Issues With The Packet Loss During Transmission

Published Date: 02 Nov 2017

INTRODUCTION

Internet is the global system of interconnected computer networks that are interlinked and helps in providing information from one computer to the other. The data is passed in the form of packets that contain all the information and via the several network layers and packets are encoded and sent to the destination address. The information that the wrapper (a system that filters the network) contains helps the destination computer know what type of data is in the packet and where the data arrived to the final destination. Hence, packets play a very important role in the internet activity. Several internet protocols help in the transmission of packets from the source to the destination without any complications being made since every address should be made reachable from every other address which is supposed to be the basic goal of the Internet.

There are several instances where-in packets which are sent from a source are lost due to miscellaneous reasons and the packets disappear into black holes and fail to reach the destination. The main idea is to find such reachability issues over the global internet and try to locate the sources of the problems.

Hubble is the real-time tool that has been used to find the reachability problems within the network between any two peers. The packets being delivered or undelivered are monitored by performing several operations using the Hubble by monitoring feeds regularly and by watching the prefix movement of the physical path.

Packet loss can play a vital role in losing information because it could create severe mutilation of the data received and all the over-burdened network nodes play an important role as failures are noticeable in all the types of the digital communications.

Packet loss can be fixed by identifying the location of the failure. It is a common tendency that sometimes when the network congestion plays a vital role, the router switch throws away the packets traversing towards their destination from specific node carrying vital information, resulting in the loss of data.

IMPORTANT IDEAS

The main idea that is being dealt is regarding the message passing using the packets and where the packets are lost when the they are undelivered. It could be the Blackhole or some physical path being lost due to dis-connectivity in the transmission mediums like the cables, wires, etc.

The authors Ethan, Harsha, John, Arvind and David in their paper discuss about the routes that the lost packets follow. The Hubble is a tool that helps to discover and track the problems that the packets face during their path like the physical medium, the low internet connectivity speeds, the destination traffic the Autonomous System (AS) affects or is it the problem/defect of a router that is transmitting the packets. They build a system that provides information about such reachability issues and that system helps in diagnosing problems. It is found out that the real-time status of router data and the direct BGP feeds that every Autonomous System takes, it is not possible to pinpoint the reason for the failure, without any access to the complete topologies. Hence, after the operation of Hubble is successfully run, it is found out that the trace-routes have less than 90% or stages to reach the origin of the Autonomous System for the prefix. It is clear that the failure rate of 10% is a substantial number that can have various problems like having multiple problems with a single route or cascading of processes at routers.

The operations that Hubble performed are declared to be successful as its discovery rate is pinpointed to a massive 85%, 73% of the events are identifies and a whopping 96% of the events were discovered in the first hour of its operation. Even though most of the cases were the partial reachability issues, many packets were still left undelivered because of reasons being the physical medium like the cables, wires, routers, etc.

The paper concludes that the constant BGP feeds alone do not suffice to help us know the problems that arise.

In the paper produced by the authors Yiyi Huang, Nick Feamster, Renata Teixeira, they discuss about the paths that have to be selected to monitor, detect and confirm the existence of a failure that occurs in a network.

The paper deals with the network’s internal characteristics using the data and information retrieved at the end point from the packets while they were being traversed from the source to the destination. Using the Network Tomography technique, it has been identified that detection of the failure is evident but finding the location of the fault is tough. The packets that have been sent from the source were lost after reaching a point, where an alarm was raised and traps were made to find such types of failures. But, running pings and trace-routes could not help in finding out the location of the fault, but, one task of knowing the failure has been achieved.

The active monitoring techniques that are used for the quick detection of the failures requires probing paths that are frequent, but, since in real networks, the paths are large and vast, the detection tradeoff is minimized as it is tough to find the location of the failure from huge networks. Also, the common notion that the conventional network tomography algorithms presume are supposed to be the link failures, traversing paths that links observe and the detection time. It is declared that the short failures are the most difficult to determine since not all the monitoring systems could identify the failures accurately unless it is an Autonomous System like Hubble. Hence, it is clearly evident that in the real-networks, when the link failure occurs, the loss of the transient packets may be provoked to raise false alarms due to re-routing technique and create a wrong and inconsistent overview of the topology.

The paper concludes that the effectiveness of the network tomography for finding the links using the VINI test-bed is high to identify the loss of the packets but not the location where the loss has occurred.

