iXBT Labs - Computer Hardware in Detail

Platform

Video

Multimedia

Mobile

Other

Internet2 in Russia — Already Reality?



Corbina Telecom, Russian all-around telecoms operator, has started the Internet2 project in Moscow, which will connect the clients of Corbina Telecom — both present and future — into a high speed data network possessing much wider features in comparison with the networks of the previous generation. On December 15, 2004 the company arranged a press conference, where several features of this data network were demonstrated.

We'll return to them a tad later, and now a bit of history.

In distant 1968, an organization named ARPA (Advanced Research Projects Agency), department of the ministry of defense of the USA, built the first computer network based on the principles, which are still used in modern Internet. It consisted of four computers. Many organizations and universities connected to ARPANET in the following ten years.

All basic protocols were implemented by 1978, which are used even now in Internet. The main protocol is IP v4 (IP — Internet Protocol). In 1982 a European UNIX Network is founded (EUnet). Before that only the USA, Canada, and Great Britain were included into the network.

The system of domain names appeared in 1984. In 1989 the number of connected computers reached a hundred thousand. By the end of 80s the Network comprised over dozens of countries.

WWW technology was developed in 1991. In 1991 Russia joined Internet. By 1992 the network counted over a million of computers. The web grew by three and a half thousand (!) times for 1993.

In 1994 Internet celebrated its 25th birthday. Since that time the Network has spread globally, but it hasn't essentially changed. A lot of new technologies were invented, communications channels got better, the number of computers increased to dozens of millions and the number of users to hundreds of millions. As a result Internet gained popularity, it became commercially profitable not only for network providers. Already in 1994 one could order pizza or taxi via Internet (of course we mean Western countries).

In 2004 Internet has not gone that far technically as in the end of 60s. The modern Internet features the following essential problems inherited from ARPANET :

 

  • Small address space — the modern Internet uses only 32-bit addresses (four bytes for one address), that is only about four milliard addresses are possible, which is not only less than the population on Earth, but also much less than the number of electronic devices. To say nothing of technological impossibility to use all 4 milliard addresses (because networks are segmented into subnetworks). Nobody could expect in late 60s that each mobile phone would have to have its own network address. Moreover there was no such notion as a "mobile phone" at that time as well as the great number of electronic devices, which we presently witness. Soon there may appear a situation when the number of available IP addresses is not enough for all applicants. Strictly speaking, they are not enough already.

 

  • IPv4 disadvantages include the lack of automatic address configuration mechanism. Ask any system administrator, and he will tell you what a useful feature it would be, especially when you have to switch a corporate network from one provider to another. Of course there are external services for automatic configuration (DHCP), but this mechanism is not hardcoded into the protocol — it's installed as an additional service, that is it also requires additional configuration, etc.

 

  • Low performance – Internet was created when the fastest channels provided only scarce kilobits per second. That's why the computational processes laid in the foundation of the Network in 60s are not that optimal for these days. Packet fragmentation is an example of poor algorithmic solution in IPv4. The fact is that too large ethernet packets (their maximum size is 64 kb) can be divided into several ones, because many network technologies operate with packets of smaller size. This operation is often carried out by intermediate routers, through which the data flows. The problem is that packet distribution takes up much system resources of the router. Thus, this process not only hampers file transfers of users but it also consumes additional resources of intermediate routers.

 

  • Inadaptability to transferring data critical to delays. Transferring voice and video via modern Internet is always a challenge, because Internet does not guarantee the quality of service. Voice traffic over Internet can arrive with random delays, at random intervals, with losses, which results in voice distortions and quacking typical of VoIP. The fact is that when the IPv 4 protocol was in the development stage, practically all network interaction came to ordinary file exchange. Today there appeared a lot of new applications in Internet, including streaming ones (for example, Streaming Audio and Streaming Video). And they require exact specification and constant maintenance at the specified level of such parameters as throughput, delays and delay variations. Of course there have been attempts to introduce QoS into modern Internet, which basically come to allocating a fixed bandwidth for traffic streams. But these are only "struts", which are not always helpful. QoS procedures are basically effective within high-speed corporate networks, they are much more difficult to set up in Internet.

 

  • Because of address space shortage corporate networks hide all their computers under a single IP address (NAT — network address translation), which reduces performance and leads to purchases of additional equipment to connect the corporate network to Internet. Though NAT also has an indisputable advantage — hidden under a single address, a network is considerably more difficult to get access to for a malicious cracker.

