Sunday, September 10, 2006

Full Transcript of Dana Gardner's BriefingsDirect Podcast on Network Services and SOA Convergence with Cisco's Bill Ruh

Transcript of BriefingsDirect[TM] podcast with Dana Gardner, recorded Aug. 23, 2006.

Listen to the podcast here.

Dana Gardner:
Hi, this is Dana Gardner, principal analyst at Interarbor Solutions, and you’re listening to BriefingsDirect. Today, a podcast discussion on a mega convergence point, the intersection of advanced intelligent network services and applications as services, otherwise known as Services Oriented Architecture (SOA). And joining us to discuss this rather weighty topic is Bill Ruh, the vice president of Advanced Services at Cisco Systems. Welcome to the show, Bill.

Bill Ruh:
Dana, thank you very much.

Gardner:
Let me provide some more background on you and what you do at Cisco. You've been 20 years plus working in enterprise middleware and integration, and you are currently leading the Services Oriented Network Architecture (SONA), as well as the Application Oriented Networking (AON) services teams worldwide at Cisco. Previous to that, you were a senior vice president and CTO at Software AG, and you’ve worked at Mitre and IBM.

Furthermore, you’re an author, and in July of 2004 published a book called, "Enterprise Integration: the Essential Guide to Integrations and Solutions." Previous to that came, "Enterprise Application Integration, " as well as "Inside CORBA" and previous to that, "IIOP Complete. " So it sounds like you’ve got a pretty good handle on integration, Bill.

Ruh:
I’ve seen a lot of the trends and a lot of the technologies come and go over the last 25 years to help us solve this problem.

Gardner:
Based on some of the presentations I’ve seen you deliver, we are at this mega-tipping point -- not something that is just a continuum of what’s happened in the past, but something that seems quite different. This is largely due to SOA, virtualization, network convergence, intelligent network services, a higher level of standardization and open source. This all seems to imply a bit more than the typical roadmap, something that is quite new. I think you called it "integration for everyone." I wonder if you could start our discussion by sort of explaining what you mean by "integration for everyone."

Ruh: It certainly does feel like we’re about ready to enter and move up to another plateau in technology. In my experience it’s like client-server, or certainly like the Internet age. Then, we went up to a new plateau with the technology, and this has that same feel. I call it "integration for everyone" in that previously when we have talked about integration, it has been very much technical, the very technical integration of tying together systems.

But, when we start to look at where we’re going now -- in both the number of things and the kinds of things -- what we’re trying to do with those things in tying them together now is very different. In fact, we’ve got a movement from where we used to just have a few devices to where we have the Internet. We’re tying together RFID tags. We’re tying together sensors. So we’ve got a real change in the kinds of things that we’re tying together, where it’s not just technology things. It’s things that are going to be more business-oriented.

The idea of sensors, the idea of tags, the idea of "what is the device on the network" -- that is changing. We also see that the kind of things we want to do with integration is changing. It used to be that integration was really about taking information from one system and integrating it with information from another system. The kinds of integration we’re talking about now is also fundamentally changing, because what people want to do now is more about interactions than about production and transactions.

Now that I know that a temperature gauge in a refrigeration unit has gone off, well, what do I want to do with that? I want to not only sense it, but I want to respond to it. I want to tell somebody about it. This idea of collaboration means that I’m not just integrating applications in a technology sense. I’m integrating people and processes into this. So, when you look about "integration for everyone,” what we’re really talking about is not just technology integration. We’re now integrating all these devices and all these sensors and all these people and all these cell phones.

We’re now taking those integrations and its not just transferring information. We want to collaborate. We want to bring in this idea of communication. When I talk about "integration for everyone," it means that it’s integration for me. It’s integration for my son or daughter. It’s integration for everyone so that they get all the services and capabilities they need through whatever device or channel or whatever they’re interacting through. And that’s a fundamental re-definition of what integration is about.

Gardner:
Okay, it sounds like this portends a great deal of opportunity, albeit with quite a bit of complexity. So, if I’m a CIO in a corporation and I’m trying to be strategic, work with the business side of the house in formulating how we’re going to take advantage of these trends and this new plateau of integration, how do we manage the complexity? What are some of the priorities that you see facing today’s CIO?