In the paper produced by the authors Italo Cunha, Renata Teixeira, Darryl Veitch and Chrisophe Diot, they discuss about the main idea of using the trace-routes and its effectiveness in finding the path changes using various classical methods to optimize the probing capacity is dealt.

A path mapping system named DTRACK is used in this approach to find the path sampling rate allocation. The other path mapping system method named Trace-Tree mapping has not been preferred since it does not use the prediction techniques to figure out the paths like the BGP messages. Probing all the paths that the packets traverse is a tedious job since there are many paths that do not change with the same frequency while on the hunt for the paths that change. Hence, detection time is probed to be large.

The DTRACK system operates independently in finding out the path the packets take by carefully monitoring the current route, the last change that happened in the route and the time based metrics measures the timescale for an event-time.

The paper is precise to the point that the path tracking strategy occurs in two steps namely the path change detection and the path remapping. It is understood that both the steps involve the DTRACK to employ a trace-route and it detects the change that occurs and then it re-maps the trace-route to ensure the accuracy in the route management.

The paper concludes that the path of the packets that are lost could be found, but, not the location where it occurs.

In the paper produced by the authors Fernando Sanchez and Zhenhai Duan, the most affected parts of the network when packets are undelivered and lost are discussed.

The Border Gateway Protocol is the most affected protocol since it has the weakest security for the prefixes of the packets, also, the new filtering scheme named Region-Based BGP is introduced which helps in the improvement of the BGP prefix attacks as it can detect hijacking attacks by performing the partition of internet into various regions so that no Autonomous System could have any access to the prefixes of the BGP. The other technique named Secure-BGP (S-BGP) is the most comprehensive security enhancement for the BGP attacks as it helps in preventing both types of BGP Prefix attacks. But, S-BGP is not widely used due to its high management costs and also the processing of the packets is a tedious job along with the cost.

In this paper, it is declared that Hubble is a large-scale system that is used to carefully adhere and monitor the reachability problems on the Internet, specifying the black holes that exists and black holes being the most important places where the BGP routes exist to the corresponding network prefixes and helps in identifying the BGP path changes.

The BGP prefix hijacking attacks generally either the Origin Spoofing form or the Path Spoofing form. The path spoofing form generally blindfolds the network by considering the wrong path i.e. a false path is pretended to be the actual route to the prefix of an Autonomous System.

The BGP feeds are always reachability issues in the real-time by carefully monitoring the anomalies of the Autonomous Systems and also that BGP acts as a regulator to establish the routes for the internet traffic, it is advised that the packets movement is optimized to a highly locatable trace-route.

3 CONCLUSION

Given a large amount of prefixes that are affected by various path inflation and trace-routes issues, the continuous probing and identification of the reachability problems in order to identify over a real-time network and scale, it is concluded that the popular BGP feeds alone cannot suffice in helping to discover the complications that arrive while various tasks are being performed. The reachability problems that arise with respect to the distinct prefixes are more relating to the physical path like the cables, cords, wires, etc that are used for the transmission of the packets.

Using the various spoofing techniques for finding the alternative routes, using various methods to implement the transmission issue to improve the global reachability, diagnosing the basic performance would be ideal and preferably the extent of the bandwidth and the internet speeds would help the packets reach the destinations without any hindrance.

Also, the ISPs spoofing of the traffic and the filtering could help Hubble to provide better results in finding the location where the fault has been detected. Though many several ideas have been procured to deliver to operators the information required for improving the global reachability, helps the system in identifying, diagnosing and processing the BGP streams.

It has been observed that majority of the problems relating to the reachability are found to be more than what could be expected and the Hubble tool has been a fore-runner over the DTRACK tool to successfully identify the reachability issue using various paths and by using the tunneling approach, we could utilize the Hubble data which could help in identifying and increasing the number of the stages that are defined for various purposes.

Determining the smallest number of subset of paths helps in indentifying and short-listing different combinations of network paths from the huge lot of network with regular bandwidth speed and since traffic over the links would intervene the packets, dumping of packets during bottleneck. Packet loss is associated with the Quality of Service (QoS) considerations, hence, since the regular internet connectivity is generally associated with the TCP/IP protocols, the packet loss rate is below 0.3% (3 packets are lost in every 1000 packets) and this can be tolerated, anything more should be adhered and addressed.