Comparison of sdn and gmpls control techniques in optical transport networks by Name: Muhammad Atif Yaqub Reg #: ms(EE)-sp13-008 In Partial Fulfillment of Requirements
Formation of Label switched path in GMPLS
Download 193.35 Kb.
|
- Navigate this page:
- Formation of LSP in GMPLS network – Fig.2
- 1.3.3 Mechanism for data Transmission variety
- 1.3.4 Structuring of GMPLS
- 1.3.4.2 Duplex Label switch path
- 1.3.5 Scaling 1.3.5.1 Forwarding Adjacency Label switch path
- Figure 3. Forwarding Adjacency
- 1.3.5.2 Categorized label switch path
- Fig.4 Hierarchy of network
- Figure 5. Hierarchical LSPs
- Figure 6. Fault-Management Process in GMPLS
- Figure 7. Protection techniques used in GMPLS
- 1.3.7.1 Numberless connections
- 1.4 GMPLS unresolved problems The GMPLS protocols are not standardized and there are many issues which need solution, which are mentioned below. 1.4.1 Safety
- 1.4.2 Connectivity with other networks
- 1.4.3 Stability of network
- Chapter-2
- SDN Architecture 2.2.1 Data plane
- 2.2.5 Network specification
- 2.2.6 Advantages of Decoupling of control and data planes
- 2.2.7 Interaction of SDN layers 2.2.7.1 Software defined networking core controller
- 2.2.7.2 Use of vendor based Network Software
- 2.2.7.3 Use of Network Management System
- 2.2.8 SDN layers 2.2.8.1 Infrastructure layer
1.3.2 Formation of Label switched path in GMPLS Now we have an example to create LSP path from source to destination that has packet switching at start. The TDM network used having SONET systems and after that DWDM system used network two networks. Each layer has its different Labels. The source is LSR1 and destination is LSR4. In the initial layer from source we have OC-12 link between packet switching layer to TDM network. After that It is network of TDM WITH OC-48 capacity. Then multiple OC-48 lines combine in the OXC1. 
Formation of LSP in GMPLS network – Fig.2 This cross connect is connecting multiple wavelengths of OC-192 data rate. Now we want to create LSP from LSR1 to last LSR4. Now it is very important that the LSPs should be established such that 1st of all the downstream path LSPs should be established. These LSPs should be tunneled with the higher level LSPs to have complete path between source to destication. As per Fig.2 the LSP between LSR1 and LSR4 will have to carry all the labels of in between network (TDM + WDM). First of all a Request for label is forwarded downwards to the network from upstream node as the downward nodes will share the LSP first. In this Fig.2, the TDM system DSCi will forward the request to OXC1 mentioning the receiver which is DSCe. This message received by OXC1 causes an action of one label switch path creation between OXC1 and OXC2. With creation of this LSP, the main LSP between DSCi and DSCe is made which is called LSPtDi. What information is carried in this request for label creation is demand for creation of generalized label for the label switch path. It carries the type of label switch path (protocol of the layer and the what kind of payload it will carry for example VC-4, VC-3. Some special properties like type of signal, automatic protection mechanism, duplex Label switch path and the proposed labels are also available in this label request. The downward system sends RESV/label mapping message which has one generalized label and can also have many generalized labels. As the source LSR gets these generalized labels from downward systems then it makes a label switch path to the destination LSR with RSVP/PATH message. The actions performed in our case are as under.
