LinkMeUp. Issue 2
Hello, colleagues.
In the new issue of discuss
1) SDH/PDH (Synchronous Digital Hierarchy/Plesiochronous Digital Hierarchy)
2) System of WDM channels in optical networks
3) Technology of redundancy to Ethernet network. Alternatives STP — ERPS, RRPP.
Download podcast.
Nevoshedshie podcast topics that we found interesting, we made Overtime.
I'm sure every telecommunications engineer knows or at least heard the name as a stream E1. But even the most advanced heard of the Nyquist theorem, the sampling frequency and PCM. The history of the development of networks PDH ( Plesiochronous Digital Hierarchy) began with the fact that by the mid 70s it became clear that further frequency seal is impossible in a single physical communications channel, it is limited to PP cable, the attenuation increase, the complexity of the filters and amplifiers, and so began the transition to CSS.
In the early 80-ies was developed by 3 such systems (in Europe, North America and Japan). Despite the same principles, the systems use different multiplexing ratios for different levels of hierarchies. Description of the joints of these interfaces and levels of multiplexing are given in recommendation G. 703.
The technology is quite simple – stream E1 consists of 32 proteins, each of which has a speed of 64 kbps, 2 service, which are transmitted to alarm, sync. The Association is due to the multiplexer PDH (typical NATEKS multiplexer, Raisecom, etc.). Without going into details, it looks like this — the incoming bits in streams are compressed to the number of times that is a multiple of the level of output and inserted at a certain position, then adds the service information of the higher flow.
When multiplexing of user threads in the multiplexers PDH uses a technique called bit-stuffing. This technique is used when the speed of the user flow is somewhat less than the velocity of the combined flow such problems can arise in the network consisting of a large number of multiplexers, despite all the efforts of centralized synchronization of network nodes (in nature nothing is perfect, including the perfectly synchronous network nodes). As a result, the multiplexer PDH periodically faced with a situation where he "lacks" a bit for presentation in the combined flow of one or the other of the user thread. In this case, the multiplexer simply inserts in the merged bit stream insert and notes this fact in the service bits of the combined frame. When demultiplexing the combined bit stream insert is removed from the user stream, which returns to its original state. The technique of bit-stuffing is used in both international and American versions of the PDH.
The physical layer PDH technology supports different types of cables: twisted pair, coaxial cable, fiber-optic cable. The main option for the subscriber access to channels T-1/E-1 is a cable of two twisted pairs with RJ-48.
Two pairs are required for the arrangement of duplex transmission mode data rate of 1,544/2,048 Mbit/s For the representation of signals are used:
in the channels of T-1 — potential B8ZS bipolar code;
channels E-1 — potential bipolar HDB3 code.
This technology has several drawbacks:
One of the major drawbacks is the complexity of the operations of multiplexing and demultiplexing of user data.The term "plesiochronous" used for this technology, says the reason for this phenomenon — lack of synchronism of the data flow when combining low-speed channels into higher-speed. Initially, an asynchronous approach to the transfer of frames has spawned the insertion of a bit or multiple bit synchronization between frames. The result is to retrieve the user data from the combined channel, you need to fully demultiplex the frames of the joint channel. For example, if you want to get data of a single user channel 64 Kbps with frames of channel TK, to make the de-multiplexing of these frames to the level of the frame T2, and then to the level of the T1 frames, and then demultiplex the T1 frames themselves.
The third drawback is too low by today's standards speeds the hierarchy PDH. Optical fibre cables allow to transmit data at speeds of several gigabits per second over a single fiber, which provides consolidation in a single cable of tens of thousands of custom channels, but it is property of the PDH technology is not implemented — it is the hierarchy of speeds, end with level 139 Mbps.
Therefore, the next stage of development of the SDH network is fully synchronous network.
System SDH (Synchronous Digital Hierarchy) provides the standard levels of the information structures, that is, a set of standard speeds. The basic speed level — STM-1 155,52 Mbit/s Digital velocity higher levels are determined by multiplying the flow rate of STM-1 are respectively 4, 16, 64, etc.: 622 Mbps (STM-4), 2.5 Gbps (STM-16), 10 Gbit/s (STM-64) and 40 Gbit/s (STM-256).
