DSG is holding weekly meetings to create an space for researchers to learn and discuss about their latest findings in the area.
There is the tradition in several research groups of all around the world universities, to hold weekly meetings to allow their peers to talk and discuss about their latest work, sometimes served with some sweets to boost brainy conversations. At DSG we have decided to give it a try :).
So... How does it work?
Each week someone agrees to take the responsibility to open a discussion forum about any interesting topic. This person is responsible for sending an e-mail with a brief summary to the meeting mailing list (dsg-weekly [at] ac [dot] upc [dot] edu) and prepare some talk/outline/presentation/demo about it. The appointed day we all meet at some place and start a productive-informal discussion about the topic.
Other weeks we can pick some recent papers relevant to distributed systems and watch the conference talks during the session if it is available. Otherwise, one of us can read the paper in advance and make a small presentation. Then, during the meeting we will try to answer questions like:
Why was the paper accepted? What problems does it solve? Has anything been overlooked? Are there too many assumptions? Can there be a better approach? Any specific cases where the solution may not work or result in considerable overhead? Weakness?
The goal is to be as critical as possible in order to better understand the ongoing work and the motives behind it. The session might result in identifying open problems that need to be addressed and will serve to keep us updated with the current progress of research in our area.
When is it going to be?
Initially, we plan on meeting on Thursdays at 15:30. But in any case, we can move the event to other days if it increases the participation.
Where will it be?
We will reserve room C6-E101 for such meetings, since it has enough space and the needed equipment.
How do I help?
You can help us in (at least) 3 different ways:
What topics will be discussed next Thursday?
You can find out which is the topic of the next meeting by checking the announcements of this page; we will keep updated the sessions that we hold. Also, you can subscribe to dsg-weekly if you're interested in receiving information via e-mail about the meetings.
So, see you next Friday!
Abstract Caught between the infinite promise unleashed by technology proliferation and the unprecedented scale of resource depletion, waste and inequity, we inhabit a space where critical alternatives are sought more than ever. As a reflection of the above, we find in HCI, a slant towards technological fixes to existing sustainability problems, as opposed to a more holistic approach that includes behavioural and societal change. It is within this context that this paper is situated, where we propose a socio-ecological approach and argue our case for a life-cycle lens towards building systems that are in line with current understanding of the earth’s finite resources. We do so by presenting an illustrative case study of what such critical alternatives might look like, by examining the Fairphone movement. We contribute to a deeper understanding of how social value laden enterprises along with open technological design can shape sustainable relationships between our environment and us.
Related paper presented at ICT4S: http://www.atlantis-press.com/php/pub.php?publication=ict4s-env-15&frame...
Tessy Cerratto Pargman http://teresacerrattopargman.blogs.dsv.su.se I am an associate professor of Human-Computer Interaction (HCI) at Stockholm University (SU) in Sweden since February 2003. My research belongs to the research field of human-computer interaction (HCI) and Technology Enhanced Learning (TEL). I am interested in the relationship between writing, meaning and technology from socio-cultural and critical perspectives of literacy and tool use. I work with a particular focus on design, adoption and use of technologies for participatory and collaborative purposes. I have after my sabbatical at UCI developed a special interest in issues pertaining to information technology and sustainability in HCI. I lead the research group on Critical Computing (CRITIC) and since january 2016 I am head of Interaction Design and Learning (IDEAL) research unit at the Dept. of Computer and Systems Sciences at Stockholm University.
Somya Joshi is a research scientist with expertise in the field of Human Computer Interaction when applied to Social Innovation & Sustainability. Her specialisation falls within the applied context of technological innovation, particularly in how it translates into transparency in governance, environmental conservation and health services within the developing world. She has experience working with a range of partners from academia, industry, NGOs, as well as international development organisations, towards the common goal of facilitating inclusive development. Currently, Somya is a Senior Lecturer and Research Fellow at eGovernance-Lab, within DSV (Department of Computer & Systems Science) at Stockholm University. She is responsible for course development and teaching on two Masters programs - namely: ICT4D, as well Open Governance & e-Democracy. In addition to this, she is teaching an Undergraduate course on Participatory Design and HCI - with a specific focus on Sustainability. Somya is at present project managing and researching within pan-European projects that aim to bring together the political-ecology of participation and sustainability within the same discourse.
