distributed systems. Blockchain provides the trust layer that allows
direct interaction between peers. Hence, blockchain could be used as
middleware to coordinate various components of distributed systems and
implement various tasks such as payment, trust, reputation, provenance &
However, main characteristics of blockchain are still relatively unknown
from a performance point of view, but we already know that scalability
is the main issue to solve for a wider adoption. In fact , the
technology is evolving rapidly to address the scalability issue with the
introduction of various form of consensus, such as Proof-of-Stake, side
channel, lightning network, sharding, cross blockchain interaction,
Because of this, new methodologies are required to experiment with
blockchains in an environment that would allow for reproducibility,
configurability and performance evaluation. The topic of this thesis is
to propose a framework for experimenting with the various blockchain
protocols and middleware, in particular by focusing on the Ethereum
The use of virtualization and emulation techniques will be explored: the
work on the Distem emulator, targeting other kinds of distributed
systems, already showed that they are suitable to achieve
experimentation at scale, with controlled heterogeneity of performance
(CPU, network, I/O), or controlled fault injection & load imbalance. We
will build on and extend these techniques, including exploring automatic
or assisted techniques to uncover performance or resilience issues.
Experimentations performed to validate the framework will focus on the
evaluation of Ethereum and the interaction with various forms of side
chain structures to have a better scalability while ensuring the same
level of security and transparency.
PhD work plan
To start with, the current state of the art will be explored, to better
understand the current state of performance evaluation around blockchain
technologies, on one hand, and the relevant experimentation techniques
(simulators, emulators, testbeds).
After that, we will aim at entering a virtuous cycle between the design
of an experimentation framework, and experiments on blockchains that
will provide interesting insight and drive the design of the framework.
This PhD thesis, funded by a CIFRE grant, will be jointly supervised at
both LORIA (Nancy) and iExec headquarters (Lyon). The PhD candidate will
spend a majority of the time in Nancy, but is expected to spend about
three months per year in Lyon.
Master in CS + experience in research (internship, publication)