Dynamic resource allocation techniques using smart load balancer

Dynamic resource allocation techniques using smart load balancer algorithm Md Baitul Al Sadi, Isaac J. Cushman, Lei Chen, Rami J. Haddad Department of Information Technology, Department of Electrical Engineering Georgia Southern University Statesboro, USA {ic 00214, ms 12508, lchen, rhaddad}@georgiasouthern. edu

Outline • • • Motivation Abstract Introduction Open. Stack Cloud Design Resource Management, Allocation and Provisioning • Proposed Solution • Conclusion

Motivation • Growing complexity of data types in the cloud medium. • Security concerns with cloud access. • Bandwidth allocation predictions indicate shortage of availability in the near future with Internet of Things.

Abstract • We present a new framework that will allow for a smart load balancer to efficiently allocate resources to increase application processing speed for data and request response of memory stored by mobile devices in a secure manner.

Introduction • Mobile Cloud Computing (MCC) offers the ability to offload data processing. • MCC is offered as “everything-as-aservice” – Platform – Software – Infrastructure • MCC can potentiality meet the computational limitation of the mobile device.

● Mobile device applications require complex data type storage and execution. ● Scalability and redundancy are fundamental when upholding confidentiality, availability and integrity. ● Core goals of MCC: – – Computational power Storage Bandwidth Data heterogeneity

Open. Stack • Features: – Open Source – Large development support – Ability to launch public and private cloud • Configurable cloud design to fit scalability needs. • Ability to exist in a private entity but access a public domain.

The following Open. Stack projects, listed by their project names, are defined in [8]: • Keystone: Authentication and authorization service that operates as the identity of the cloud network. • Glance: Operates as the image service for the cloud network; this software is responsible for creating, editing and provisioning virtual machines. • Neutron: Establishes the internal and external bridge connections between each of the nodes and the other Open. Stack services. • Nova: Manages the lifecycle of compute instances. This includes spawning, scheduling and decommissioning of virtual machines on demand. • Cinder: This software is the block storage on the cloud network; it provides persistent block storage on the instances created by Glance. • Horizon: This software is the web-based system that allows for the cloud provider to quickly access and manages each of the services in the cloud outside of the command line interface.

Design of Private Cloud • Physical hardware: – Dell Poweredge R 820 Servers • Run on Ubuntu server 16. 04 • Uses Open. Stack software to create databases. • Metal As a Service (MAAS): – Treats hardware as if they are Virtual Machines

Network Topology for Private Cloud

Controller, Compute and Network Nodes • a minimum requirement of 3 network servers • controller ○ Runs the virtual machine identity and image services, management portion of compute node and the dashboard • compute ○ Organizes and calls tenant virtual machines or instances, connects network plug ins and firewall services • network ○ Runs the networking plug in and several agents to provide switching, routing, NAT and DHCP

Resource Management, Allocation and Provisioning • Important to prevent overloading in single machine. • Aims to provide Quality of Service (Qo. S) in the form of confidentiality, integrity and availability • Offer features like: – resource optimization, – diminishing of response time and down time – maximizing the throughput – avoiding of overload

Load Balancer ● The job of a load balancer is to: ● ● store data in optimal locations to point user to data in quickest path ● Current methods for load balancing require single specific target to meet. ● Common configurations: ● ● ● Priority Round Robin Least loaded VMs

Proposed Solution • Load balancer with intelligent decision making. • Should consider a heterogenous data configuration – Emphasis on priority requests – Promotes fairness in the system – Minimal decision delay for large networks

Dynamic Resource Allocation • User Generated Data: User generated data can be referred to as the data generated by the user. • Application data: All mobile application driven data can be classified as application data. • System data: All data associated with the system information.

Security Class level

Smart Load Balancer • Need before Greed – Idea comes from classic reward disbursement made famous by MMORPG video games. • Goal is to determine each user that is more in need of immediate and or high resolution. • Decision making will exist in application hosted by the cloud controller

Smart Load Balancer Algorithm

Smart Load Balancer Algorithm 1: Gather current network requests. 2: Establish handshake between VMs and Server 3: Initialize SLB to organize requests Case 1: Priority Target is requesting BW a. Check leasing history for overuse b. Grant leasing time to PT with allocated BW Case 2: Many other VMs are requesting BW a. If no PT, begin round robin b. If PT is active, allocate remaining BW 4: If overuse by PT is found, PT will have cap placed End

Front View of Hardware used for Mobile Cloud Network

Conclusion • Complex data types in networking creates a problem with maintaining a consistent protocol for cloud security. • Proposed smart load balancer and cloud in cloud method provides a more secure, intelligent connection. • Future work goal is to implement full system to test the viability of the smart load balancer. • Possible changes to the structure if request latency or unfairness in the system occurs.

Thank you for your time! Are there any Questions?