1.3.3 Mechanism for data Transmission variety As we know that Multi-protocol label switching systems have the intelligence of identifying the packet headers and then performing required actions of data forwarding as per the information carried by the headers including labels and the destination system information. The data and control planes are separated when the labels are exchanges between systems. Generalized multi-protocol label switching uses the same kind of approach that mentioned above. In it the control and data planes are separated physically or logically. We can have the option to have different path for the control messages other than the data path. We may have separate SDH signal for control. Generalized multi-protocol label switching will not give orders that the control messages should be transmitted in this way. The cost is involved in choosing any one option for transmission of the control information. It is also not feasible to have just one SDH link to transport the control information in a scenario when the systems are far apart in a SDH ring. A good idea for use of the control messages in the existing network without involving any cost is to transmit the control information through DCC channel of the SDH signal. These overhead bytes will be used in each SDH link of joining nodes in the ring, so we will have no single point of failure. This arrangement does not cause any bandwidth wastage as it may happen if the control channel uses the space of data plane. 1.3.4 Structuring of GMPLS In GMPLS the label switch paths are created from the last access side. These paths may also require the LSP formation of intermediate systems that may have different structures upto the destination node. Each of the intermediate devices has unique qualities which GMPLS has to address so that LSP creation is fast and efficient. In order to have insight about these differences following methods needs to be clear. 1.3.4.1 Proposed Label There is one possibility available that we discussed earlier that there is an option with the upstream system that it can propose a label to the next downward system to speed up the process of LSP paths formation. But the downward system can reject this label and can create its own made LSP. This is a lengthy process which may take too much time. If the switching capacity of TDM system is e.g 622 Mbps then it will become long process to allocate labels for multiple traffic. Proposed labels are used to speed up the process of making LSP between source to destination. The SDH system is our example does not wait for the labels from downward OXC layers and has its own label. It reduces the time consumption if making the LSP end to end. But if the downward system not willing to use the proposed label then it can use its own chosen label instead of proposed one. 1.3.4.2 Duplex Label switch path As we know that SDH TDM systems use ring topology for provisioning of the protection path in case of fiber cut conditions. Like TDM protection we have to protect the label switch path to have the same protection of the path in case of any path failure. It can be done by having two simplex LSPs. One LSP works as protection to the other. Generalized multi-protocol label switching has the option for use of duplex LSPs with the signaling protocols like RSVP/PATH. It becomes useful to avoid unnecessary sharing of control information, route formation in optical networks. 1.3.5 Scaling 1.3.5.1 Forwarding Adjacency Label switch path Forwarding Adjacency label switch path is used in GMPLS to get data of other LSPs. It os contructed among 02 GMPLS systems as has its own characteristics and may be added in routing protocols like OSPF and IS-IS. It can be used in the link state algorithm of routing protocols to transport information of other label switch paths. It will ultimately lower the load on database and time consumption in label search from the table is reduced. These Forwarding adjacency label switch paths can be a number of unnumbered data which may be attached with the other links information. Figure 3. Forwarding Adjacency 
In above Fig.3, we observe that the LSPtdm is the PSP which is connecting 02packet switched networks. It makes the sense of just one LSP path between two packet networks instead of combination of different networks. 1.3.5.2 Categorized label switch path As we know the network categorization in access, METRO and WDM to carry the traffic from source to destination, therefore the LSP hierarchy should be used for better us eof the network. The data from source to destination will pass through different LSPs of different networks. The TDM systems have fixed bandwidth for each LSP. Fig.4 Hierarchy of network 
Therefore it is not feasible to use a complete 10 G wavelength for just a data of 10 Mbps. For this purpose it is better to aggregate the lower rate signals into higher rate for better use of resources. In Generalized multi-protocol label switching the LSPs are grouped together to aggregate the traffic in main LSP having large capacity. The below Figure shows the architecture. Figure 5. Hierarchical LSPs 
1.3.5.3 Link Aggregations Since the optical networks are supposed to have hundreds of the optical fiber lines between 02 adjacent nodes and each fiber may also have many wavelengths in it. This will give huge bandwidth availability for the Generalized Multi-protocol label switching network. In order to efficiently use the network resources it is better to use bundling of the network resources. With this approach of bundling all this information is shared with routing protocols. With this bundling we may have lost some data. But this method has great advantage that the load on routing and signaling protocols is reduced too much extent. This bundle carrying heavy data has only one control channel for exchange between systems. The bundling can be done of point to point as well as label switch paths, which are shared with the routing protocols. We have some limitations for the use of bundling the links. All these mentioned below.
