Paul Bramsen and Sam Kumar and Andrew Chen and Dibyo Majumdar

EECS Department, University of California, Berkeley

Technical Report No. UCB/EECS-2017-56

May 11, 2017

http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-56.pdf

Internet-connected devices are rapidly becoming ubiquitous. This so called Internet of Things (IoT) carries significant security implications thanks to all the sensors, actuators, and computers that make up the devices. If they are not properly secured, IoT devices enable malicious actors to spy on, digitally attack, and even physically harm victims. Unfortunately, many IoT devices are currently chock-full of security holes. We believe that this is largely the result of traditional network abstractions being a bad fit for the IoT. The Global Data Plane (GDP) is a new architecture for global IoT storage and communication that makes data the network's ``narrow waist" by presenting users with secure single-writer append-only logs.

In this report, we present some improvements to the GDP and show that the append-only log is an abstraction that can underpin powerful applications which operate on mutable data. First, we present changes to some of the GDP protocols to improve the performance and security of the GDP. We describe these changes in detail and argue for their merit based on empirical data. Next, we present the Global Data Plane File System (GDPFS), a distributed filesystem that expresses mutable files on top of append-only logs and provides efficient data access within files. Because it is built on top of the GDP, the GDPFS has the potential to scale very well and run securely while giving application developers a traditional interface for managing data.

Advisors: John D. Kubiatowicz

---

BibTeX citation:

@mastersthesis{Bramsen:EECS-2017-56,
    Author= {Bramsen, Paul and Kumar, Sam and Chen, Andrew and Majumdar, Dibyo},
    Title= {Exploring a New IoT Infrastructure},
    School= {EECS Department, University of California, Berkeley},
    Year= {2017},
    Month= {May},
    Url= {http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-56.html},
    Number= {UCB/EECS-2017-56},
    Abstract= {Internet-connected devices are rapidly becoming ubiquitous. This so called Internet of Things (IoT) carries significant security implications thanks to all the sensors, actuators, and computers that make up the devices. If they are not properly secured, IoT devices enable malicious actors to spy on, digitally attack, and even physically harm victims. Unfortunately, many IoT devices are currently chock-full of security holes. We believe that this is largely the result of traditional network abstractions being a bad fit for the IoT. The Global Data Plane (GDP) is a new architecture for global IoT storage and communication that makes data the network's ``narrow waist" by presenting users with secure single-writer append-only logs.

In this report, we present some improvements to the GDP and show that the append-only log is an abstraction that can underpin powerful applications which operate on mutable data. First, we present changes to some of the GDP protocols to improve the performance and security of the GDP. We describe these changes in detail and argue for their merit based on empirical data. Next, we present the Global Data Plane File System (GDPFS), a distributed filesystem that expresses mutable files on top of append-only logs and provides efficient data access within files. Because it is built on top of the GDP, the GDPFS has the potential to scale very well and run securely while giving application developers a traditional interface for managing data.},
}

EndNote citation:

%0 Thesis
%A Bramsen, Paul 
%A Kumar, Sam 
%A Chen, Andrew 
%A Majumdar, Dibyo 
%T Exploring a New IoT Infrastructure
%I EECS Department, University of California, Berkeley
%D 2017
%8 May 11
%@ UCB/EECS-2017-56
%U http://www2.eecs.berkeley.edu/Pubs/TechRpts/2017/EECS-2017-56.html
%F Bramsen:EECS-2017-56