All information in the system is transmitted in the SDH containers. The container is a structured data transmitted in the system. If the system PDH generates traffic that you want to pass on the system of SDH, PDH structured data first into containers and then the container header is added and pointers, the result is a synchronous transport module STM-1. Over the network containers STM-1 transmitted in the SDH system at different levels (STM-n), but in all cases once formed STM-1 can only emerge with another vehicle module, e.g. a multiplexing transport modules.
Another important concept that is directly linked to the General understanding of the SDH technology is the concept of a virtual container VC.As a result of adding to the container traktove(route) header to get virtual container. Virtual containers are the ideological and technological communication with the container so that the container C-12 corresponds to a virtual container VC-12 (E1 stream), C-3 to VC-3 (transmission of the stream E3), C-4 container VC-4 (STM-1 stream).
Because of the low speed signals are multiplexed in PDH frame structure of high-speed SDH signals by byte multiplexing method, their location in the frame of high-speed signal is fixed and determined or, say, predictable. Therefore, low-speed SDH signal, for example 155 Mbit/s (STM-1) may be directly attached or allocated from the high-speed signal, such as 2.5 Gbps (STM-16). This simplifies the process of multiplexing and demultiplexing the signal and makes the SDH hierarchy especially suitable for high-speed fiber-optic transmission systems, having a high performance.
Since the adopted method for the synchronous multiplexing and flexible mapping structure, low-speed signals PDH (e.g., 2 Mbit/s) can also be multiplexed into the signal SDH (STM-N). Their location in the frame STM-N is also predictable. So tributary low-speed signal (up to signal DS-0, i.e. one time-slot PDH, 64 kbps) can be directly added or extracted from the signal STM-N. Note that this is not the same with the above process of adding/allocating a low-speed SDH signal to/from high-speed SDH signal. Here it refers to the direct addition of/selection tributario low-speed signal such as 2 Mbit/s, 34Мбит and 140Мбит/s to/from SDH signal. This eliminates the necessity of using a large amount of equipment multiplexing / demultiplexing (interconnected), which increases the reliability and reduces the likelihood of deterioration of signal quality, reduces the cost, power consumption and complexity of the equipment. Adding/selection services further simplified.
Spectral seal of the channels (eng. Wavelength-division multiplexing, WDM, literally-division multiplexing wavelength) — a technology that allows to simultaneously transmit several information channels over a single optical fiber at different carrier frequencies.
WDM technology can significantly increase channel capacity (by 2003 reached the speed of 10.72 Tbit/s, and by 2012, 20 Tbit/s), and it allows the use of already laid fiber-optic lines. Thanks to the WDM manages to organize a two-way multicast traffic over a single fiber. The advantage of DWDM systems is the ability to transmit high-speed signal for ultra-long distances without the use of intermediate points (without regeneration of the signal and intermediate amplifiers)
In the simplest case, each laser transmitter generates a signal at a certain frequency from the frequency plan. All these signals before inputted into the optical fiber, are combined by multiplexer (MUX). At the receiving end the signals are similarly separated by a demultiplexer (DEMUX). Here, as in SDH networks, a multiplexer is a key element. Signals come in on wavelengths of the customer equipment, and transmission occurs on the lengths of the corresponding frequency plan ITU DWDM.
Historically first arose dvuhventsovye WDM system working at the Central wavelengths of the second and third Windows of transparency of the quartz fiber (1310 and 1550 nm). The main advantage of such systems is that due to the large spectral separation is completely absent the influence of channels on each other. This method allows either to double the speed of transmission over a single optical fiber, or to organize a full-duplex communication.
Modern WDM systems based on standard frequency plan (ITU-T Rec. G. 692) can be divided into three groups:
Coarse WDM (Coarse WDM CWDM) system with a frequency channel spacing of at least 200 GHz, allowing to multiplex up to 18 channels.
(Currently used CWDM operate in the band from 1270нм to 1610нм, the space between the channels 20 nm (200Ghz), you can multiplex 16 spectral channels)
dense WDM (Dense WDM — DWDM) system with channel spacing of at least 100 GHz, allowing to multiplex up to 40 channels.
high-density WDM (Dense WDM High — HDWDM) system with channel spacing of 50 GHz or less, allowing one to multiplex at least 64 channels.