 

  • Security problems. Internet access providers cannot be sure whether a person on the other end of the line is really who he/she claims to be. Without special measures taken, users cannot be sure that their financial information will not be stolen or modified when transferred via Internet. You cannot usually be sure that the information transmitted via Internet is really sent from the address specified in the IP packet. However widely used SSL protocol features practically solve this problem.

 

  • Impossibility of data broadcasting via Internet. Transformation of Internet into a universal information transfer medium, particularly into a TV and radio broadcasting medium, is impeded by the necessity to send a copy of the TV data stream to each user (though many users are watching the same TV program simultaneously).

 

All these problems are solved by the new version of the IP protocol – IP v6. IPv6 development was started in 1992. Since 2003 it has been supported by manufacturers of most telecommunications equipment (corporate level).

 

IPv6 is a new step in Internet development. This protocol is developed allowing for the growing requirements to the Global Network, which stopped being just a file distribution network long ago. Today Internet serves to transmit huge amounts of data, including audio and video streams, secret information. To say nothing of problems with insufficient IP addresses, which may appear in the nearest future. That's why the initiated introduction of IPv6 is very important. Let's hope that it will go nice and easy and will solve many problems of Network users.

 

 

 

  • IPv 6 offers 128-bit addresses (16 bytes, it's 4 times as many as it used to be), which are configured absolutely automatically, clearly for a corporate user, which allows to save the working time of system administrators. You can fit as many devices as you need in the 128-bit address space. The total number of addresses in the 128-bit address space is approximately 3.8 * 10^38 (^ - power), it's a huge figure (IPv 4 offered only 4.3 * 10^9).
  • IPv 6 offers multiple improvements of the packet format, thanks to which IP packet routing is faster and consumes less computational resources of the routers. For example, instead of fragmenting data packets on the way, the sides calculate the maximum possible packet size between them beforehand and send packets no larger than this size. It saves their time, the time of intermediate routers and increases transfer rates. IPv 6 does not require calculating the check sum of IP packet headers on each router (it's assumed that all modern routers have built-in error control). Header check sum calculation is not an expensive operation, but when millions of routers stop doing that for trillions of packets each second, the speed of the entire network will grow considerably. Besides, the IPv 6 packet header size is now fixed at 40 bytes (in IPv 4 the header size varied from 20 to 60 bytes), thus IPv 6 overhead comprises a fixed value – 2.6% of the overall traffic volume (in IPv 4 this value varied within 1.3 - 3.9 %).
  • IPv 6 guarantees the quality of service, that is constant maintenance of network throughput and packet transfer time parameters. Video and IP-telephony, sent across the continents via IPv 6, will be of much higher quality and will not suffer from data transfer delays.
  • Corporate networks may have as many IP addresses as they need. Administrators will give up the NAT technology resulting in spared working time of system administrators, saved money for extra equipment, and increased data transfer rates. Besides the corporate network security will be enhanced.

  • IPv 6 supports network multicasting. A good example is an internet radio station. At present each user, connected to the radio station broadcasting a 128Kbit stream, gets its data stream, that is 10 users will require a 1Mbit channel, 100 — 100 Mbit/sec (though the data is the same!). So we cannot speak of any mass character so far. Multicasting allows to send only one copy of data (be it audio, video, or any other stream) via a channel shared by several users. That significantly saves the network throughput, and thus gives "green light" to video and audio broadcasting technologies via Internet on a mass scale. A small reservation — via a future Internet, which will support IPv 6.

 

The network supporting IPv 6 is called Internet2. That's how its logo looks like:




Like the first Internet, it was born in the USA. To be more exact, scientists had to look for an alternative to Internet because of its practically uncontrolled growth and overloading with parasitic (from the scientific point of view) traffic. Building a new network (codenamed Internet2) has become this alternative, this network initially connected dozens of universities and several corporations. Except for data exchange, this network served as a good testing area to break in new technologies. It was planned high speed from the very beginning (100Mbit and gigabit rates), so it was necessary to develop new routing data transfer protocols as well, because the existing protocols would not have been optimal in the new network.