Ruh:
I think you hit on the major challenge here. Software is always complex, and systems are complex, so it’s great for me to say, "Wow, we’re going to take all these sensors and plug them in, and you’re going to be told when some sensor goes off in your house or some sensor goes off in your business, and your business processes are going to be suddenly kicked off and you’re going to respond to things much quicker." But the fact is that we all know that trying to do that in real time is difficult.

The CIO is the one who is going to bear the brunt of why we can’t do this. The last time I remember, "Why can’t I do this," was in the PC era. You remember everybody had dumb terminals, and suddenly it was easy for people to go buy PCs through the finance department, etc. The CIOs suddenly had all this complexity and they had to tie it all together. Suddenly, they had networks showing up everywhere, all different flavors of networks, all different kinds of protocols -- and that became a nightmare to manage. Cisco was actually born out of that era in solving the problem of how to integrate and network together all these different PCs into different applications.

We’ve moved up that stack and the CIO is going to see that people are going to be building sensors and cell phones, and people are going to be saying in their businesses, "Why can’t you do this? I can go buy this capability." This will happen in the same way we saw the PC coming in the back door 20 years ago. Fortunately -- and maybe it’s really driven by this complexity -- the dialogue has turned to SOA.

SOA isn't really new. I was involved in Smalltalk and CORBA and the object-oriented movement back in the mid-1980s. A lot of these techniques and approaches have been talked about for a long time. What makes it different is that we’re now starting to get where the protocols are being made that allow standardization to occur. This idea of protocols is the same thing we saw with Internet protocol (IP). What we’re seeing is a movement up the stack to things that ride on top of IP that allow services to actually communicate, coordinate, and collaborate better -- in the same way it used to be hard to connect two things together and send data from one device to another, which is less of a problem today.

We’re going to see that protocols and services in the network are really going to allow us to communicate easier and reduce some of the complexity. So, the network’s going to tie into this movement of SOA. It’s going to provide services, infrastructure-level services, to take some of the complexity away from CIOs in terms of the infrastructure and allow them to focus on the business processes.

Gardner: We’ve discussed some of the technical integration issues, but it seems to me that in a typical enterprise today, there also is going to be the need for a cultural or organizational integration. For example, the application development and deployment folks are in one group or one culture. Network services and administration in another, and then voice telephony (VOIP) and communications in another. It seems to me that that won’t stand, that disintermediation between these different groups is not conducive to this larger integration proposition that we’re discussing. How do we bring application and network services together? How do Cisco’s AON, the Application Oriented Networking, and SOA relate, both technically and organizationally?

Ruh:
You know, I think you hit the nail on the head. I don’t think this is necessarily being driven by Cisco. This is being driven by this larger services-oriented trend. When you look at it, we’ve organized our IT and our businesses that integrate with IT around the Internet model and the client-server model.

When you move to a service-oriented model and you start to move more intelligence in the network, certainly you’ve got the need for the networking person to become more cognizant of how they play and support the applications -- and it’s no longer a black box. We see the need for the networking folks to take on a larger responsibility in that intelligence, and provide services that may have been more traditionally built into our applications -- duplicated in many instances -- so people have multiple identity services, instead of one that logically probably fits in the network.

What we see is that the networking folks have to come up and deal with some of the complexity of the services being put in the network. At the same time, the applications folks within an IT organization have to think of the network not as plumbing. They can't just say: "I’ve opened up a socket," or "I open up some communication mechanism and the network takes care of getting the bits from here to there." They have to realize that there are services inside the network. We’re going to see that the organizational dynamics that you suggest are going to change because of that.

I’d add one other thing. As we get to more services or other organizational dynamics, we’re going to see that the end users are going to want the tools that they have on their desktops to allow them to quickly and easily utilize those services in the network as well as on our servers, so that they can apply them and connect them into their desktop applications. We see this in the case of Google, which has done mash ups and other things out there.

