Microsoft It's all about innovation

Jawad Khaki, Corporate VP, Windows Networking & Communications



For Microsoft, IPv6 is the platform for software innovation. Microsoft Windows XP Peer-to-Peer Software Development Kit (SDK) beta, which the company released in February 2003, relies entirely on IPv6 for communications. It would enable new scenarios in networked computing. We interviewed Jawad Khaki, VP of Windows Networking on the future of networked applications and the role of IPv6. This interview was conducted at Global IPv6 Summit in Madrid in May 2003.

IPv6style:How far do you think Microsoft has come, in terms of exploiting the promises that IPv6 brings?

Khaki:Microsoft has been very aggressive in moving the world of IP version 6. But it’s a big job, as you know. We think it is important that all Windows platforms have IPv6 support. That’s true today. We thought also it was important that there would be more infrastructure using IPv6 that enables new applications to be written. So we have come up with Peer-to-Peer SDK. It was also important for us to show an example of IPv6 applications that people could write. That’s Three Degrees. So, in terms of showing what needs to be done, I think we’ve come a long way.

I think that scenario for consumers, or what we need to do for consumers, is largely in place from technology point of view. Now it’s about partnerships with operators and service providers in the deployment of these scenarios.

IPv6style:The Peer-to-Peer SDK you just mentioned is accompanied with IP stack update containing Teredo, for automatic tunneling of IPv6 over IPv4 network Would you say that as a software vendor, Microsoft simply cannot wait for service providers to offer IPv6 connectivity?

Khaki:You can look at it in a couple of ways. First, it would take a long time for the infrastructure to change. Infrastructure will not change overnight. People have deployed NATs (Network Address Translators) which they will not get rid of anytime soon. Many people don’t even know they got NATs. This is why Teredo is very important, because it enables you to transparently go through NATs, especially when you have double NATs.

The second thing is, people will focus on new scenarios. If I tell you to use a new digital gadget instead of using pen and paper, you would say, “Yes, I would like to start using the new gadget but I still want to use pen and paper, too.” I think this is how we should look at IPv6. We should ask, “What are the new scenarios that IPv6 enables?”, and let people deploy those new scenarios. Then they will say, “Hmmm, I got two routers, etc. I can save money by combining the infrastructure.” I think such cost savings will happen as we make things easier to deploy, easier to manage, and more secure.

So to me, Teredo is a necessary piece to jumpstart IPv6 deployment. Teredo solves a very important customer problem today: end-to-end connectivity using IPv6.

At the same time, I am very excited when carriers deploy IPv6. But we can’t wait for everybody. Japan is very progressive here. The rest of the world is lagging in public IPv6 connectivity services.

IPv6style:How will Microsoft IPv6-enable existing applications, such as MSN Messenger?

Khaki:Applications we provide will be updated with IPv6 over time. It’s the question of getting the work done. As for software running in Windows, like Internet Explorer, we have already done the work. I’m sure the Office team and MSN Messenger team will support IPv6 at some point when they are ready. The reason why it is like this is, you need to get the operating system platform ready first. Once you get the operating system ready, applications can move. That’s the work we are doing right now.

Peer-to-peer applications are really exciting, because it is all about being connected all the time, on broadband or wireless networks. I think two things will happen. One is that new kinds of applications will come around, like Three Degrees. There will be even better examples of peer-to-peer applications, I hope. Then, existing applications like Office might find peer-to-peer very useful infrastructure to provide more value to the end user. If you work on a Word document, you want to be able to collaborate with your friends across the network. Or you are using PowerPoint and you want to show the slides across the network. In many cases, there would be more than one person, maybe four or five people at different parts of the world. Peer-to-peer with IPv6 enables applications taking advantage of end-to-end connectivity.

IPv6style:How do you think will IPv6 make way into enterprises?