 




The network has been growing rather fast and it now covers quite a large area of the USA. Its core is the Abilence backbone, its throughput being 10 Gbit/sec, and each connected computer gets no less than 100 Mbit/sec. The current network load can always be seen at this web site:

 




sample Abilene load diagram

 

Clients are connected to the network via "points of Presence" — gigapops. They allow to use effectively the throughput of the entire network.

This network is mostly used for scientific purposes — remote control of experiments, observatory access, distributed processing of huge data arrays, and certainly digital video. Multicasting features open up practically unlimited prospects (within reasonable limits, that is the overall network throughput, of course), that's why the network is actively used for video conferencing as well as for regular video broadcasting. For example for distance learning — lectures are broadcasted over the network live or from the archive.

And though the coverage of Internet2 is constantly expanding, the increasing number of new networks is connected to it, and there are even points of presence in Europe, objectively speaking, "they" have the most interesting features. And what about Russia?

In our parts, as we have already written in the very beginning of the article, Corbina Telecom was the first to announce putting into operation a fragment of the network working with IPv 6 and having a 10 Gbit/sec backbone. So far this network segment covers only two backbone nodes of the company and a network of Corbina's client — Niko company. The length of this network fragment is 28 kilometers.

This segment is called Internet2, though formally it's not yet connected to the rest of Internet2 network. According to Alexander Malis, vice-president of the company, negotiations concerning this issue are carried on (in the final stage), global reunion is just a matter of time.

Backbone speed of the 10 Gbit/sec network is impressive. The company claims that at this data rate one second is enough to send half an hour of a high resolution video (HLTV). But it should be noted here that Corbina connects its clients to Internet2 at data rates from 100 Mbit to 1 Gbit, that is the same video fragment will be sent for 10 seconds. Just think about it — connections FROM 100 Mbit, it's like data rates in most office LANs... No 128 Kbit, and not a bit more!

We cannot but be glad to hear Malis saying that even now Corbina is ready to accept application forms to connect clients to Internet2, if they are within the MKAD (they will be connected within several months). Plus the company invests about 30 million dollars to create and develop multicast networks in 45 Russian regions.

And what about the existing Corbina clients? The company presently has about 1700 clients, it's planned to switch them to Internet2 simultaneously with connecting new clients. In other words, they prefer to connect new clients directly to Internet2, while existing clients will be switched to the new network as required. Besides high data rates, upgrade to the new network will be stimulated by lower traffic price. According to company forecasts, the transition will be completed within two years (that's a short term!).

But who will enjoy the new technology? That's a nasty thing for us, end users, so far — the company counts on corporate clients first of all. That is according to Malis, those who are ready to pay approximately 3 thousand dollars a month for such a channel. Yep, the traffic will be cheap (approximately $5 for a gigabyte), but no one promised small volumes :)

You shouldn't forget that the new technologies require new equipment supporting the "timely novelties". It goes without saying that the old equipment won't do for IPv 6. The IPv 6 has a new packet format, new routing technology, built-in multicast, etc. So clients will have to pay for new hardware as well. The comforting thing is that most equipment sold after 2003 already has the built-in IPv 6 support (or will have it with new firmware). Generally we mean routers, as in end computers one will just have to install IPv 6 stack, which is rather easy both in Windows and in *nix-like systems (most such systems already support IPv 6).

Now about compatibility. We have backward compatibility, that is one can see an IPv 4 network (old Internet) from the IPv 6 area, but not vice versa. In other words, clients connected to common Internet will not see the Internet2 network. One remark here — they can see the new network, but not all. Only those who can set up tunneling into the IPv 6 space using one of the IPv 4 -- IPv 6 gates, for example Hurricane Electric's IPv6 Tunnel Broker. Hurricane Electric provides free access to the IPv 6 space via its gate. You just register, enable IPv 6 stack support on your computer, and properly set up traffic tunneling.

Drawing a preliminary conclusion, one can say that Corbina Telecom is doing a "nice and easy" revolution — access to this service, even for corporate clients, cannot be left unnoticed. At last Internet2 has definitely gone out of the "wow, how great, but this is not for us" talk stage and started its inexorable course across Russia. OK, not the entire Russia, but that's just a start.