Frequency plan for CWDM systems is defined by the ITU standard G. 694.2. The scope of technologies CWDM — urban network with distance up to 50 km. the Advantage of this type of WDM systems is low (compared to other types) the cost of equipment as a result of smaller component requirements.
Frequency plan for DWDM systems defined by the ITU standard G. 694.1. The field of application — backbone network. This kind of WDM systems puts higher demands on the components than CWDM (spectral width of the radiation source, temperature stabilization of the source, etc.). The impetus for the rapid development of DWDM networks gave the appearance of inexpensive and efficient erbium fiber amplifiers (EDFA) operating in the range from 1525 to 1565 nm (third window of transparency of the quartz fiber).
We have found that SDH was originally designed for the use in ring topologies and already carries in itself a protection from loops — APS. But in Ethernet this issue and the question of shirokolistvenno storm stand upright.
The struggle began in 1985 with the invention of the STP. Then the RSTP, the early 2000s saw the light of MSTP. All of them have the same problem.
First, the convergence time, even in the modifications — this can be a few seconds. For voice and video streaming that is already quite noticeable.
Second, the safety. STP has no authentication and is vulnerable to attacks. And in addition, one Topology Change package can put an entire network.
Thirdly, the path between two adjacent switches can be through the root, then there will be sub-optimal.
Then it is not practical for large networks.
The change comes a new STP 802.1 aq. Humanly speaking, this SPB Shortest Path Bridging.
It converges faster, allowing you to recycle all of the links, in contrast to STP. Works in conjunction with protocols MMRP (Multi Mac Registration Protocol) and ISIS. In 2012 the standard was officially approved by IEEE.
It is fully compatible with STP family.
Another alternative — TRILL — Transparent Interconnection of Lots of Links. Uses the concept of routing Bridges. Something like a Link-State routing Protocol, but only for switching.
Although, this is for Layer 2 networks, very large scale data centers.
The next thing that comes to mind — or MLAG, MC-LAG Multi-Chassis Link Aggregation Group. That is a modification of the usual aggregate ports when you have the links start on one device and end on different. It not only has a redundant link, as in a normal LAG, but the backup device. The problem with it is that there are no universal standards and then someone on that much. To cross two vendors — will not work.
But all these really interesting topics I propose to discuss in the next releases. But today let's focus on another topic — ring topology Ethernet networks.
SDH originally designed to work in the ring and redundancy. So as not to appear to wish to implement this in a normal Ethernet? Appeared and made — ERPS — Ethernet Ring Protection Switching.
It is only on the rings. No full mesh or part mesh is not supported.
As far as I know its implementation is have the Cisco and juniper and other vendors. But because I work in Huawei, you can tell in detail about our proprietary Protocol RRPP, which is the same thing. It stands as Rapid Ring Protection Protocol.
Two of its main advantages is that the convergence time of less than 50 MS. Almost 2 orders of magnitude less than that of STP, and then it supports large L2 network. That is, if STP in the ring may be about 14 devices, RRPP, and with it the ERPS that are not restricted.
The principle of operation.
Take the example of the network scale of the city. It is scattered on the inner perimeter 25 of the aggregation switches. It would be ideal if the scheme of the star from the point of view of stability, but it is not rational, logical ring. Thus we get a fall protection link and failure of a single device.
Of these 25 switches select one main — master. This switch becomes responsible for the integrity of the ring and no loop.
Each switch has two ports, one East, another West, respectively the ports of different switches need to look at each other East to West.
The Master likewise, and one of them he chooses active, and the second blocks. Turns out just such a chain of 25 switches. And the frame with the 25th to get to the top, will have to go through 24, 23, 22, etc., that is, through the whole chain.
Regularly sent master Hello messages. If the blocked port is not received this package within Fail_таймера or information received about the drop link, RRPP unlocks it.
One large ring can be the petals — the slave ring. That is, if, for example, two switches serve any district, the petal starts on one and ends on another. Inside it there are the same laws of redundancy. That's kind of a Daisy.
With the real use of ERPS I have not encountered. At the same time, RRPP quite used in operators ' networks in those areas where all the equipment of Huawei.
