This preview shows pages 1–2. Sign up to view the full content.
This preview has intentionally blurred sections. Sign up to view the full version.View Full Document
Unformatted text preview: SecondNet: A Data Center Network Virtualization Architecture with Bandwidth Guarantees Chuanxiong Guo, Guohan Lu, Helen J. Wang * , Shuang Yang, Chao Kong, Peng Sun, Wenfei Wu, Yongguang Zhang MSR Asia, * MSR Redmond MSR-TR-2010-81 ABSTRACT In this paper, we propose virtual data center (VDC) as the unit of resource allocation for multiple tenants in the cloud. VDCs are more desirable than physical data cen- ters because the resources allocated to VDCs can be rapidly adjusted as tenants needs change. To enable the VDC ab- straction, we designed a data center network virtualization architecture called SecondNet. SecondNet is scalable by distributing all the virtual-to-physical mapping, routing, and bandwidth reservation state in server hypervisors. Its port- switching based source routing (PSSR) further makes Sec- ondNet applicable to arbitrary network topologies using com- modity servers and switches. SecondNet introduces a cen- tralized VDC allocation algorithm for virtual to physical map- ping with bandwidth guarantee. Simulations demonstrated that our VDC allocation achieves high network utilization and low time complexity. Our implementation and experi- ments on our testbed demonstrate that we can build Second- Net on top of various network topologies, and SecondNet provides bandwidth guarantee and elasticity, as designed. 1. INTRODUCTION With the advent of Amazon EC2, Google App En- gine, and Microsoft Azure, the dream of computing-as- a-utility is becoming a reality [25, 28]. By outsourcing computing to the cloud, utility computing frees busi- nesses and consumers from the cost and burden of plan- ning, purchasing, operating, and maintaining physical hardware and software, and at the mean time, it offers elasticity to meet dynamic demands in resources and good economy with a pay-as-you-go billing model . The Service Level Agreement (SLA) of todays utility computing [3, 26, 4, 27] are centered around computa- tion (dollars per hour per virtual machine or VM), stor- age (dollars per GB per month), Internet trac (dollar per GB transferred), and the availability of these re- sources. Nevertheless, no abstraction or mechanisms and hence no SLAs are available to capture the require- ments on the interactions among the allocated VMs, such as bandwidth guarantees among the VMs. In this paper, we propose virtual data center (VDC) as the abstraction for resource allocation. A VDC is defined as a set of VMs with a customer-supplied IP address range and an associated service level agree- ment (SLA). The SLA specifies not only computation and storage requirements (such as the number of VMs, CPU, memory, and disk space of each VM), but also bandwidth requirements for the VMs. The bandwidth requirement is a key addition and offers the significant benefit of performance predictability for distributed com- puting. A VDC gives the illusion of a dedicated physical data center. This requires VDCs to be isolated from one another in all resource access and usage. A VDC is inanother in all resource access and usage....
View Full Document
This note was uploaded on 12/01/2011 for the course EE 5373 taught by Professor Chao during the Spring '11 term at NYU Poly.
- Spring '11