It’s not just the networking administrator or the networking person. It’s not just the applications person. The end user is going to start taking advantage of services and that changes the dynamics entirely. So, we’re going to see a real change in the dynamics between these organizations. And that’s why the network is so important. It’s the one thing that connects them all -- and those services have to be there to reduce that complexity.

Gardner:
Another trend that’s afoot today that we haven’t mentioned, but has an impact, is the social networking phenomena, characterized often as Web 2.0, where we're seeing communities and collaboration and particularly interactivity. This back-and-forth nature of a discussion, of sharing of knowledge, is growing prominent, not only in blogging, but more mainstream in enterprises and in how people do business. Given that interactivity is a buzzword, and I don’t think just a short-term one, how does the network play in a modern, advanced sense in supporting interactivity solutions?

Ruh: Within Cisco, we talk about unified communications -- all the different mechanisms that people use to communicate and collaborate, and that communication is certainly changing. You just outlined that. We go from Voice-over-IP (VOIP), to instant messaging, to collaborative applications, sharing documents, and then we get into blogging.

This whole idea of communication is fundamentally being re-written overnight. Unified communications is going to be the focus for these next-generation applications that are going to support that environment. When you really look at the vision for where unified communications will go in the next five years, it includes the idea: "Why can't I take my blogs, and why can’t I take something like Google Earth and then plot where the blogs are?" Maybe I want to look at blogs that are from a certain part of the world, or that talk about a certain part of the world.

Communication is fundamentally changing to be IP-based. So, the network has provided services to allow those individuals to integrate with everybody. If I’m on a blog and I want to talk to someone and I want to collaborate with someone, I can bring up my collaboration tools. I can bring up my VOIP capability -- all of these things that are integrated. Those services allow me to take all of the services like Google Earth and start to use those tools as a part of the collaboration.

The idea that services can get integrated to the communications is important, and the network becomes the focal point for the services that support that infrastructure, as well as IP being the basis for all these different tools working together.

Gardner:
It sounds like what we’re gaining here is the opportunity to not just view communications, applications, and data as separate, but that we need to pull together with our own brains -- with our wet-ware. How do we take this intelligent information network and manage it? It sounds like we have this opportunity, but I think people are still stuck thinking about these things as separate and unrelated.

Ruh:
That’s obviously going to take some time, in the same way the Internet took time. I remember in 1994 and 1995 putting up Web sites. At the time I was with the Mitre Corp., and we did a lot of early work there. A lot of folks looked at this and asked, "What use was it?" and, "Who is going to manage it?"

Between 1994 or 1995 and 2000, organizations suddenly had a change, and now they have Web administrators and Web servers and everything else coming in. The organization had to morph to support that, so there’s always this chicken-and-egg relationship. What do you do first?

The important thing is that you don't decide to put the organization in place and then the technology. What happens is that the business begins to be driven by some of these external forces. So, external forces -- sensors, unified communications, and moving to VOIP -- and all these IP services -- are starting to come together, and they’re coming into the organization. With the PC, the user said, “Well, I’ve got this spreadsheet, and this is how I want to interact with things.” People start to focus on what the application is.

"If I can go home and do all the things I’m doing at home, well, when I go to work, I want to have exactly the same thing." That’s the thinking that drove the PC. That’s what drove the Internet, and that’s what’s going to drive this whole era. People can do this at home, and they’re going to say, "Well, why can’t I do that at work, and why can’t I have that same capability?" And they start bringing it in the back door. What happens then, of course, is that the IT world has to respond to that.

What we’ll see is that organizations are going to have to decide whether they want to get a little ahead of the power curve here. They know this is all happening, and that means they have to look at their network, look at their systems and say, "I’m going to do services-oriented here. I’m going to do SOA, and I’m going to do services-oriented infrastructure. Let me figure out a few of the core elements I need to put in. Security is a network service. That’s great, and I’ve got to put that in there. I’ve got to start looking at what kind of services go where."

You get a few core elements in and from that what is going to happen is that IT organizations are going to learn what they need to do to really fit into this wave or this trend. Number two, is that they are going to be viewed as being responsive to the business side of things. This is the approach that people have to look at. "What are some of these core services that I can put in the network? What are some of the core services that I’ve got within my servers and how can take advantage of this trend and support what is certainly going to be a wave that is going to hit?"