Article based on information from habrahabr.ru
In the new issue of discuss
1) SDH/PDH (Synchronous Digital Hierarchy/Plesiochronous Digital Hierarchy)
2) System of WDM channels in optical networks
3) Technology of redundancy to Ethernet network. Alternatives STP — ERPS, RRPP.
Download podcast.
Nevoshedshie podcast topics that we found interesting, we made Overtime.
SDH/PDH
I'm sure every telecommunications engineer knows or at least heard the name as a stream E1. But even the most advanced heard of the Nyquist theorem, the sampling frequency and PCM. The history of the development of networks PDH ( Plesiochronous Digital Hierarchy) began with the fact that by the mid 70s it became clear that further frequency seal is impossible in a single physical communications channel, it is limited to PP cable, the attenuation increase, the complexity of the filters and amplifiers, and so began the transition to CSS.
In the early 80-ies was developed by 3 such systems (in Europe, North America and Japan). Despite the same principles, the systems use different multiplexing ratios for different levels of hierarchies. Description of the joints of these interfaces and levels of multiplexing are given in recommendation G. 703.
The technology is quite simple – stream E1 consists of 32 proteins, each of which has a speed of 64 kbps, 2 service, which are transmitted to alarm, sync. The Association is due to the multiplexer PDH (typical NATEKS multiplexer, Raisecom, etc.). Without going into details, it looks like this — the incoming bits in streams are compressed to the number of times that is a multiple of the level of output and inserted at a certain position, then adds the service information of the higher flow.
When multiplexing of user threads in the multiplexers PDH uses a technique called bit-stuffing. This technique is used when the speed of the user flow is somewhat less than the velocity of the combined flow such problems can arise in the network consisting of a large number of multiplexers, despite all the efforts of centralized synchronization of network nodes (in nature nothing is perfect, including the perfectly synchronous network nodes). As a result, the multiplexer PDH periodically faced with a situation where he "lacks" a bit for presentation in the combined flow of one or the other of the user thread. In this case, the multiplexer simply inserts in the merged bit stream insert and notes this fact in the service bits of the combined frame. When demultiplexing the combined bit stream insert is removed from the user stream, which returns to its original state. The technique of bit-stuffing is used in both international and American versions of the PDH.
The physical layer PDH technology supports different types of cables: twisted pair, coaxial cable, fiber-optic cable. The main option for the subscriber access to channels T-1/E-1 is a cable of two twisted pairs with RJ-48.
Two pairs are required for the arrangement of duplex transmission mode data rate of 1,544/2,048 Mbit/s For the representation of signals are used:
in the channels of T-1 — potential B8ZS bipolar code;
channels E-1 — potential bipolar HDB3 code.
This technology has several drawbacks:
One of the major drawbacks is the complexity of the operations of multiplexing and demultiplexing of user data.The term "plesiochronous" used for this technology, says the reason for this phenomenon — lack of synchronism of the data flow when combining low-speed channels into higher-speed. Initially, an asynchronous approach to the transfer of frames has spawned the insertion of a bit or multiple bit synchronization between frames. The result is to retrieve the user data from the combined channel, you need to fully demultiplex the frames of the joint channel. For example, if you want to get data of a single user channel 64 Kbps with frames of channel TK, to make the de-multiplexing of these frames to the level of the frame T2, and then to the level of the T1 frames, and then demultiplex the T1 frames themselves.
The third drawback is too low by today's standards speeds the hierarchy PDH. Optical fibre cables allow to transmit data at speeds of several gigabits per second over a single fiber, which provides consolidation in a single cable of tens of thousands of custom channels, but it is property of the PDH technology is not implemented — it is the hierarchy of speeds, end with level 139 Mbps.
Therefore, the next stage of development of the SDH network is fully synchronous network.
System SDH (Synchronous Digital Hierarchy) provides the standard levels of the information structures, that is, a set of standard speeds. The basic speed level — STM-1 155,52 Mbit/s Digital velocity higher levels are determined by multiplying the flow rate of STM-1 are respectively 4, 16, 64, etc.: 622 Mbps (STM-4), 2.5 Gbps (STM-16), 10 Gbit/s (STM-64) and 40 Gbit/s (STM-256).
