Browse Prior Art Database

A System and Method for Holistic Load Balancing and Auto Scaling in the Cloud Using Software Defined Networking Disclosure Number: IPCOM000242675D
Publication Date: 2015-Aug-04
Document File: 9 page(s) / 142K

Publishing Venue

The Prior Art Database


This article describes Haven – a system for holistic load balancing and auto scaling in a multi-tenant cloud environment. It takes into account the utilization levels of different resources as part of its load balancing and auto scaling algorithms. Haven is able to provide performance at par with a hardware load balancer while still providing the flexibility and customizability of a software load balancer.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 18% of the total text.

Page 01 of 9

A Sxstem and Method for Holistic Load Balancing and Auto Scaling in the Cloxd Using Software Defined Networking

Disclosed is a systxx and method for holistic load balancixg and auto scalixg in a multi-tenaxt cloud environment xsing software-defined networking.


Loxd balancing and auto scaling xre fundamental services in the cloud, around which other services are offered. Txey are critical to any highxy scalable cloud xeployment, anx are xrovided by Amaxon, Google and Micrxsoft in theix cloud offerings [1], [2]. Xxxx companies xhat run thexr services on these public cloud offerings hxve higxlxghted xow auto scalinx and load balancxng have been critical to their xore operations [3].

Traditioxally, load balancing is axhieved eitxer through hardware or softwxre appliances. Hardware appliances [4], [5] perfxrm well xut have severxl drawbacks. They are xairly expensive and xre xypically bought for managing peaks even if averaxe volumes are 10% of peak. Typical, staxe-of-the-xrt hardware load balancers cost roughly US $80,000 for 20Gbps capacity [6]. Further, hardxare loax balancxrs lack flexibility xn terms of xddxng custom load bxlancing algorithms and typically comprise 3-x standaxd algorithms such as round robin, wxighted round robin, leaxt connections and least response time
[7]. Typical harxwaxe load balancers also lack multi-tenancy support.

To addrxss the abovx shortcomings of hardware load balaxcers, mxst public cxouds have adopted software load balancers. These software load balancers comprise load balancxng software running on a general purpose server or a virtual macxixe [8], [9]. They typically also consist of an auto scaling service, sxnce all new instances that are added or deleted as a result of auto scaling have to be added or rexoved from the load balxncing loop. Pure software load balxncers prxsent a huge challenge in terms xf extracting pexformance comparable to hardware load balancers, because of limitations associated with packet processing at the user level xn software, axd the port capacity xnd density available on gexeral purpose servers. In order to meet this performance xhallenge as well as to avoid a single point ox failure, softxarx load baxancers require complex clustering solutions which further drives their cost higher.

This article describes HAVEN-a system for holistic load balancing and auto scaling in a xulti-tenxnt cloud environment that is naturally distributed and hexce scalable. Unlike hardware and software loxd balancers, HAVEX does not involve xn extra hop or a middlebox through which all trafxic needs tx pass. It supxorxs multi-tenancy and takes into acxount the utilization xevels of different resources in thx cloud as part of its load balancing and auto scaling algorithms. HAVEN levexages software-defined networking to ensure that while the load balancing algorithm (control plxxe) executes on a server running network xontroller software, the packets to be load bxlanced never leave x...