Gardner:
We’ve had quite a few acronyms in our discussion, and we could probably come up with a couple of hundred more. Help me out with Services Oriented Network Architecture (SONA). I’ve got a pretty good handle on SOA, and I’ve got a pretty good handle on AON, but I’m a little bit less clear on SONA. What do you all mean by that?

Ruh: SONA is a Cisco-defined architecture. We recognize that we have a very diverse product set, and those products fit together. So part of SONA is to show how all of our products that we bring to market fit together into an overall network. And rather than the network as just the transport, it's the network as a platform.

With that you can understand how unified communications fits with mobility, data center, networking, and security. How does all this fit on top of your network infrastructure and work together? One piece of this is to show how the entirety of what we’re doing in terms of our products fits together as a single, integrated platform.

In addition, what SONA is about is for us to show how these are not only just products with features and functions, but that they provide specific services in the network -- identity services, VPN services, presence services from mobility -- so that the applications people can begin to understand it. If they're going to use a presence service within the SONA architecture, when the networking people provide those mobility services, and there is a presence service in there, they can take advantage of that presence service for their application.

SONA is about how the entire Cisco product line fits together. It’s also about how those products then provide services and how those services fit with the application. Finally, it really lays out the vision for how we’re taking our portfolio going forward and how these things are going to continue to evolve, how they’re going to continue to work together. As an architecture, we show how we can create an infrastructure that can improve your overall SOA investment.

As you invest in SOA, the infrastructure through SONA is there to allow you to scale. How are you going to do bandwidth and manage bandwidth and quality of service? How are you going to provide a converged network? How are security and mobility going to fit in? Which standards and protocols are going to be used? How do you get the ability to virtualize your storage and your compute environments and other devices? SONA really is an architecture, and it’s a way for the customer to see how the network plays in their overall service-oriented architecture and what services we believe really belong in the network going forward.

Gardner:
Help me understand, if I put myself in the role of the developer. Traditionally as a developer I probably haven’t been too concerned with the network and I probably feel that’s up to the operations people to deal with. I’m focused on the logic, on the presentation, on the access and integration. But it sounds to me as a developer that I may have a palette of different objects and services to choose from as I’m crafting my application. And increasingly I’m going to have more on that palette in the form of network services.

Is that the way I should be thinking as a developer -- that I should be looking toward network services as part of my typical library of resources when I’m putting together an application? Or is there some other intercept where these network services as you described them will come into play with applications?

Ruh: We don’t want developers to look and say, "Oh, I’m programming the network." Obviously that’s not the intent here. In fact, to them, these services should just exist and be available to them as part of their overall palette, as you described.

The issue that we’re trying to educate people on is that some of these services do belong in the network. From a developer’s perspective today they may think about identity services or mobility services. They need to be aware that decisions need to be made about it. Maybe this ought to be in the network, and they need to work with networking folks and architecture folks. If we need these as common services, hopefully they’ll also say, "Yeah, we know that they should be in the network."

And this has happened before. VPN used to be in application code -- and it has migrated down to the network. Multi-cast is another great example: Programmers used to build multi-cast and now they know it’s in the network, and they expect it to be there. Firewalls used to be at the application layer. It's the same with quality of service, replication, back-up, and encryption -- all these things. We’re really continuing down the path of educating developers that these services belong in the network.

From a programmer’s standpoint, it either just happens, which is the best case, or they can configure it as a service. The way we’re moving is to open up and provide some of our functionality to services that a system’s architect or an application person can take advantage of as they build out their applications.

Gardner:
Perhaps I had it a little bit backward. Instead of my seeing more items on my palette, perhaps I should expect to see fewer, as components that have perhaps traditionally been part of an application or middleware layer, migrate to the network layer. Then, I really don’t have to worry about them, and so perhaps the reuse and integration of these services into a platform reduces my options as a developer. Do you think that’s right?

