Row hammer exploit in cloud environment

Thumbnail Image
Date
2018-01-01
Authors
Venkataraman, Adithya
Major Professor
Akhilesh Tyagi
Advisor
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Authors
Research Projects
Organizational Units
Journal Issue
Is Version Of
Versions
Series
Department
Electrical and Computer Engineering
Abstract

The rapid increase in the adoption rate of cloud computing, across numerous businesses, has resulted in extensive use of virtualization tools. Virtualization technology utilizes a software layer (hypervisor) to enable sharing of hardware between multiple tenants that are co-located on the same multi-processor system. This enables the consolidation of servers and user machines into a very small set of physical systems. Physical machines are replaced with virtual machines (VM), running on the same physical system, to achieve better utilization of the hardware. Consequently, cloud users work on and store their data in the same physical machine.

A crucial part of a cloud setup is preventing information leakage between tenants. While the hypervisor enforces software isolation, shared CPU, cache or memory, has the potential to leak sensitive information. This article aims to provide an overview of the security concerns in virtualization technology, particularly in relation to row hammer bug that affects the DRAM chips.

As DRAM process technology scales down in dimension, it becomes increasingly difficult to prevent sub-micron electrical interaction between DRAM cells. This leads to unintentional effects where, activating the same row or same set of rows in DRAM (row hammer) corrupts data in nearby rows. If row hammer is coupled with some resource sharing features that can be enabled in hypervisors, then there is a high likelihood of corrupting a piece of data that belongs to another VM. This article is a survey of some prior works in cloud-based row hammer attacks and sheds some light on the exploit mechanics.

Comments
Description
Keywords
Citation
DOI
Source
Copyright
Mon Jan 01 00:00:00 UTC 2018