Improving the efficiency of flow table updates in software defined networks

مقدم أطروحة جامعية

Atiyyah, Aya Adnan Muhammad

مشرف أطروحة جامعية

al-Majali, Sufyan

الجامعة

جامعة الأميرة سمية للتكنولوجيا

الكلية

كلية الملك الحسين لعلوم الحوسبة

القسم الأكاديمي

قسم علم الحاسوب

دولة الجامعة

الأردن

الدرجة العلمية

ماجستير

تاريخ الدرجة العلمية

2018

الملخص الإنجليزي

The network requirements have changed greatly in the past decade, since the growth of file sharing and hosting applications on servers.

Current organizations and research groups use higher computing specifications, server and desktop virtualization, cloud computing and remote storage services to meet their demands.

These technologies introduce a dramatic change in the network resources.

A new networking architecture called Software Defined Networking has been developed.

Software Defined Networking splits the control plane and the data plane, places less intelligence in the data plane and moves it to a fully centralized software program that manages all the network operations (control plane).

In a basic Software Defined Networking operation, when packets are delivered to the network devices and switches, a switch performs a flow table look up to match the packet information with an existing flow entry.

The flow entry fields are used to forward a packet to the right route.

The switch requests a new flow entry from Software Defined Networking controller(s) for each undefined flow.

Software Defined Networking switches have a limited usable memory space, which limits the switch from loading a large number of flow entries.

The memory limit causes the switch to update its flow table more frequently.

Since a new entry in a switch table is required for each undefined flow, the network traffic will increase.

The fine-grained in entries increases the number of entries and may lead to a memory overload.

The impact could become of multiple folds during a distributed denial of service attack.

Therefore, a flow update algorithm should be selected by the controller when the flow tables are full.

This research presents a new flow table update policy, Volume-Based Flow Prioritization (VBFP).

The policy is based on collecting statistics about the generated flow, rank the flows according to their frequencies and volume, and prefer frequent flows with large volumes over others.

The new policy has been fully implemented and simulated in C++ and compared to standard policies.

According to our simulation results, the new policy showed up to 48% improvement over the standard policies used in Software Defined Networking switches (HARDTIMEOUT and IDELTIMEOUT).

التخصصات الرئيسية

تكنولوجيا المعلومات وعلم الحاسوب

الموضوعات

• برمجيات الحاسوب

عدد الصفحات

88

قائمة المحتويات

• APA
• MLA
• AMA

لغة النص

الإنجليزية

نوع البيانات

رسائل جامعية

رقم السجل

BIM-890516