I'll be presenting our paper titled " Augmenting Elasticity Controllers for Improved Accuracy ", which is due to be presented at ICAC 2016. Abstract: Elastic resource provisioning is used to guarantee service level objectives (SLO) at reduced cost in a Cloud platform. However, performance interference in the hosting platform introduces uncertainty in the performance guarantees of provisioned services. Existing elasticity controllers are either unaware of this interference or over-provision resources to meet the SLO. In this paper, we show that assuming predictable performance of VMs in a multi-tenant environment to scale, will result in long periods of SLO violations. We augment the elasticity controller to be aware of interference and improve the convergence time of scaling without over provisioning. We perform experiments with Memcached and compare our solution against a baseline elasticity controller that is unaware of performance interference. Our results show that augmentation can reduce SLO violations by 65% or more and also save provisioning costs compared to an interference oblivious controller.
In this talk, I'll present our work on achieving reliable elastic scaling in multi-tenant environments. This paper is accepted for publication in CCGrid 2016 and is a rehearsal talk for the conference. Looking forward to your comments and feedback.
Abstract:
Elastic resource provisioning is used to guarantee service level objective (SLO) with reduced cost in a Cloud platform.
However, performance interference in the hosting platform introduces uncertainty in the performance guarantees of provisioned services.
Existing elasticity controllers are either unaware of this interference or over-provision resources to meet the SLO. In this paper, we show that assuming predictable performance of VMs to build an elasticity controller will fail if interference is not modelled. We identify and control the different sources of unpredictability and build Hubbub-Scale; an elasticity controller that is reliable in the presence of performance interference. Our evaluation with Redis and Memcached show that Hubbub-Scale efficiently conforms to the SLO requirements under scenarios where standard modelling approaches fail.
(The presentation will take place at room C6-E101 for availability reasons)
Work to be presented: XXXV Sunbelt Conference of the International Network for Social Network Analysis (INSNA) June 23 – June 28, 2015
Davide Vega, Matteo Magnani, Roc Meseguer, Felix Freitag
First talk: 26 weeks in Uppsala. Fika, Nature, Snow... BBQ! How these concepts are related? They could describe any Scandinavian country, but for me they will be always linked to Uppsala (Sweden), a University-city where I spent 26 weeks as part of my PhD. The talk intends to describe my experience in both, the university and the city. I will describe the Swedish university model and how it is related with the research and academic expectations there. It will be a good opportunity to share our experiences (as PhD's) in different institutions (you are welcome to share yours too!). As always, comparisons are a good tool for reflection and I think the talk could open some interesting discussion about our own institution and, more particularly, our research group.
This meeting is more related to EMJD-DC doctorate students and tutors, but everyone is welcome. The idea is to discuss about the organization (timetables, logistics) of the upcoming EMJD-DC summer event. It would be nice if participants come prepared with some proposals based on what they expect from the event. Sorry for the late notice.
(Please note that this week's meeting is at 18:00)
The Community-lab testbed introduces nodes at the edges of the Community network in order to give an environment where Community network members can easily deploy experiments on the nodes. However, there are limitations with the number of physical nodes that the Community-lab has deployed. Therefore, we propose that we can extend this number of testbeds in a virtualized way, without the need of adding new physical nodes to the network. Also, it would seem to be a good addition to have as a new automated service for community members in the Cloudy.community Operating System so that members can deploy their own nodes with ease. Furthermore, we postulate two solutions, using virtualization, by using a virtual machine (QEmu emulation); or by using linux containers (LXC containers). Both solutions can have their advantages and disadvantages, and work within different environments. We are currently in the testing phase of our solutions, in order to gather an evaluation for a paper that we are going to submit in the coming weeks. Therefore, we appreciate any feedback on the way we deploy our solutions and also for the measurements needed to have a more complete evaluation of the community-lab virtualized nodes.
(Please note that this week's meeting is at 15:00)
Abstract: A natural succeeding process for the Internet was to create Social Networks (e.g. Facebook, among others), where anyone in the World can share their experiences, knowledge and information, using personal computers or mobile devices. In fact, Social Networks can be regarded as enabling information sharing in a peer-to-peer fashion. Given the enormous number of users, sharing could also be applied to the untapped potential of computing resources in users’ computers.
By mining the user friendship graphs, we can perform people (and resource) discovery for distributed computing. Actually, employing Social Networks for distributed processing can have significant impact in global distributed computing, by letting users willingly share their idle computing resources publicly with other trusted users, or groups; this sharing extends to activities and causes that users naturally tend to adhere to.
We describe the design, development and resulting evaluation of a web-enabled platform, called Trans-SocialDP: Trans-Social Networks for Distributed Processing. This platform can leverage Social Networks to perform resource discovery, mining friendship relationships for computing resources, and giving the possibility of resource (not only information) sharing among users, enabling cycle-sharing (such as in SETI@home) over these networks.
Paper presented at:
IFIP'12
Proceedings of the 11th international IFIP TC 6 conference on Networking
Paper available here:
http://www.gsd.inesc-id.pt/~lveiga/prosopon/pubs/24-IFIP-Networking-2012...
