SCS
for 10 Gigabit Ethernet
Harry Forbs is the technical director of the
company ITT NS&S. (http://www.ittnss.com/)
Publishing house "Open Systems" (http://www.osp.ru/)
Methods which invent in order to «compel» the
UTP systems properly to support speed 10 Gbit/s on the large account are attempts to compensate absence at these
systems of screen. No balancing will deliver from all problems simply because
the ideally balanced cable does not exist, and against them screening is the
most effective method of fight.
Seemed,
quite recently we worked in the Ethernet networks built on a base concentrators
10 Ìáèò/ñ, and gygabyte speeds appeared by something fantastic, from the distant
future. But technical progress walks ahead by sevenmiles steps. Digital
technologies of storage and treatment gradually engage all new and new types of
information in the orbit. Graphic arts and video, multimedia and various vocal
appendixes is all of it finds wide application in corporate networks. Mass
development of the systems with support of the Gigabit Ethernet technology, on
the prognoses of experts, will result in that in two years of about 70% all
ports of local network will be gygabyte .
If
to the work stations the Gigabit Ethernet channels are tricked into, means, on
server and interswitchboard connections speed must be yet higher, I.e. 10 Ãáèò/ñ in accordance with the tenfold
«hierarchy» of the Ethernet growth. And in course of time, at first for the
most «demanding» users, and then and for all other, technology 10 Gigabit
Ethernet will be inevitably inculcated on the areas of access. Therefore
far-sighted enterprises already today reflect about its use.
Even if such speeds while seem irrelevant, it
is necessary to give itself a report, that it is early or late they will «come»
in your network. Especially important to take into account it at development of
cable infrastructure which must be created with a supply, at least on five—ten
years ahead.
10 Gbit/s ON COPPER
Although decisions for 10 Gbit\s at many is associated foremost with the
optical cable systems, on the horizontal areas of the structured cable systems
(SKS) cables with the copper twisted pair will prevail yet long. Interest to the
potentially inexpensive copper decisions 10 Gigabit Ethernet warms up
circumstance that cost of active equipment with optical ports 10 Gbit\s remains ever-higher — about 10 thousand of
dollars calculating on port.
In November, 2003 the IEEE institute formed the
working group of 802.3an, setting before it the problem of development of
standard of 10GBaseT for the information transfer at a speed of 10 Gbit\s on cables with the twisted pair. At saving of
such Ethernet, as format of shot, this
standard foresees the simultaneous transmission of traffic on each of four
twisted pair of cable on speed 2,5 Gbit\s in the
full-duplex mode (in both directions), that provides in a sum necessary 10 Gbit\s. Although final ratification of standard of
10GBaseT is set only on the summer of 2006, many enterprises already were today
caused anxiety by the choice of the cable system which would guarantee valuable
support of network technology 10 Gbit\s. What
variants are offered by a market?
THE
UTP PROBLEMS
The unscreened cable systems (Unshielded
Twisted Pair, UTP) already once or twice refuted the prognoses of sceptics,
rising all higher and higher on the «stair» of the Ethernet speeds: 10 Mbit\s, 100 Mbit\s, 1 Gbit\s... However situation with support
of technology 10 Gbit\swhile remains extremely misty. The
point is that the standart systems of the UTP Category 6 will be able assured
to support technology of 10GÂaseT only at length of channel no more than 55 ì. But, as is generally known, standard length
of copper cable channels makes 100 ì. Exactly it undertakes in a calculation in swingeing majority of the
SKS projects, and to look over this fundamental description, probably, makes
sense only for very specific cases, we will say for the compact centers of the
data processing.
The row of producers offered the cables of the
so-called extended Category 6, which, in obedience to their statements, able to
provide the transmission of informative streams 10 Gbit\s on required 100 ì. However in default of standard they remain a
firm, yes besides frequently by the exotic enough technical decisions. In
addition, all more frequent in this communication it is spoken about the
necessity of development of new methods of installation of such cables UTP,
that also scarcely will be instrumental in their wide distribution.
Alike, complications with creation of the
standard systems UTP for valuable support of technology 10 Gigabit Ethernet
grow as snow lump. So will producers be able to create such systems which will
satisfy to all requirements of developers, installators and end users SKS?
It is known that by the «heel of Achilles» of
the UTP systems at data communication at a speed of 10 Gbit\s there are the intercable aiming, or Alien FEXT
(AFEXT), I.e. aiming from the twisted pair of neighbouring cables. The digital
methods of treatment of signals allowed to score great successes on business of
indemnification of aiming arising up into a cable, and on the intercable aiming
serious attention was not paid long time. The especially large problems of the
AFEXT aiming are caused in the short channels UTP, and noticeable progress is
not while visible any in the decision of this problem. As it turned out
recently, not less complications arose up wherein short channels combine with
long, even if the algorithm of «optimization of power is used» (power backoff).
It supposes the use in the short channels of signal of less power, than in
long, that must, by plan of developers, to limit the level of the intercable
aiming. However in practice a situation can even become worse.
DECISION
— IN SCREENING
Methods which invent in order to «compel» the
UTP systems properly to support speed 10 Gbit\s on the large account are attempts to compensate
absence at these systems of screen. The apologists of application of the UTP
cables «for the decision of any tasks» like to repeat, that such ideally
balanced cable absolutely radiates nothing outside and protected from all
external aiming. But if it was actual so, then the intercable aiming did not
represent a hindrance. The representatives of companies producing the screened
cable systems supposed always, that no balancing would deliver from all
problems simply because the ideally balanced cable does not exist, and against
them screening is the most effective method of fight.
The producers of the UTP cables offer mass of
devices, to do them suitable for support 10 Gigabit Ethernet. This and division
of cable channels at the level of commutation panels (separation of commutation
cords from each other and increase of distance between the ports 10GBaseT), and
creation of air-gaps between cables at their gasket in kabeleprovode, and
increase of length of commutation cords, and combination of short and long
channels, and the use of technology of power backoff. But, alike, it is
only-only succeeded to manage with one problem, as here is other.
A situation arising up during modernization of
the cable system very a model in that behalf, which the cables of the
Categories 5å and 6 are
in. We will assume, cables are laid in the main system of building between a
basic instrument room room and vehicle on every floor. In course of time for
support of users of high-performance appendixes (say, systems of the automatic
planning SAPR) working on one of floors, it is decided to organize channels 10 Gbit\s on the UTP cables. Ordinary practice of gasket
of cables in vertical baskets consists that a new bunch of cable is placed next
to present. Thus, the UTP cables intended for the systems 10 Gigabit Ethernet
will neighbour with the cables of Categories 5å and 6, which inevitably will bring in the
contribution to the increase of the intercable aiming AFEXT.
So, will different devices and limitations to
become to the systems the UTP so suitable physical environment for networks 10 Gbit\s offered by suppliers be allowed, what are the
screened decisions? And they will be able to remain acceptable at price, to
save the sufficient degree of flexibility and did stock by recommendations,
that do both designers and users of modern networks need? These questions while
remain without an answer. And as the prosecutions of standard on the systems of
10GÂaseT approach
completion, optimism concerning application in them of the UTP cables is
reduced.