1.3.6 Consistency This is very important feature of this control plane is its capability for dynamic restoration of faults in the network. It is important that whenever a fault occurs in the network it must be separated from the network and restoration must be done after separating from network. The label switch paths that are already tunneled with other label switch paths must have such kind of consistency. The process of the faults resolution is shown in below Figure.6 Figure 6. Fault-Management Process in GMPLS 
Generalized multi-protocol label switching gives protection of the faulty channels among both joining systems which is called span protection while source to destination protection is path protection. Routing protocols broadcast all the information of this kind of protection with the whole network and the protection route is calculated. After its calculation a signaling is used with the RSVP and CR-LDP protocols. The secondary path is calculated and when the fault occurs the signaling protocols switch the signals to protection paths. Whenever a fault occurs the dynamic fault recovery is achieved with shifting traffic to protection path. The 02 dynamic procedures of restoration are line and path. In line protection alternate path is selected at the system in between source and destination, while in path receiver the restoration is started from the source system. The below mentioned Figure.7 shows that system 1 starts for new path. Since much large time taken for switching to protection path but it is bandwidth efficient as no media is already allocated for this protected path. Figure 7. Protection techniques used in GMPLS 
1.3.7 Efficient Use of available resources It needs some kind of optimization when we talk about use of TDM network with IP-based control plane. It needs to be discussed the major tendency of TDM network to use IP addresses of packet network. 1.3.7.1 Numberless connections In the time division and wavelength switching systems we do not use the IP addresses for each such traffic. It is called numberless connectivity technique. The reasons behind this technique are as below.
These links which are not having any numbers are identified by the router number and some information of the link number. The other necessary information is carried by the signaling protocols as done with the links having identified by IP addresses. 1.4 GMPLS unresolved problems The GMPLS protocols are not standardized and there are many issues which need solution, which are mentioned below. 1.4.1 Safety As we know that the already used packet based systems use the IP address to identify the traffic and the packet header has the complete information about originator and the final end node IP addresses. These IP addresses are exclusive in the whole world and have better security phenomena. But the labels which are used in GMPLS and MPLS are just limited to the internal network and have nothing to do with the external network. Therefore these labels cannot be used for security policy to give access control to the systems. In order to have security in these networks it is required to use techniques like that used in ATM and X.25. 1.4.2 Connectivity with other networks GMPLS success is due to its property for connectivity and working with other network techniques like ATM and Frame relay. Such connectivity with other networks results in sharing of data and control information of 02 ATM networks when GMPLS used in between 02 ATM networks. Following issues are observed in making such connectivity.
1.4.3 Stability of network Whenever a new network element id placed in network or someone removed, the much large control data is shared by NEs with each other than that of old packet based systems. Generic multi-protocol label switching performs TE which has information about traffic parameters and using important protocols like Link Management Protocol. As per my study this kind of network consumes large time to have stability than those old type packet-based systems. 1.4.4 NMS Old Internet protocol based systems use only the IP address to have reachability to that systems, but when we talk about Generalized MPLS it has to deal with thousands of Label switch paths to check for traffic monitoring and quality of services. It makes the NMS more difficult to use in such networks. Chapter-2 Software defined networking
Software defined networking is a technique that is required to be used in the latest computer networking centers in the world. This technique performs separation of data and control planes which provide common control scenario in the network. The control plane is found in the software defined networking controller. The path for the packet movement is determined with this controller and then shares this information with all network elements. It all special applications that run with the application layer. These applications can run dynamically or through the management system. This dynamic nature of control processes enables the SDN network to provide fast service extensions. This control plane minimizes the hardware cost of the optical equipment with the use of control information in software that runs over common hardware instead of working on optical chips.
