Distributed Systems Seminars (DSS)
Corso di Laurea Specialistica in Ingegneria Informatica

AA 2007/2008

Principal Instructor Roberto Baldoni


NEWS!!!

Schedule

Thursday 17:30-19:00 room A3 (Via Ariosto)

Friday 15:45-17:15 room A3 (Via Ariosto)

Friday 17:30-19:00 room A3 (Via Ariosto)

Classes will start on April 24, 2008


DSS is a course in computing systems, with a strong focus on operating systems and distributed computing. The course is aimed at students seeking broad background in the areas of importance to modern systems researchers, or having an interest in possible research topics. The course involves a certain amount of reading, and all students who take the class are expected to read the papers for the experience to be satisfactory.  Most papers are chosen from the best conferences and journals in systems, and most are available on-line.The topics we'll span a range of hot areas within modern systems. 

Each seminar will be composed of one or two lectures. The seminar will consist of a formal presentation, and is expected to go into considerable depth by focusing on some aspect of the material and treating it thoroughly. After each seminar, students are required to email to me (baldoni@dis.uniroma1.it) and to the instructor of the specific seminar one page discussing the contributions and at least one major weakness of each required reading.

In addition to sending one page document each week, to pass the exam each student will be assigned to a project and he has to write a high-quality paper about the project.  Topics for the projects will be related to the themes discussed in the seminars and the list of projects will be available approximately two weeks before the end of the course (tips on how to prepare a paper will be given during the seminars). The project will be done in three stages:


Schedule

Thursday 17:30-19:00 room A3 (Via Ariosto)

Friday 15:45-17:15 room A3 (Via Ariosto)

Friday 17:30-19:00 room A3 (Via Ariosto)

Classes will start on April 24, 2008


Tentative list of Seminars

Week Speaker Topic Required Readings for weekly assignments Suggested readings
24 April Roberto Baldoni

Handling Arbitrary failures.

Nowadays, there are many protocols able to cope with process crashes, but, unfortunately, a process crash represents only a particular
faulty behavior. Handling and monitoring tougher failures (e.g. sending omission failures, receive omission failures, arbitrary failures) is a real practical challenge due to malicious attacks or unexpected software errors. The seminar addresses a methodology to build protocols resilient to arbitrary failures and techniques to implement accountability of distributed system spanning multipledomains.

"A methodology to design arbitrary failure detectors for distributed protocols" Roberto Baldoni, Jean-Michel Helary, Sara Tucci Piergiovanni

"PeerReview: practical accountability for distributed systems" Andreas Haeberlen, Petr Kouznetsov, Peter Druschel
8-9 May Roberto Baldoni

E-gov: servizio pubblico di connettività . SPCoop aims at developing a nationwide e-Government Service Oriented Architecture,
in order to guarantee a flexible approach for integrated application services, respecting local and central administrations’ autonomy. An appropriate mixture of organizational initiatives, together with the promulgation of appropriate laws, and the development of innovative
technical rules, seems to be the success factor of the approach. The seminar will focus on all these aspects pointing out the complexity of the overall picture.

lecture slides

"The Italian e-Government Service Oriented Architecture: Strategic Vision and Technical Solutions" R. Baldoni, S. Fuligni, M. Mecella, F. Tortorelli

Servizi di interoperabilità evoluta e cooperazione applicativa -documenti di riferimento:

Quadro tecnico d'insieme

Porta di dominio

Accordo di servizio

Servizi di registro

Servizi di sicurezza

Nomenclatura e semantica

15-16 May

Roberto Beraldi

Exploiting partial knowledge in probabilistic search algorithms for wireless networks

A recurrent problem arising in a wireless network is to search for a node with a given property. Probabilistic protocols, both sequential (random walks) and parallel (probabilistic flooding, or gossiping), may often be used for this purpose, especially if the network is dynamic. This talk discusses how such probabilistic protocols can highly benefit from even a modest level of knowledge about where the target is positioned in the network. Two protocols, a polarized version of the gossip protocol and a biased random walk are discussed.

