thing we do with SaaS is we sell software to Salesforce.com, Concur, Edmonds and others who themselves deliver SaaS. They need management just like everybody else needs management.
Behnia: We also can help manage SaaS applications for the enterprise. If you want to know how many licenses you have on your SaaS-based applications, our asset management system has been extended to be able to account for those licenses. If you want to do response time monitoring, you can do that.
The changing role of IT
Behnia: This is a thread that really ties both the cloud piece and SaaS nicely together: We see the role of IT as evolving and changing from being the sole provider of services to the aggregator, broker and the integrator of these services, whether they're internal or external services - because when there's a problem with Salesforce.com, you don't call Salesforce, you call your internal IT help desk.
Beauchamp: In a sense you're now talking about things like core network services, things that people in that business have always talked about. But now you're now starting to talk about it for the data center. IT becomes this aggregation manager for wherever this stuff sits. And it may very well be there's nothing in your data center. Your data center goes away, even for larger companies. [Instead] it's going to be popping all over the place.
Behnia: We see multisourcing as where people take an individual service and then ask who's the best provider at this price point and efficiency?
Talk about the evolving competitive arena today - you're dealing with some pretty powerful companies that didn't start as management companies.
Beauchamp: I went to see Jim Barksdale and Marc Andreessen when Netscape was a little company; they were the hottest thing on the planet. I said we ought to talk about enterprise management. At some point one of them said, the whole management layer's going away. There'll be no need for a management layer because the Web is going to make that irrelevant. We're going to have our management products based on Web technologies, and we're going to build a complete management stack. Same thing with Oracle and Microsoft -- I met with Larry Ellison and Jim Allchin and they said the same thing. Everyone says, 'We're going to build a complete management stack.' And every time the same thing happens. The customer starts saying: Why are you the logical person I'm buying this from? Am I going to trust Oracle to manage Sybase? So it begins to get wobbly. The other thing that happens is in the vendor's board room when they have a bad quarter and money's tight. They're saying, how much money are we spending on management and why? It's outside of our core. So they ultimately bail out.
Now HP started early and they're in it. IBM with the Tivoli deal got in it. And then there's us. And you can argue that CA certainly has a lot of stuff, right?
CA is primarily a mainframe company still, as best as we can tell. They've done some acquisitions of some nice products. And they talk BSM, and they talk cloud computing. I don't take them lightly, and we compete with them every day.
What's the defining philosophical difference between your company, HP, IBM and CA?
Beauchamp: The management business, in the case of HP, is I think 2.5% of revenues. It's an interesting business. But at some point their board of directors has to talk about the other 97.5% of the revenue. By the way, [for customers] HP says you really need to use our servers, because the only way you're going to get the full robust capability out of this is going to be on our platform and our storage. There's just an inherent conflict of interest around buying the agnostic management layer from the people that are also selling you the entire soup-to-nuts platform. I would say that in general customers would view that as lock-in.
You're seeing basically a reverticalisation of the stacks, like back in the BUNCH days. [Burroughs, Univac, NCR, CDC and Honeywell were mainframe rivals to IBM collectively known as the 'BUNCH.'] You're seeing companies do what they did back then: We've got our own processors, we've got our own OSes, we've got our own services, we've got our own hardware, our own storage, our own app layers, our own management layer. You can buy it all from us. And look how happy you're going to be. And then Oracle came along and said, I've got an idea, why don't we do a horizontal database that runs across all of them, disaggregate what was pricing leverage? And Oracle just went from nothing to gigantic, basically busting those stovepipes.
So then everybody else followed suit. I'm going to be the operating-system company, I'm going to be the database company, I'm going to be the desktop company. I think the same gravity that caused [vertical integration] to fail in [the BUNCH] time will cause it to fail again. The integration is not meaningful. And it gives the vendor enormous leverage.
Behnia: And it's not practical given how much customers have already invested. These aren't green field, brand-new data centers where the customer's going to buy 20,000 brand-new servers and forget about all their 105,000 applications that have been written.
Beauchamp: CA built an incredible franchise back in Charles's [Wang] day just buying up a whole bunch of companies and basically taking that whole space, and saying, if you want one of those, we've got a bunch of them, right? There's nothing inherently wrong with that. It's just a different strategy.
Tivoli almost got it right back in client server. It's just -- the technology didn't really work, and it was client server based ... and some other things. But it was a pretty doggone good story. When that Tivoli framework came out, we looked at buying it. We went into discussions to acquire Tivoli before IBM did and I was involved in those discussions. I told my boss, if they can do what they say they can do, they're going to put us out of business.
So when I got the keys to the car in 2001, I basically said, let's do this again, but let's do it right this time. Let's not make it a big, heavy framework, and let's break it down into bite-sized pieces. Let the integration occur naturally. When you buy the second piece, it should find the first piece and work together. Not that you can't upgrade this one unless you upgrade this one. We basically just redid the framework thing, modernised it, because it's got to work that way. The world cannot accept hundreds of point products that just manage one thing. It's illogical.
Each new computing paradigm sets up a new set of gorillas. So how big a risk is it that cloud is so different that it's going to open the door to a vendor who approaches management completely differently?
Beauchamp: There's a moment of fear that we experience every time something disruptive and new emerges. Once that wave passes, the next fear for us was: We know cloud's going to be big. We know customers are going to make architectural decisions. We have to win those bake-offs. So we set up a task force across the organisation, and we just focus on winning that architectural [bake-off]. So we've got product initiatives. We've got distribution initiatives. We've got marketing initiatives. You know there's support, how we go to market. We're starting to win these deals. Now the next fear is ... is something entirely new coming along, and management just becomes irrelevant? And a customer told me why he didn't think that would be true. He said: Eventually something is plugged into the wall, right? It's not really in the cloud. It actually is sitting over there and it's running. And so I just don't see the uber-clouds annihilating all the management layers. Private cloud to us is just another operating system inside our existing customer base.