2.2.1 Data plane Data plane is the movement of data between the routers and switches through the interface ports. The switches and routers can be connected with high data rate lines like in Gbps. Two important information databases are available in every equipment in the network. The 1st is the database to differentiate the packets and the other forwards the packet to its destination through interface ports. 2.2.2 Control plane It contains special procedures and methods that are used to find the path to route the packets from source to destination. In the switches the control plane checks if the packet MAC addresses matches the database then it forwards to the destination, otherwise if the database not contains the information of packet then it sends the packet to all switches. Routers use their own routing protocols for calculating the path. Fig-8 SDN Architecture 
Following are four important parts of this architecture.
The common control nature has many benefits like it is very easy to change network structure with high level software versions. It becomes quite difficult with system dependent software. The common control works immediately whenever new network changes are observed. The control plane has complete information of the network changes and then acts accordingly. 2.2.3 Data forwarding Data forwarding is very important and unique feature in SDN. The forwarding information does not carry any information of the source hardware and hides its information. 2.2.4 Data distribution In SDN the data distribution control is also common. The network operating system carried this distribution plane. It performs downloading network distribution information of the network elements and gets all necessary information about the network elements and connectivity between them for proper distribution. 2.2.5 Network specification Network specification is done in network operating system with a required behavior of network, which is supposed to be not done by control plane. It is a virtual technique in which the specific configurations are mapped on physical network setup basis. Below Fig-9 show the detail.
Following are important benefits of using the separate control and data planes.
2.2.7 Interaction of SDN layers 2.2.7.1 Software defined networking core controller It is the main part in the operation and maintenance layers of the SDN and keeps all the network related information with it. It performs main operations to provide required actions in the network. 2.2.7.2 Use of vendor based Network Software This is software which is made by the equipment manufacturer. It is installed in the system which manages the network. The user gets it accessed from another computer and works on it through some commands. The specific vendor based touch of this software asks user to use specific commands which are only based on the vendor equipment. This software is responsible to manage the whole network either single layered or multi layered. 2.2.7.3 Use of Network Management System Network Management system is used for some low level configurations of the transport systems. It is used for making the arrangement so that the network elements are online and accessible for further operations. This software makes configurations of transport nodes regarding their ID, IP and the link establishment between the network elements. This software is also vendor based and each vendor used its own specific network management software. Like one vendor uses U31 software for this purpose. This is GUI based software which enables users to have performed all operations efficiently.
This is the layer which is related to hardware equipment installed in the network. These equipment are not having any control information in them as SDN has the centralized control plane. The data plane is not having any intelligence in it and all is available in control plane. 
Fig-10 SDN in (a) planes (b) layers (c) architecture The intelligence is completely available in the network control and application plane. It is seen that these networks are established along with Open Flow standards so that different kinds of networks elements can be used together for data forwarding. The old transmission equipment like SDH are not having this flexibility as many built-in hardware requirements must be fulfilled in these equipment. The below Fig-11 shows the SDN control and data devices. 
Fig-11 Open flow enabled SDN devices In the above Fig-11 the SDN elements are shown. One is the control element and other is forwarding one. The control plane has a match table which has parts of rules, actions and statistics of these data packets. The table shows the process through which the packets have to move. We have the option to choose one of the rule out of these mentioned for a packet to distinguish. The actions with the packet may be sending to the out ports, send to control plane, discard, forward to old forwarding line which is not open flow. Directory: uploads -> Z2015 uploads -> The Case Against the nypd’s Quota-Driven ‘Broken Windows’ Policing What Is It? uploads -> For want of a nail uploads -> Ownership Models There are two different types of ownership models uploads -> Police Militarization uploads -> City of rising star regular meting thursday, july 14, 2016 uploads -> Township of winfield uploads -> Stop Militarizing Law Enforcement Act of 2014 uploads -> The Black Panther Party’s Ten-Point Program What We Want Now! Z2015 -> Deaths in 2015. Names are reported under the date of death. Names under each date are reported in alphabetical order by family name orpseudonym. Deaths of non-human organisms are also reported here if they first have their own Wikipedia Z2015 -> A project study report Download 193.35 Kb. Share with your friends:
|
