HOME | INFOTECH | HEADLINES |
January 23, 1998 |
'Applets run on clients, servelets on servers. We'll
Aren't you fed up that each time you surf the Net it gets cloggier?
|
T O D A Y |
Novell's Glenn Ricart VSNL board Sun posts India results Techna's Jap deals |
Well, instead of generating all this repetitive traffic over and over again, wouldn't it be great if your office server or some other machine not very far away, could hold the site temporarily and then save the round trips of data through the telecom networks of several nations.
Dr Glenn Ricart, chief technical officer of Novell Inc, the fifth largest software company in the world, claims to have a solution. He calls it the Border Manager.
But that's not all, he is also thinking of putting Java into data packets over the Web. 'Jacket', or Java packets, is what he calls it.
All this effort is part of the 'intelligence Internet infrastructure' initiative of his company.
The solution, he says, could be of particular significance to bandwidth and infrastructure challenged nations like India.
He was in Bombay recently, on his way to the India Development Centre at Bangalore, Novell's largest R&D set-up outside of the US. Srinivas Rao Konakanchi cornered him then to bring to the Rediff faithful an exclusive interview:
Why is the Internet a source of irritation to many governments across the world?
There are lots of problems. No one is in charge. There are no censors. There is no supervision. No one polices it. No wonder it has become a source of irritation for governments. It's just like what happened in China after the Tiananmen Square incident when reporters faxed their stories. The fax machine became an irritant to the Chinese government.
After the initial euphoria, many nations have discovered that the Internet violates their sovereignty. For instance, Middle Eastern nations feel that a large number of sites that can be reached over the Internet are not commensurate with their national religions. But no one can control this.
And many more feel it is a threat to government functions themselves.
What's the big issue about protocols today?
This is a big problem. The key issue is that the protocols are ageing. I believe that these protocols will not be able to deal with gigabit Ethernet, fast fibre that runs under the oceans and also with the merger of telephony and video.
Telephone and video signals are not very well handled by the Internet. If you have tried any of the Internet telephone products you will know that they don't work very well as these protocols are not very well suited to it. So we will see a reinvention of these Internet protocols.
What's IP version 6 and what's so hot about it?
It is the new version of the Internet Protocol, which has two important characteristics for customers.
One. It extends the addressing and that's what everyone is talking about because we are going to run out of Internet addresses by 2003 or 2004. And IP version 6 is going to show the way out. More importantly, in IP version 6, they are adopting the IPX style of addressing.
IP version 6 provides for greater security and authentication header extensions which makes it an inherently more secure protocol than IP version 4 is.
I think companies will begin using IP version 6 for the additional security it provides. I guess that will propel IP version 6.
So will the requirements for addressing. Out of the existing networks that are already on the Internet, I am trying to do some research and find out what fraction or percentage is going to need new addressing.
It is definitely not going to be 65 million new addresses that are needed. It's going to be a much smaller number.
Another fact is that the number of addresses available right now is in the billions range. So we are okay there. But we are witnessing compounded growth and we will need many more new addresses.
The other issue is that the governing structure will change and I think the ISOC will have to step up and take a bigger role, take over from the National Science Foundation in the United States that issues the names and addresses. They will have to be truly global.
What other changes do you foresee in the next couple of years on the Internet?
There is a law known as Metcalfe's Law that says the value of the network increases proportional to the square of the number of connections. This means that it is much better when you get many more people.
The difficulty is that the number of crackers, hackers, the bad guys, is also growing rapidly and that is a very serious issue. It means we have to step up Internet security dramatically.
Perhaps the most important change I see coming on the Internet is that we will have to get some rational economics in place. Right now it costs the same to retrieve off a Web server in Sweden as it costs to retrieve off a Web server in Bombay. That does not make any economic sense. It has to be that things which are closer should cost less. So we need some rationalisation of the economics of the Internet.
Some people suggested advertising would provide that. But we have recently seen that the rates for banner ads have gone way down. I really don't know?
But I do feel that the 'intelligent Internet infrastructure' from Novell will make a big difference.
Isn't bandwidth growth a critical issue?
As far as bandwidth goes, it will continue to grow. My guess is that bandwidth will grow much more slowly as we insert more and more 'intelligence Internet infrastructure' instead of bandwidth. But we will see it growing.
The best part of fibre is that you can change the electronics at both ends and get much more bandwidth through it. This is going to be very important. Another issue is copper, the lifetime of it is only about 50 years and it requires a lot more maintenance compared to fibre and you cannot increase bandwidth on it as much as you can on fibre.
So infrastructure is going to make a difference.
Is the network computer going to take off?
It is a good thing for us if it succeeds but as you can see we are not making a huge bet on it. I personally think people will have a big machine and a little machine. The NC makes a lot of sense at places where people are not doing heavy-duty computing.
Why is that everyone still perceives Novell to be a legacy file and print software company?
That's not true. About a year back we explained that our server platform, NetWare, is not just about file and print services.
We said that we are going to add Java and make it a Java server. Well, we did it by taking the license from Sun. We also said that we are going give you a directory service. We just did it and it is available today, not in 1999 like Microsoft promises.
Today NDS (Novell Directory Services) runs on NetWare and UNIX. We have just announced it for Windows NT. We also recently announced a tie-up with IBM for System 390.
So we are in all ways uniting your entire network and it doesn't really matter to us what technologies you have got on it.
Today, with our software, anyone can have a single login, a single view of the entire network and a single way of setting up an entire network and that does not mean that there's just one technology in that network.
Then we have set up services - directory enabled services - ManageWise, GroupWise, file and print also use NDS.
The exciting new entrant is the Border Manager that has just taken off in a big way. Now all these services that earlier ran on IPX now run on IP version 4 and IP version 6 too. That's really a quite a lot of effort.
