Hot Lab: The News
Lab Scope and Goals
Software from the Lab
Experiments
Publications
Projects
History
Cloud Storage Lab Team

Hot Lab: The News

UCC 2014 tutorial participants: Check out the CloudFusion transport, the NubiSave storage controller, the Splitter-NG data coding framework, and the StealthDB native cloud database.
BlackSeaCom 2014 links: Find here the ongoing implementation of dispersed processing in Python, along with a bit stream (de)multiplexer in C.
In the winter semester 2013/14, the storage team meets about bi-weekly in INF 3080. Visitors are invited to join. Feel free to contact us to ask for topics and schedules.
The Cloud Storage lab is about to coordinate the IEEE TCSC Technical Area on Distributed Data Storage and Computing.

(jump to top)

Lab Scope and Goals

The Cloud Storage Lab is a focused laboratory setup for research on orchestrated and user-controlled autonomous distributed storage. It is hosted at Technische Universität Dresden, Department of Computer Science, Institute for Systems Architecture, and supported by OSSPRI as well as public and private funding partners for individual projects and results. For researchers and students, it offers an experimental area for hands-on research on cloud storage integration, storage controllers and gateways, data distribution policies and similar topics.

(jump to top)

Software from the Lab

We create applications, tools and filesystems for distributed and cloud storage and make it available as open source software to the community. The main user-visible applications developed by us:
  • NubiSave - An optimal cloud storage controller. The controller can be used standalone or as part of a storage integrator or gateway. The main user-centric application is the storage flow editor which determines the flow of data from a virtual folder across several splitting and modification modules towards the storage targets represented by local directories, removable media, or network, online and cloud storage services. Random screenshot: extended dialogue for storage service selection - screenshot
  • NubiVis - distributed data visualisation. A web application which allows users to see where their data is stored and how changes to the configuration (e.g. cloud downtime) affects their file availability. The NubiVis application accesses the NubiSave distribution database through a RESTful web service.
  • NubiGate - virtual gateway machine which combines NubiSave, NubiVis, an extended OwnCloud web interface and other tools into one pre-configured system. This should be the primary method of deployment of the software for production scenarios.
Further tools used for research purposes:
  • peaCS testsuite - see below for experiments results
  • cloud resource allocation simulator
  • syncme and adsyncer - file synchronisation tools
We also develop a number of filesystems to integrate cloud services at a system level. All filesystems are based on FUSE to make them (somewhat) portable.
  • nubisave splitter - replication, erasure coding and bitsplitting of files (Java, see NubiSave as well as the new Splitter-NG codec abstraction framework)
  • cloudfusion - extensible optimised adapter/transport to access SugarSync, DropBox, Amazon S3, Google Storage and WebDav accounts (Python, see CloudFusion)
  • flowfs - reconfigurable distributed storage flows (C)
  • datasetfs - virtual datasets from individual files (C/mysqlfs, see DSCreator)
  • fusim - FUSE simulator filesystem with varying quality characteristics (Python)
  • bpelmonitorfs - system-level access to business process status (Python, see PyPEL)

(jump to top)

Experiments

We perform both long-running and sporadic experiments.
The Content Delivery Network (CDN) for random NubiSave lab tests: peaCS source code:
peacs.git
SVG diagram example:
read performance, all-in-parallel strategy

(jump to top)

Publications

Our peer-reviewed publications:

  • J. Spillner, J. Müller: PICav: Precise, Iterative and Complement-based Cloud Storage Availability Calculation Scheme. 7th IEEE/International Conference on Utility and Cloud Computing (UCC), pp. 443-450, London, UK, December 2014. (read preprint, check proceedings (upcoming), view slides)
  • J. Spillner, A. Schill: Algorithms for Dispersed Processing. 1st International Workshop on Advanced in Cloud Computing Legislation, Accountability, Security and Privacy (CLASP), London, UK, December 2014. (read preprint, check proceedings (upcoming), view slides)
  • J. Spillner, S. Tilsch, A. Schill: NubiVis: A Personal Cloud File Explorer. 11th International Conference on Mobile and Ubiquitous Systems: Computing, Networking and Services (MobiQuitous), London, UK, December 2014. (read preprint, check proceedings (upcoming); demonstration)
  • J. Spillner, A. Schill: Towards Dispersed Cloud Computing. 2nd IEEE International Black Sea Conference on Communications and Networking (BlackSeaCom), pp. 175-179, Chișinău, Moldova, May 2014. (read preprint, check proceedings, view slides)
  • J. Spillner, A. Schill: Orchestration of Distributed Storage Targets through Storage Flows. 5th IEEE International Conference on Cloud Computing Technology and Science (CloudCom), pp. 349-354, Bristol, UK, December 2013. (read preprint, check proceedings; demonstration)
  • J. Spillner, M. Quellmalz, M. Friedrich, A. Schill: peaCS - Performance and Efficiency Analysis for Cloud Storage. Workshop of Cloud Storage Optimisation (CLOUSO) at the 2nd European Conference on Service-Oriented and Cloud Computing (ESOCC), Málaga, Spain, September 2013. (read preprint, check proceedings, view slides)
  • J. Spillner, J. Müller, A. Schill: Creating Optimal Cloud Storage Systems. Elsevier Future Generation Computer Systems (FGCS) 29(4):1062-1072, June 2013. (read article in HTML/PDF; DOI 10.1016/j.future.2012.06.004; proof online June 2012)
  • J. Spillner, A. Schill: Flexible Data Distribution Policy Language and Gateway Architecture. 1st IEEE Latin American Conference on Cloud Computing and Communications (LatinCloud), pp. 1-6, Porto Alegre, Brazil, November 2012. (read preprint, check proceedings, view slides)
  • J. Spillner, G. Bombach, S. Matthischke, J. Müller, R. Tzschichholz, A. Schill: Information Dispersion over Redundant Arrays of Optimal Cloud Storage for Desktop Users. 4th IEEE/ACM International Conference on Utility and Cloud Computing (UCC), pp. 1-8, Melbourne, Australia, December 2011. (read preprint, check proceedings, view slides)
    Best Paper Award