Khaki:Adoption of IPv6 in enterprises will happen in a couple of scenarios. One will be new applications like peer-to-peer, because Peer-to-Peer SDK only runs on IPv6. New applications written to our peer-to-peer APIs will require IPv6. IPv6 can be used on ISATAP, native, dual stack, or whatever. The other scenario is, people will do a specific project, let’s say streaming of media content. They would be willing to deploy IPv6 side by side with IPv4 for that project. Such scenario has to be one that require end-to-end connectivity, communication, media streaming, etc. Mobility and security are fringe benefits that come with IPv6 and some customers may want them, although standardization work on them are still not quite complete, and that obviously needs to get done.

At some point, once people have dual infrastructure, they will opt to optimize it. They would say, “I don’t want to run IPv4 and IPv6 side by side.” Then they will move to single IPv6 network, which I think will be quite a few years from now.

IPv6style:Right now, your primary focus is on consumer market?

Khaki:Correct. Our initial focus for IPv6 is on consumers, because their pain is already huge. The pain, for example, is that real time communications using Messenger is difficult because of NATs. Application developers cannot deploy new applications because NATs do not have the intelligence to do necessary protocol translations. So we have done all the work necessary to enable consumer scenarios. And we are excited about that. I think this will lead to enterprise adoption. It’s like messaging. When messaging applications came around, people started using them for their work. And it came to the point where we need to provide enterprise messaging applications. So I think there would be new peer-to-peer applications that would result in similar deployment in the enterprise.

IPv6style:Then the enterprises will feel the pain from NAT.

Khaki:Right. They will realize that they can’t take advantage of cool innovation from peer-to-peer developers that can bring benefit to them to help them get their work done. Eventually, they will have a choice to make: do they want to remain in old ways, or move along with changes? At that point, they will have to plan on IPv6 deployment.

There is also a scenario with distributed software. If you are in a large enterprise with a remote connectivity or a WAN link, it doesn’t make sense to go to n machines, n times (on the other end of the link) for file distribution. What you really want to do is transfer your data to one machine on the other end and make the data replicate itself over there. So that’s peer-to-peer content distribution in the enterprise that leads to cost savings.

IPv6style:What is your opinion about seemingly “slow” adoption of IPv6 in general?

Khaki:I don’t think IPv6 adoption is slow. “Slow” or “fast” depends on what your expectation is. If you had an unrealistic expectation, you would say it’s slow. But if you’ had realistic expectation, you would say it’s on course. From my point of view, adoption of IPv6 happens when new customer solutions come around. And I’m excited, for example, that several thousand developers have already downloaded our Peer-to-Peer SDK beta. People are downloading it immediately to play with it. To me, that looks like an adoption of peer-to-peer. Adoption of IPv6 in the enterprise is an unrealistic goal at this point. It is unrealistic because IPv6 solutions are not completed yet, from manageability, security, or deployability point of view. More work needs to be done by companies like Hitachi, Cisco, Juniper, or Microsoft to focus on the scenarios for the enterprise, how the network needs to be configured and managed, and how we make it secure using IPsec. Open questions need to be answered. Once they are answered, we’ll see enterprise deployment. Right now, we see no enterprise deployment of IPv6. Am I disappointed? No, because I don’t expect it.

IPv6style:As VP of Windows Networking and Communications at Microsoft, do you care if IPv6 won’t get deployed anytime soon?

Khaki:I do. But I don’t really think in terms of IPv6 deployment. I think about solving customer problems. And IPv6 is a way to solve customer problems. IPv6 solves customer problems because it offers end-to-end connectivity. End-to-end connectivity is necessary for innovation. That’s why I care. If IPv6 won’t get deployed, I feel very bad about it. That’s clear. But I am realistic about the pace of deployment. I’m excited and happy about the progress Microsoft has made, and the work Japanese industry has done. They are really leadership work. We were out there, giving customers software and hardware to go deploy. That is really what progress is about. But as we actually work with the customers, we learned that we didn’t think about everything. We had to be open-minded about it, and come back with the feedback, and incorporate the feedback in our products and go out again. Manageability of IPsec is one example. I care about IPv6 deployment. It’s important to me. It’s important to me because it solves customer problems.