What's exciting about the Border Manager?
Border Manager was really changing the way that people were working… at what we were doing. It changed the way we do things at Novell.
Now you have a server, a directory and a set of services that run on the directory and also run other platforms as well.
But now we have an entrance into a new fourth category - and this is the 'intelligent Internet infrastructure'. This is really some kind of magic.
As companies do more and more of their work over networks, access the World Wide Web, link up with other companies, do more and more wide-area networking, they are discovering that a bigger and bigger fraction of their dollars spent on information technology are going to the telecommunications companies.
These companies provide the bandwidth between different countries. But now we can actually trade off servers - Border Managers - against that bandwidth.
When we took a look at what we are sending over our own backbone between Indian and the US, we found that 90 per cent of the information was duplicated. For example, people plugging into our inner-web home page were accessing the same document many times over.
Then we said let's take a look at what's coming into the country from the outside. We can save a lot of traffic if we can have a Border Manager to catch information that's common so that we can go and cut down our traffic by a factor of five, six or seven, depending on what's going on.
You have to realise that you don't have to always buy bandwidth but instead you can buy virtual bandwidth by putting in the Border Manager. Another thing is that it's much faster.
What's happening is that as the Internet gets bigger and bigger, the diameter gets bigger, the number of hops that information bits have to go through to get from one end to another also goes up.
You know it connects to a router and then it connects to another router and so on. Sometimes the chains are of 20-30 (routers) long. So even if you do have a good bandwidth, between the client and the server, it takes a lot of time because it stops at each of these routers and gets repeated down the chain/line.
If I can go and put in a Border Manager here which knows what you want and can turn the information around quickly enough, you get much faster response times.
So the key benefits are - there's faster response time, less traffic going over the Internet and it's much less congestion out there.
Sometimes the World Wide Web is called the world wide wait. Well, now you can do something about that and bring down a lot of cost as well.
What happens when people need to access live information?
What happens when you want live information? Like the BSE Sensex that comes with a lot of graphics and update and order forms. Well, keep in mind that all this represents the sending of huge amount of bits - 100 bits for variable information like numbers and say 10,000 bits for graphics information. But with the Border Manager we can do this by getting 10,000 bits from the Border Manager and fetch the 100 bits over the telecom line.
Well, is this a big deal? Yes it is. It lowers the cost of the network, it gives faster return to the customer and if your telecom infrastructure is a constraint, as it is in India and some other nations, you can now bypass this constraint by trading off new bandwidth for intelligence. We are the only company that's been able to do this. It's a new market opportunity.
What kind of an opportunity?
Well, there is the routers market. Then there is the servers market (Novell competes in this segment) and there is a third marketplace between the routers and servers and that is for the Border Manager kind of products.
And so everyone can use a Border Manager. It does not matter if you aren't on NetWare. You could be using UNIX or Windows NT or both.
That's a lot of companies and a lot of people and especially if the telecom infrastructure is limited or constrained, it's especially important.
Although we just announced the product last quarter and our sales have been booming, I feel that sales are especially important in countries like India where telecom infrastructure is developing very rapidly but is still largely constrained on resources for knowledge workers.
What's next on your technology agenda?
Infrastructure's for one. Caching is next. We'll be doing object management, prioritisation of network requests.
For example, if you are trying to move a mail message that can wait for a few seconds at the side and let the path go through for a videoconference so that there is no clogging, that can happen.
We want to put intelligence into the network.
Until now we have had Java at the client and the server ends. Applets run on the client and servelets on the server. We will execute Java in the network. We will have Java packets - Jackets, servelets and Jackets.
They are the third part of the infrastructure, as we get more intelligence into the network.
These jackets are very flexible in nature. For example, if Bombay, Bangalore, Delhi and the US are in a videoconference, Jackets can optimise network traffic.
The Jackets merge the packets of only those people who are speaking. So we are only shipping out of India the packets of the video people who are speaking to whichever city in the world.
These Jackets optimise the network traffic at any point of time. This technology has made it possible for us to start building intelligence into the network.
How are you looking at Novell's flagging revenues and the loss of its market share to Microsoft?
We are a billion-dollar firm and the fifth largest software company in the world. We have over 65 million customers using our products worldwide.
We continue to be successful, we continue to make 600 to 700 million dollars a year worldwide on the server software market and that continues to be strong despite what some people are saying about the NT server making inroads at the cost of the NetWare server. We have seen just about like 3 pre cent of our customers leave and go to Microsoft for an NT server.
Let me tell you something, Microsoft plays a lot of different games. One is that if for instance a customer buys an NT server, Microsoft says he switched, even if he kept the NetWare server. Most people do. It's a logical thing to do. People want a collection of these things as it provides a number of software application you can't get elsewhere.
But, what we are finding is that while people are adding NT servers, they are not taking out NetWare servers.
But isn't it a fact that Windows NT is rapidly gaining market share?
I think in the overall scheme of things the market is growing, we continue to ship increasing numbers of NetWare servers worldwide and unit shipments are going up.
Windows NT is also going up. When you take a look at IDC numbers, they show that NT is going up, NetWare is going up, OS2 is coming down and UNIX is coming down.
So most of the net growth for NT is coming from OS2 and UNIX space, and from NetWare it's about 3 per cent.
That's not very significant.
Our directory is the only real enterprise-ready directory in the marketplace. Microsoft is promising something with NT 5.0 in 1999. We'll see how that goes.
GroupWise, right now, is the number two product behind Lotus Notes. Did you know that GroupWise's eight million users outnumber Microsoft Exchange's users (2 million) by four times. It's really the hot spot for our growth.