Cisco's Charlie Giancarlo dives deep on applications oriented networking
"Opportunity is missed by most people because it is dressed in overalls and looks like work." Thomas Edison
I picked up that gem of a quote yesterday from Intel Senior Fellow Justin Rattner, who spoke at MIT Technology Review's Emerging Technology Conference in the People's Republic of Cambridge, Mass. Rattner was outlining Intel's efforts to develop technologies tailored to the needs of consumers in emerging markets like China and India, including a specially hardened PC that can be used by entire villages to access the world. According to Rattner's stats, nearly 80% of the U.S. personal computer market has been penetrated, compared to less than 5% in emerging markets.
Rattner was joined on stage by Cisco Chief Development Officer Charlie Giancarlo, who joined me for lunch after the presentation. During his on-stage discussion, Giancarlo talked about how the speed of data center interconnects is coming to match the speed of computer backplanes, which enables customers to deconstruct the server as we know it - separating processors and memory, for example, from the disks - and creating greater opportunities for virtualization of computing. In essence, advances in networking are redefining the "computer" as we know it today and, as a result, changing the role and shape of the operating system of the future. All of this could go a long way toward resolving the scaling problems customers face, Giancarlo claims.
Over lunch, I dived deeply into Cisco's Application Oriented Networks (AON) strategy, which was rolled out in June. AON has been met with a mix of confusion, skepticism and intrigue. It sounds great to talk about how the network will provide better support for applications - including Web services-based apps - but what exactly is Cisco planning to do and what will customers be required to buy and manage?
Some folks positioned AON as simply Cisco's introduction of an XML switch, but Giancarlo says that's a mischaracterization. Cisco is beta-testing what he calls a multiprotocol message router (an MMR, and I'll get to that in a second), but that's only one step on a longer road toward ensuring that not only do applications perform better but that the network provides some intelligence that improves business processes.
For example, he says, the router could look inside a message and determine that some action needs to be taken - say, a P.O. needs to be approved by a senior manager - before the information is forwarded. In addition to the new router, Cisco is providing lower-level functionality such as compression and app acceleration that also improve performance.
So what's an MMR?
Well, it's helpful to use an analogy to understand it. A traditional router moved and translated packets sent among diverse networks - an IP network, a Novell or Apple LAN or an IBM SNA environment. Similarly, the MMR will move and translate messages among diverse message-oriented apps platforms. The messages could be in an XML format or in the format used by TIBCO, IBM's MQ Series, EDI messages or proprietary message formats, among others. The router connects these systems and rapidly shuttles information back and forth, improving apps performance. (Message switching has also garnered the attention of Intel, which recently acquired XML switching company Sarvega. That move will likely help Intel improve the performance of its server systems.)
More important perhaps, the router (with its graphical interface) becomes a single point of change that frees an enterprise from modifying each of the applications that support its intra-company or inter-company processes every time business needs change. As mentioned above, it would also enable enterprises to embed business logic in the network that could ensure that business processes are handled in particular ways.
Two questions about this.
1. Do we need or want the network to do this?
2. How will this differ from what Juniper will ultimately deliver to customers? Juniper has made application acceleration the centerpiece of its attack on the enterprise and it's clear that it will be a major front in the war with Cisco. What will Juniper do differently?
I'll explore both in upcoming postings.
For now, here are a couple of NW stories on the topic of AON. First, a piece on AON's introduction, and, second, a column by Kevin Tolly on what AON means.
Back to Vortex Blog
Comments
John,
Your AON blog caught my attention: every major new initiative meets with skepticism and confusion initially, and the good ones usually clear up the fog within a year with real products, customers and value propositions. I think its too early to tell about AON, but I would say one thing: its high time that the application and the network worlds converged. One thing no one seems to talk about is the huge separation in the corporate enterprise between the network and the applications.
Large corporations are increasingly rationalizing the network and the data center into a horizontal corporate function, while the applications are still stovepipes in lines of business. Network and data center folks have little if any knowledge of applications, and application folks have little knowledge and interest in the network. To them its dial tone.
To me the big question is not how AON will be pulled of technologically, but organizationally. We at Business Signatures run into this often, and our approach has been to package half of our application into a software appliance in the network so the network folks can treat it like a router or switch, and the rest of our application as a LOB specific configurable system so the application folks can get their hands on it. It works remarkably well in keeping both sides of the organization happy.
A critical success factor for AON will be not how to engage the network organization: if Cisco can't do that we should all go home; the key challenge will be how to get the LOB application folks engaged with a compelling AON value proposition they actually can get their hands on. I suspect middleware partnerships will be key, because its middleware that sits in the "middle" of the network and the app. If that is the case IBM and Websphere will become the control point for AON: and that's an interesting possibility. How will Cisco ensure it doesn't lose the leverage point in its own play? Remember Oracle and the Network Computer initiative. I believe it was Dell who leveraged that initiative to its advantage, not Oracle.
Posted by: Peter Relan on October 13, 2005 06:58 AM
Post a comment