To be presented in Middleware'14.
Title: Stay-Away, protecting sensitive applications from performance interference(30 minutes)Abstract: While co-locating virtual machines improves utilization in resource shared environments, the resulting performance interference between VMs is difficult to model or predict. QoS sensitive applications can suffer from resource co-location with other less short-term resource sensitive or batch applications. The common practice of overprovisioning resources helps to avoid performance interference and guarantee QoS but leads to low machine utilization. Recent work that relies on static approaches suffer from practical limitations due to assumptions such as a priori knowledge of application behaviour and workload.
To address these limitations, we present Stay-Away, a generic and adaptive mechanism to mitigate the detrimental effects of performance interference on sensitive applications when co-located with batch applications. Our mechanism complements the allocation decisions of resource schedulers by continuously learning the favourable and unfavourable states of co-execution and mapping them to a state-space representation. Trajectories in this representation are used to predict and prevent any transition towards interference of sensitive applications by proactively throttling the execution of batch applications. The representation also doubles as a template to prevent violations in the future execution of the repeatable sensitive application when co-located with other batch applications. Experimental results with realistic applications show that it is possible to guarantee a high level of QoS for latency sensitive applications while also improving machine utilization.
Title: Monitoring System for Community-Lab(30 minutes)
Abstract: In this session we will talk about the monitoring system for Community-Lab, its various capabilities, current limitations and future work. The metrics monitored include OS-provided resource consumption metrics, synthesized metrics such as open/closed ports and connectivity information. It also provides a view of the network topology and the general network health between the nodes. The design of the monitoring system ensures that no information is lost during network partitions, provides a complete historic access and supports aggregation of data at different granularities. The short talk will particularly focus on the different features of the monitoring system and will provide a short guide on the capabilities that can be leveraged by researchers to choose nodes before deploying a slice. Also people are encouraged to provide any suggestions for requirements or features they deem beneficial for researchers/admins. More information on the monitoring system can be found on the wiki.
Prof. Rajkumar Buyya
Future Fellow, Australian Research Council (ARC),
Director, Cloud Computing and Distributed Systems (CLOUDS) Lab,
The University of Melbourne, Australia
CEO, Manjrasoft Pvt Ltd, Melbourne, Australia
A talk explaining the collaboration between our research group and Pangea as part of APC.
On June 1-9 several activities organised by APC will take place in the campus (Leandro is co-org). Some of the activities are:
Cloud vendor lock-in and interoperability gaps arise (among many reasons) when semantics of resources and services, and of Application Programming Interfaces is not shared. Standards and techniques borrowed from SOA and Semantic Web Services areas might help in gaining shared, machine readable description of Cloud offerings (resources, Services at Platform and Application level, and their API groundings), thus allowing automatic discovery, matchmaking, and thus supporting selection, brokering, interoperability end even composition of Cloud Services among multiple Clouds. This series of two lectures starts with an introduction to Semantic Web and Semantic Web Services technologies, languages and tools. Then an overview of applications of semantic technologies is provided, in particular to discovery and management of Web and Cloud Services, APIs and resources. Finally, the outcomes of the mOSAIC EU funded project are presented: a Cloud Ontology, a Semantic Engine, and a Dynamic Semantic Discovery System. The second lecture consists of practical laboratory exercises and hands-on on Semantic tools for Ontology building and Semantic Web Services annotations. Expected results: Participating students will know about basic and advanced techniques for Semantic processing, Semantic Web Services, Cloud Computing and Cloud Portability / Interoperability issues. In addition, the visibility of the exchange program between the two Universities will hopefully be raised by this lecturing activity, and this might lead to further student exchanges. Day 1: Tuesday 12-14 (as part of the CLC-MIRI course) A5-106 Day 2: Thurdsay 12-13 (for the research group) C6-220 Day 3: Friday 12-14 (as part of the CLC-MIRI course) A5-106
Abstract:
Energy consumption in IT infrastructure is becoming a critical issue nowadays. Looking forward, energy is going to be dear. The energy cost is going to exceed the infrastructure and maintenance cost; thus, as computer scientists, we should welcome the energy era by designing energy efficient architecture, resource management, etc.
In this talk we will speculate the energy consumption of different cloud architectures to find out about the energy efficiency of Community Cloud Computing(C3) architecture. Then we will discuss the possible energy- credit economic modeling and scheduling methods and their challenges; we will conclude with a case study of providing "Energy as a Service" (EaaS) on community clouds.
After the talk, Leandro and me are going to show you a couple of "good" examples based about how to disseminate your knowledge and interests on Internet providing open data and open tools. Maybe it could start an interesting discussion about it!
For people willing to attend remotely, we will share the presentation via Google Hangouts.