All information in the system is transmitted in the SDH containers. The container is a structured data transmitted in the system. If the system PDH generates traffic that you want to pass on the system of SDH, PDH structured data first into containers and then the container header is added and pointers, the result is a synchronous transport module STM-1. Over the network containers STM-1 transmitted in the SDH system at different levels (STM-n), but in all cases once formed STM-1 can only emerge with another vehicle module, e.g. a multiplexing transport modules.
Another important concept that is directly linked to the General understanding of the SDH technology is the concept of a virtual container VC.As a result of adding to the container traktove(route) header to get virtual container. Virtual containers are the ideological and technological communication with the container so that the container C-12 corresponds to a virtual container VC-12 (E1 stream), C-3 to VC-3 (transmission of the stream E3), C-4 container VC-4 (STM-1 stream).
Because of the low speed signals are multiplexed in PDH frame structure of high-speed SDH signals by byte multiplexing method, their location in the frame of high-speed signal is fixed and determined or, say, predictable. Therefore, low-speed SDH signal, for example 155 Mbit/s (STM-1) may be directly attached or allocated from the high-speed signal, such as 2.5 Gbps (STM-16). This simplifies the process of multiplexing and demultiplexing the signal and makes the SDH hierarchy especially suitable for high-speed fiber-optic transmission systems, having a high performance.
Since the adopted method for the synchronous multiplexing and flexible mapping structure, low-speed signals PDH (e.g., 2 Mbit/s) can also be multiplexed into the signal SDH (STM-N). Their location in the frame STM-N is also predictable. So tributary low-speed signal (up to signal DS-0, i.e. one time-slot PDH, 64 kbps) can be directly added or extracted from the signal STM-N. Note that this is not the same with the above process of adding/allocating a low-speed SDH signal to/from high-speed SDH signal. Here it refers to the direct addition of/selection tributario low-speed signal such as 2 Mbit/s, 34Мбит and 140Мбит/s to/from SDH signal. This eliminates the necessity of using a large amount of equipment multiplexing / demultiplexing (interconnected), which increases the reliability and reduces the likelihood of deterioration of signal quality, reduces the cost, power consumption and complexity of the equipment. Adding/selection services further simplified.
Optical networks and systems for spectral multiplexing
Spectral seal of the channels (eng. Wavelength-division multiplexing, WDM, literally-division multiplexing wavelength) — a technology that allows to simultaneously transmit several information channels over a single optical fiber at different carrier frequencies.
WDM technology can significantly increase channel capacity (by 2003 reached the speed of 10.72 Tbit/s, and by 2012, 20 Tbit/s), and it allows the use of already laid fiber-optic lines. Thanks to the WDM manages to organize a two-way multicast traffic over a single fiber. The advantage of DWDM systems is the ability to transmit high-speed signal for ultra-long distances without the use of intermediate points (without regeneration of the signal and intermediate amplifiers)
In the simplest case, each laser transmitter generates a signal at a certain frequency from the frequency plan. All these signals before inputted into the optical fiber, are combined by multiplexer (MUX). At the receiving end the signals are similarly separated by a demultiplexer (DEMUX). Here, as in SDH networks, a multiplexer is a key element. Signals come in on wavelengths of the customer equipment, and transmission occurs on the lengths of the corresponding frequency plan ITU DWDM.
Historically first arose dvuhventsovye WDM system working at the Central wavelengths of the second and third Windows of transparency of the quartz fiber (1310 and 1550 nm). The main advantage of such systems is that due to the large spectral separation is completely absent the influence of channels on each other. This method allows either to double the speed of transmission over a single optical fiber, or to organize a full-duplex communication.
Modern WDM systems based on standard frequency plan (ITU-T Rec. G. 692) can be divided into three groups:
Coarse WDM (Coarse WDM CWDM) system with a frequency channel spacing of at least 200 GHz, allowing to multiplex up to 18 channels.
(Currently used CWDM operate in the band from 1270нм to 1610нм, the space between the channels 20 nm (200Ghz), you can multiplex 16 spectral channels)
dense WDM (Dense WDM — DWDM) system with channel spacing of at least 100 GHz, allowing to multiplex up to 40 channels.
high-density WDM (Dense WDM High — HDWDM) system with channel spacing of 50 GHz or less, allowing one to multiplex at least 64 channels.