"The Polarized Gossip Protocol for Path Discovery in MANETs" R. Beraldi

"Service discovery in MANET via biased random walks" R.Beraldi

 
23 May Carlo Marchetti Three-tier software replication. software replication is a set of techniques suitable for increasing the availability of a software service by replicating the same software components using distinct processes of a distributed system (replicas). When the service maintains an internal state (e.g., a reservation service), and in order to continue feeding correct clients with correct replies even in the presence of failures, it arises the need of maintaining consistency among the states of replicas. Furthermore, if clients and/or replicas may be hijacked, it is important to preserve the correctness and the confidentiality of the replies returned to clients by the service. In this lectures students are preliminary introduced to the basic concepts of software replication, and then to recent research results on software replication protocols, in which a three-tier architecture is used for efficiency and for security purposes.

"Fully Distributed Three-Tier Active Software Replication" R. Baldoni, C. Marchetti, S. Tucci Piergiovanni, A. Virgillito

"Separating agreement from execution for byzantine fault tolerant services" Jian Yin, Jean-Philippe Martin, Arun Venkataramani, Lorenzo Alvisi, Michael Dahlin:

R. Guerraoui and A. Schiper. Software based replication for fault tolerance. IEEE Computer, 30(4), Apr. 1997

Additional info (e.g., slides of the seminar) can be found following this link

29-30 May Adnan Noor Mian Searching and Service Discovery in Mobile Ad hoc networks. Mobile and pervasive environments are inherently scarce in resources. The devices in such environments need to cooperate among themselves for performing tasks that cannot be done alone. This cooperation is in the form of services that are offered by such devices. To get benefit from these services they have to be discovered. Service Discovery Protocols (SDPs) are used for this purpose.

In this seminar we shall discuss SDPs for mobile ad hoc networks. In the first part of the talk we shall briefly introduce SDP and its different components. In the second part we shall discuss a SDP that form virtual backbone of ad hoc nodes/devices and in the third part we shall see how random walk can be used for searching and service discovery and will discuss a protocol that efficiently implements random walk.

Network Layer Support for Service Discovery in
Mobile Ad Hoc Networks

Ulas¸ C. Kozat and Leandros Tassiulas

A Robust and Energy Efficient Protocol for RandomWalk in
Ad hoc Networks with IEEE 802.11

Adnan Noor Mian, Roberto Beraldi, Roberto Baldoni

Survey of Service Discovery Protocols in Mobile Ad Hoc Networks
Technical Report 4/06, MIDLAB/Dipartimento di Informatica e Sistemistica Antonio Ruberti
Adnan Noor Mian, Roberto Beraldi, Roberto Baldoni
5-6 June Leonardo Querzoni Search in peer-to-peer systems
The Internet traffic is today dominated by complex and chaotic  distributed systems, lacking any centralized organization or  hierarchical control. These systems usually are referred to as Peer-to- Peer (P2P) systems. These systems try to provide a long list of  features such as: selection of nearby peers, redundant storage,  ef?cient
search/location of data items, data permanence or guarantees,  hierarchical naming, trust and authentication, and, anonymity.P2P  networks potentially offer an ef?cient routing architecture that is  self-organizing, massively scalable, and robust in the wide-area,  combining fault tolerance, load balancing and explicit notion of  locality. In this seminar we explore the various search mechanisms  implemented by different Peer-to-Peer systems, and show a performance  comparison  that highlights the strengths and weakness of three  representative solutions.
slides

 

Xiuqi Li and Jie Wu, Searching Techniques in Peer-to-Peer Networks

Eng Keong Lua, Jon Crowcroft, Marcelo Pias, Ravi Sharma and Steven Lim, A Survey and Comparison of
Peer-to-Peer Overlay Network Schemes

Beverly Yang Hector Garcia-Molina, Improving Search in Peer-to-Peer Networks

Indranil Gupta, Ken Birman, Prakash Linga, Al Demers, Robbert van Renesse. Kelips: Building an Efficient and Stable P2P DHT
Through Increased Memory and Background Overhead


Based on the XOR Metric
Petar Maymounkov and David Mazi`eres. Kademlia: A Peer-to-peer Information System

12-13 June TBD