Student theses:

  • A. Gupta: Disaster Recovery in Complex Distributed Storage Pools. Master thesis, TU Dresden, December 2014.
  • M. Quellmalz: Situationsoptimale Datenspeicherung in einem kontextsensitiven Speicherdienstcontroller. Diploma thesis, TU Dresden, December 2014.
  • F. Kleeblatt: Datenqualitätsadaption in verteilten Umgebungen. Bachelor thesis, TU Dresden, November 2014.
  • J. Müller: Autonomic Calibration of Cloud Storage Controllers. Diploma thesis, TU Dresden, April 2014.
  • H. He: Advanced User-Friendly Configuration and Visualisation of Distributed Data. Master thesis, TU Dresden, March 2014.
  • D. Apsel: Inhaltsbezogene Richtlinien in Cloud-Storage-Systemen. Bachelor thesis, TU Dresden, February 2014.
  • M. Quellmalz: Parametrisierbares Dispersionsmodul für den Einsatz in einer Cloud-Speicherdienst-Integration. Study assignment (Großer Beleg), TU Dresden, November 2013.
  • M. Friedrich: Experimentelle Bestimmung einer optimalen Cloud-Storage-Parametrisierung. Bachelor thesis, TU Dresden, July 2013.
  • J. Müller: NubiSave++: Failure Resilient Distributed File System in the Cloud. Study assignment (Großer Beleg), TU Dresden, March 2012.

(jump to top)

Projects

Projects within the Cloud Storage Lab are individual or team efforts towards more powerful data handling in the cloud. The currently ongoing projects are:
  • demos @ MobiQuitous 2014 on distributed data visualisation and @ NetSys 2015 on stealthy native cloud analytics applications
  • several cloud storage talks and a tutorial @ UCC 2014, the 7th IEEE/ACM International Conference on Utility and Cloud Computing
  • watch this space: STORMCLOUD
  • ... and of course, constantly improving the existing software to defend our legacy :)
Finished projects:
  • mini-sabattical at Free University of Bozen-Bolzano to extend the work on secure cloud databases
  • demos @ CloudCom 2013 (Dec 3 or 5) to complement reconfigurable storage flows poster presentation and @ UCC 2013 (Dec 9) during the workshop posters/demos session
  • HPI FutureSOC Lab resources for cloud storage analysis: project page
  • practical course with team on NubiVis extensions towards Next-Generation Cloud File Manager
  • Tutorial: Bundling Cloud Storage Services, 2nd Karlsruhe Service Summer, supported by SUCRE, September 2013.
  • Group practical assignment (Komplexpraktikum): Where is my data?, July 2013
  • Demo: π-Box M3 at OUTPUT.DD and LNdW in Dresden, Germany, July 2013.
  • Demo: π-Box M3 at NetSys in Stuttgart, Germany, March 2013.
  • Demo: π-Box M3 (launch) at CeBIT in Hannover, Germany, March 2013.
  • Tutorial: Workshop on Optimal Cloud Storage, Dresden, Germany, June 2012.

(jump to top)

History

Around 2010, within the context of the FlexCloud young investigator group, novel ways of interacting with services and in particular data-centric services were explored. By the end of 2011, the first prototype of NubiSave had been created to store data safely across multiple cloud storage services. Shortly thereafter, it became clear that fully user-centric cloud storage would require more than a controller. Over the months, mostly-autonomous configuration, data distribution policies and semantic provider descriptions with ontologies were explored. Furthermore, visualisation of distributed data has also been considered. In late 2013, the Cloud Storage Lab was created to represent these ongoing activities in a coherent form.

(jump to top)

Cloud Storage Lab Team

The Cloud Storage Lab is operated by Dr.-Ing. Josef Spillner within the group of Prof. Dr. rer. nat. habil. Dr. h. c. Alexander Schill. It counts on contributions by many students especially for software development and experiments. In the following, current and former students are listed in alphabetic order of their family name.
  • David Apsel
  • Gerd Bombach
  • Volodymyr Bondarenko
  • Anton Caceres
  • Martin Friedrich
  • Alok Gupta
  • Matthias Hille
  • Haosheng He
  • Xiangjie Hu
  • Sharmin Khaleque
  • Florian Kleeblatt
  • Steffen Matthischke
  • Johannes Müller; scientific assistant in 2014
  • Maximilian Pensel
  • Maximilian Quellmalz
  • Sebastian Tilsch
  • Rico Tzschichholz