Frequency plan for CWDM systems is defined by the ITU standard G. 694.2. The scope of technologies CWDM — urban network with distance up to 50 km. the Advantage of this type of WDM systems is low (compared to other types) the cost of equipment as a result of smaller component requirements.
Frequency plan for DWDM systems defined by the ITU standard G. 694.1. The field of application — backbone network. This kind of WDM systems puts higher demands on the components than CWDM (spectral width of the radiation source, temperature stabilization of the source, etc.). The impetus for the rapid development of DWDM networks gave the appearance of inexpensive and efficient erbium fiber amplifiers (EDFA) operating in the range from 1525 to 1565 nm (third window of transparency of the quartz fiber).
Technology rezervirovaniya Ethernet networks
We have found that SDH was originally designed for the use in ring topologies and already carries in itself a protection from loops — APS. But in Ethernet this issue and the question of shirokolistvenno storm stand upright.
The struggle began in 1985 with the invention of the STP. Then the RSTP, the early 2000s saw the light of MSTP. All of them have the same problem.
First, the convergence time, even in the modifications — this can be a few seconds. For voice and video streaming that is already quite noticeable.
Second, the safety. STP has no authentication and is vulnerable to attacks. And in addition, one Topology Change package can put an entire network.
Thirdly, the path between two adjacent switches can be through the root, then there will be sub-optimal.
Then it is not practical for large networks.
The change comes a new STP 802.1 aq. Humanly speaking, this SPB Shortest Path Bridging.
It converges faster, allowing you to recycle all of the links, in contrast to STP. Works in conjunction with protocols MMRP (Multi Mac Registration Protocol) and ISIS. In 2012 the standard was officially approved by IEEE.
It is fully compatible with STP family.
Another alternative — TRILL — Transparent Interconnection of Lots of Links. Uses the concept of routing Bridges. Something like a Link-State routing Protocol, but only for switching.
Although, this is for Layer 2 networks, very large scale data centers.
The next thing that comes to mind — or MLAG, MC-LAG Multi-Chassis Link Aggregation Group. That is a modification of the usual aggregate ports when you have the links start on one device and end on different. It not only has a redundant link, as in a normal LAG, but the backup device. The problem with it is that there are no universal standards and then someone on that much. To cross two vendors — will not work.
But all these really interesting topics I propose to discuss in the next releases. But today let's focus on another topic — ring topology Ethernet networks.
SDH originally designed to work in the ring and redundancy. So as not to appear to wish to implement this in a normal Ethernet? Appeared and made — ERPS — Ethernet Ring Protection Switching.
It is only on the rings. No full mesh or part mesh is not supported.
As far as I know its implementation is have the Cisco and juniper and other vendors. But because I work in Huawei, you can tell in detail about our proprietary Protocol RRPP, which is the same thing. It stands as Rapid Ring Protection Protocol.
Two of its main advantages is that the convergence time of less than 50 MS. Almost 2 orders of magnitude less than that of STP, and then it supports large L2 network. That is, if STP in the ring may be about 14 devices, RRPP, and with it the ERPS that are not restricted.
The principle of operation.
Take the example of the network scale of the city. It is scattered on the inner perimeter 25 of the aggregation switches. It would be ideal if the scheme of the star from the point of view of stability, but it is not rational, logical ring. Thus we get a fall protection link and failure of a single device.
Of these 25 switches select one main — master. This switch becomes responsible for the integrity of the ring and no loop.
Each switch has two ports, one East, another West, respectively the ports of different switches need to look at each other East to West.
The Master likewise, and one of them he chooses active, and the second blocks. Turns out just such a chain of 25 switches. And the frame with the 25th to get to the top, will have to go through 24, 23, 22, etc., that is, through the whole chain.
Regularly sent master Hello messages. If the blocked port is not received this package within Fail_таймера or information received about the drop link, RRPP unlocks it.
One large ring can be the petals — the slave ring. That is, if, for example, two switches serve any district, the petal starts on one and ends on another. Inside it there are the same laws of redundancy. That's kind of a Daisy.
With the real use of ERPS I have not encountered. At the same time, RRPP quite used in operators ' networks in those areas where all the equipment of Huawei.
Comments
Post a Comment