We want to note Corbina's plans on a wide introduction of video broadcasting along its Internet2 network (indeed, why not take advantage of brilliant QoS and multicasting features). The company has prepared technically beforehand — in mid 2004 they opened a noncommercial project www.corbina.tv, which was designed to demonstrate features of modern high-speed connections to Internet. In other words, you can watch TVC, CTC, REN-TV, RBC-TV channels as well as the First channel from this site (only if you are connected to Corbina Telecom). Plus you can listen to the Silver Rain radio station. Note that all this is done within the IPv 4 frame, that is usual Internet, you only need a computer which will be able to show TV programs in real time. It's certainly very difficult to provide the quality of service of the stream in current Internet, so there may be delays and frame drops even on high speed (several Mbit) channels. But in Internet2 everything is fine with these issues — QoS is a constituent part of the protocol, and multicast saves the bandwidth considerably. That's why the new network will feature TV broadcasting at full swing.

Malis noted that Internet2 provides great broadcasting opportunities. Users will get access to dozens or even hundreds of channels and will even be able to create their own channels. You also shouldn't forget about the video conferencing feature, it will be extremely easy to set up in the new network — this service is already popular in the old Internet, while in new Internet it will be like a mobile phone, that is a standard thing.

But despite the optimistic statement in the previous paragraph, complete upgrade to IPv 6 is a very long process. Perhaps many corporate clients will switch to the new technologies, but what about those who are content with megabit channels, whose traffic is approximately $500 a month? Plus they will have to upgrade the entire fleet of switches/routers (because offices are usually equipped with devices below several hundred dollars)? And home users are even in worse situation — I may be a pessimist, but I cannot imagine that the other providers will switch their clients to IPv 6 and connect them to the global Internet2 network almost for free. Well, of course they will eventually, but to my mind it's a very long process, much longer than two-three years. And don't forget about numerous homelans, that is a community of several neighbouring residences connected into a shared LAN...

But still I'd like to finish this article optimistically. That's why I will publish the forecast of Alexander Malis (his personal forecast), which he shared in the course of the press conference. This forecast concerns the future of the Internet market in Russia:

— "In the nearest future there will be two types of Internet (for private citizens): unlimited 400--512 Kbit access at $15-20 and traffic-paid 10--100 Mbit access at 3$-10 for a gigabyte"

Well, I'd like to hope that this will happen in the nearest future...


Evgeniy Zaitsev (eightn@ixbt.com.)
January 6, 2004



Write a comment below. No registration needed!


Article navigation:



blog comments powered by Disqus

  Most Popular Reviews More    RSS  

AMD Phenom II X4 955, Phenom II X4 960T, Phenom II X6 1075T, and Intel Pentium G2120, Core i3-3220, Core i5-3330 Processors

Comparing old, cheap solutions from AMD with new, budget offerings from Intel.
February 1, 2013 · Processor Roundups

Inno3D GeForce GTX 670 iChill, Inno3D GeForce GTX 660 Ti Graphics Cards

A couple of mid-range adapters with original cooling systems.
January 30, 2013 · Video cards: NVIDIA GPUs

Creative Sound Blaster X-Fi Surround 5.1

An external X-Fi solution in tests.
September 9, 2008 · Sound Cards

AMD FX-8350 Processor

The first worthwhile Piledriver CPU.
September 11, 2012 · Processors: AMD

Consumed Power, Energy Consumption: Ivy Bridge vs. Sandy Bridge

Trying out the new method.
September 18, 2012 · Processors: Intel
  Latest Reviews More    RSS  

i3DSpeed, September 2013

Retested all graphics cards with the new drivers.
Oct 18, 2013 · 3Digests

i3DSpeed, August 2013

Added new benchmarks: BioShock Infinite and Metro: Last Light.
Sep 06, 2013 · 3Digests

i3DSpeed, July 2013

Added the test results of NVIDIA GeForce GTX 760 and AMD Radeon HD 7730.
Aug 05, 2013 · 3Digests

Gainward GeForce GTX 650 Ti BOOST 2GB Golden Sample Graphics Card

An excellent hybrid of GeForce GTX 650 Ti and GeForce GTX 660.
Jun 24, 2013 · Video cards: NVIDIA GPUs

i3DSpeed, May 2013

Added the test results of NVIDIA GeForce GTX 770/780.
Jun 03, 2013 · 3Digests
  Latest News More    RSS  

Platform  ·  Video  ·  Multimedia  ·  Mobile  ·  Other  ||  About us & Privacy policy  ·  Twitter  ·  Facebook


Copyright © Byrds Research & Publishing, Ltd., 1997–2011. All rights reserved.