Ruh: I think it's absolutely right, and I think you’ve got the point there, which is: Should every application implement virus detection and virus management? No. Should it be implemented for them? Yes. Should there be some ability for the applications to control other set policies? Those are things that should happen, but the fact is that there are a lot of things that are infrastructure that should just happen, where programmers don’t have to worry about them as they do today.

Gardner:
In wrapping up, why don’t we try to get into some more of the reality side of this in terms of implementation? If I’m a CIO, or if I’m a developer, or architect beginning to see the logic of bringing services from middleware and from a siloed basis onto a more ubiquitous platform for reuse, how do I get started, where do I put the pencil on the paper first to start in this direction?

Ruh:
What we’ve been doing in working with customers down this path is, and -- as strange as this is going to sound -- it really does start with a strategy and roadmap. The IT organization really needs to make sure that they have their strategy for their IT environment laid out, as well as the kind of roadmap they have to get there. That's what allows the dialogue to take place at the next level.

What you don’t want to do is to say, "Okay, I need services. Let me go buy services, put them in;" to do the Field of Dreams thing ... and they will come. In the past, as we’ve learned from object oriented, building common services, and having the Field of Dreams approach doesn’t work.

What does work is if you’ve got your strategy and your roadmap for your IT organization, such as where your investments are going to be made, what applications you’re going to be building. The best strategy is to come in and say, "Okay what is the best approach -- architectural approach -- to get there?" Now you can begin to integrate with the architectural group to say, "We know that we want to start to put in place certain B-to-B capabilities or certain other applications."

Maybe the best place to start is with identity services within the network or maybe the best place to start is with mobility services in the network. Or maybe really, I want to improve the data center. If I take a retail operation, if my strategy and roadmap is really that, I’m going to invest in the store operations in the stores. I may really focus on identity management and wireless services in the network. Maybe I’m going to move to wireless POS (point of sales).

Maybe I’m going to do e-learning. Maybe I’m just going to just change the nature of the store. If I've already invested in the stores, what I really want to do is I want to save money in my back-ends and I want to improve my data center operations. Then, it may be that putting in storage networking services, putting in a compute services in the network to virtualized storage and servers, maybe those are much better approaches.

The fact is you want to tie bringing in new services to the investments you plan to make that will be tied to the business problems that are out there. Then, very naturally, over the course of three years, as you execute that strategy and roadmap, you end up with these services being built in and being paid for as a part of that natural investment. Where we have been coming in, interestingly enough, is we are now getting involved with those strategies and roadmaps by doing a lot of workshops and planning with them, and then bringing in what SONA can provide in tying down to a real ROI analysis. Then, you get into the architecture and the technology and how it fits in. And, the part everybody really loves-- you finally get down to buying specific products that implement that capability.

Gardner:
So I guess two takeaways from that would be: The good news is that I can do this incrementally; that I don’t have to build it all and then throw a switch one day. It can start on a crawl-walk-run basis. But, in order to make that plausible, there also has to be a top-down mandate for these cultural silos. And all of these different parts need to act in some concert toward an iterative build-out and adoption. There has to be some word from on high that this is our strategy, this is the roadmap, and this is what we’re going to do. Does that sound about right?

Ruh: That sounds about right, and it’s getting people to realize that this can be done incrementally. It makes the most sense to have the discussion around a business problem that you’re going to invest in anyway. And then very naturally the network folks, the app folks, and the architecture folks are sitting around the table and having the right discussions -- which is: which services really belong in the network, which services are going to give us the biggest bang for the buck, and which services quite naturally belong in the other parts of the IT infrastructure? So, it really starts the dialogue down the right path.

Gardner: Well, thanks very much. We’ve been discussing mega-trends and convergence, network services, application services, and SOA. Joining us has been Bill Ruh, the vice president of Advanced Services at Cisco Systems. This is Dana Gardner, principal analyst at Interarbor Solutions for BriefingsDirect. Thanks for joining us.

Ruh: Thanks a lot.

Listen to the podcast here.

Transcript of Dana Gardner’s BriefingsDirect podcast on network services and SOA convergence. Copyright Interarbor Solutions, LLC, 2005-2006. All rights reserved.

No comments:

Post a Comment