Business service management
How is BSM, a field you're widely credited for inventing, different from traditional enterprise systems management?
Beauchamp: BSM is the general envelope of how we describe what I've been talking about, which is this self-service provisioning, managed service level, transparency and compliance. And ESM is just a name we made up for our business unit that is the non-mainframe business unit.
Here's an example of BSM. Last year a very large seed manufacturer was remaking their IT organisation. I met with the CIO right before they standardised on us, and she was explaining to me the rationale. She basically said, look, we're going from -- and I won't get this right -- two mutations per genome or whatever it is in the seed, to an eight-pack so you can plant wheat in countries you couldn't before. And, she said, in order to do that, they have environments where their scientists need to run simulations that are unbelievably computationally expensive to do. Sometimes those are just sandbox activities. So in that case they want to go outside. They use a university's grid environment to do a slow computational; it gets it done and just uses whatever resources are available at the time, kind of a grid model. And then when they go to run their big test and they really have to certify it for the government, they want to run it inside on their very scarce resource -- a very expensive, computational platform. And what they want is people to just request a service, and they want the engine to decide, am I going to jump over here to the university grid, am I going to go run it on the inside physical environment that already exists? Do I need to pick up a new VM on another machine to run it inside that environment because it needs to be segregated for compliance reasons? By the way, that's getting really complicated. They want our software to be the common portal for the 'store' where you go to request it, the common engine that decides where to host it, provision it on those environments, maintain the service level, and then report on it.
This is our value proposition around business service management, how we deliver it. And cloud is one module inside of that.
You've used the term 'ERP for IT.' Where are you on the journey of delivering that?
Beauchamp: Well we are a little cautious about using that term too much. Some customers just love it, but some people will equate it to the things that didn't go well in the ERP generation. You know, you're going to do it our way, and it's rigid, and it's going to take three years, but it's really going to take three-and-a-half, and it's going to cost a hundred million, but it's really going to cost four hundred million. So we don't use that term externally, really. But internally we'll refer to it that way sometimes, because what we're really talking about is full lifecycle management for services, where if it works right it turns into a management link, a hub and everything that plugs into it gets commoditised. If you can get an Ethernet cable into the thing, we can discover it, and we don't care what it is, we can manage it. You could swap out one vendor and plug in another. So you really begin to disaggregate the stacks of IT, which is the exact opposite of what HP and IBM want to do. They basically need those stacks.
Behnia: I think where we are on that journey really depends on the maturity of the individual enterprises. Different enterprises are at different maturity steps. One of the things that we did in the design of our architecture is to make sure that customers could deploy this as they go. So that we're not overshooting their capabilities or their needs. And this is one of the big differentiators for us, and where the analogy for ERP breaks. Because many of the ERP solutions were very monolithic.
Beauchamp: If you hire us to do it, we can get you a long way to it right now. It's not as cookie-cutter as it needs to be. But we're doing it for customers now on a regular basis. Every quarter we're taking the rough edges off -- where this product and this product connect. It's just in the productisation of the whole thing. We still have to go in with experts and help you configure it and get it set up and interview you. And it's still got a services element that's a little rich.
Acquisitions and partnerships
You've described the BladeLogic deal as a 'transitional' event for the company. Did it really turn out that way?
Beauchamp: Absolutely. When you take Blade and you add it to our network configuration automation product, and you add it to our run book automation product, you have the core of the automation suite. One of our competitors says they're all about cloud, but they have no automation tool. They've got all sorts of management widgets, but you can't provision a cloud. Without BladeLogic, we would not have been able to provision clouds. We would not have had the relationship with Cisco. We would not have been involved in as strategic a relationship as we have with Dell. We wouldn't be as involved as we are now with Accenture and Capgemini and CSC and Wipro and most of the global outsourcers and systems integrators.
How do these partnerships go beyond the usual marketing slogans - are they having a material impact on the bottom line?
Beauchamp: It depends which ones. We won an order last December, one of the biggest orders in our history. Accenture was very involved in that deal. They are actually doing the implementation of our software. The two of us worked together on that deal. Accenture views us very importantly, as their management partner in many cases. You can't say it was just the direct sell through that was material. It was the influence and the relationship. You can tell a similar story around CSC and about many of the Asian outsourcers and systems integrators.
What can you tell us about the Cisco deal? What are you seeing through your work with Cisco? Are they successfully moving the Unified Computing System (UCS) product into organisations?
Beauchamp: I think they said somewhere around 900 UCS boxes had shipped. That's their number, not mine. We'll just call it north of 100 in terms of how many that [our products] were sold with. That number is not bad in my opinion considering the fact that such a high percentage of the UCS boxes are proof of concepts. Lots of companies are doing pilots. We will be a lagging indicator. As they start to go from proof of concepts into heavy production, that's where we're going to play. Until you see a giant chemical company throw out one vendor and go with a huge UCS, we won't get huge orders out of that. So that's part one. But part two with Cisco is that because we're doing this relationship with them, our whole management stack is interesting to them as it relates to cloud initiatives and partnering and where they're engaged with the customer. So we partner with them in the field level, on a daily basis with them on very important customer initiatives where they're working on large deals or we're working on large deals.
By the way, we do the same thing with Dell. We work with Dell on deals. They're focused a little more on midmarket. But we're working with both of them.
You've done a lot of acquisitions over the years. And we talked to (Cisco's) Cham