Distributed Systems

Distributed Systems

  • Okan Ozgonenel added an answer:
    How to design a distribution system for a factory where there are 11 kV loads, and 415 V loads?

    Transformers selection, cables selections, protection units selection. The load includes induction motors and air-conditioners, pumps, and lighting.

    Okan Ozgonenel · Ondokuz Mayıs Üniversitesi

    Hi Salah,

    I think you simply need a distribution network calculations. First of all, you will have voltage drop and wire calculations. I can help you if you want.

  • Mostafa Eidiani asked a question:
    How can I performance a load scaling?

    I want to automatically determine the simultaneity factors of the loads in my distribution system

  • Chirag Bhalara added an answer:
    Are there any simulators for cloud load balancing?
    Cloud computing.
    Chirag Bhalara · Gujarat Technological University


  • Traci Ruthkoski added an answer:
    Can you give me advice on building an assembled supercomputer?
    I want some suggestion on this topic . If I have 300 desktop computers with quad core-4 core each (total 1200 core processors), can I assemble them to build one highly powerful system? Which technology and programming language should I learn to complete this task?
    Traci Ruthkoski · University of Michigan

    The Rocks cluster distribution is not for those who want to tweak their configuration constantly. BUT you can get it up and running quickly. It will give you the attributes like MPICH and a scheduler for running jobs. You will have to install your own software to run on the nodes, but I would suggest using Linux modules to keep that task under control.


  • Mark Frautschi added an answer:
    Do you know how to simulate an adjustable current-limiting diode in vanilla SPICE (MacSPICE 3f5)?

    I need a simple, passive, current limiter: adjustable 1uA-1A, one that can never be tricked during simulation into sourcing current or voltage. It can remain fixed while SPICE simulates the circuit according to NUTMEG. I am happy to set the current by editing my SPICE deck. Being able to call a SUBCKT would be nice, however, as far as I can tell, MacSPICE does not pass parameters to SUBCKTS. So, editing the current limit in the main code or in the SUBCKT would work equally well, I believe. At present the circuit needs only one current limiting diode, so, the SUBCKT makes little difference.

    In Early September 2014 I edited this question and attached a transfer curve for the desired current-limiting diode and later a simplified version of the circuit I am using to model the biological system. Hopefully these will make my request more clear. I should have included them from the beginning.

    The current-limiting diode is in the upper left, between nodes 100 and 101.

    I corrected my characteristic curve to end at Iin = Imax to eliminate the nonconservation of current problem.

  • Gerro Prinsloo added an answer:
    Challenges for applying “controller in the loop” simulation in distributed automation of energy distribution systems?

    Identify challenges for applying “controller in the loop” simulation in distributed automation of energy distribution systems and suggest ways of overcoming those challenges.

    Gerro Prinsloo · Stellenbosch University

    Dear Adil, will try and help if you can give a little more info in the question. Tell us what simulation platform are you using or planning to use ?

  • Veerabhadram Paduri added an answer:
    What is the best categorization for scheduling algorithms?
    How do we categorize scheduling algorithms in all respect for distributed computing environment?
    Veerabhadram Paduri · Vaal University of Technology

    i think and agree a new taxonomy of scheduling algorithms for scheduling preemptive
    real-time tasks on multiprocessors.this algorithm described some new classes of scheduling algorithms
    and considered the relationship of these classes to the existing well-studied classes

  • Mustapha Bouakkaz added an answer:
    What are the most efficient data aggregation techniques in terms of communication cost for monitoring in large-scale distributed systems?
    I am looking for the most efficient data aggregation techniques to monitor very big data which are produced by thousands of distributed sources. Important aspects are the communication cost and the scalability of the monitoring system.
    Mustapha Bouakkaz · Université Amar Telidji Laghouat

    My algorithm called TAG : Textual Aggregation by Graph

  • Lauro de Paula added an answer:
    How could I do to work with multiple GPUs using CUDA?

    For instance, if I have some Desktops, each one with a NVidia GPU, how could I do to divide a task between each node and, in each one divide between the CUDA cores?

    Are there any useful technique able to do that? Any suggestion of papers would be appreciated.

    Lauro de Paula · Universidade Federal de Goiás

    Hello, Christian! Thank you for your answer. I will read the text in this link.

    Hi, Joachim! I also thank you for your answer. My intention is only to divide a task between multiple GPUs and in each one divide between the cores.

  • Christian James added an answer:
    What is the best air distribution system for a cold storage chamber (for horticultural produce)?

    Let us consider a chamber with L x B x H of 18 m x 16 m x 4 m storing apples. What type of air distribution system (including locations of air inlet and outlet) should be used?

    Christian James · Grimsby Institute and University Centre Grimsby

    Providing the food has been chilled to the required storage temperature the purpose of a cold storage chamber is to maintain the temperature around the food.  The main heat load for foods such as apples during storage is that generated by their respiration.  However this is minimal compared to the initial cooling load, so the refrigeration system capacity does not have to be as large as that required for initial cooling.

    Forced air circulation systems are needed for removing initial heat, but high air velocities during storage can lead to high weight loss.  In addition, it's very difficult in practice to force air through products such as apples in bulk storage and using high velocities during storage can be very energy inefficient (see http://postharvest.tfrec.wsu.edu/pages/N1I4B).

    Personally I would recommend considering a passive cooling system for long term chilled storage of products such as apples.  

  • Riccardo Poli added an answer:
    Which pso method will be efficient for optimization of weak bus in distribution systems?

    evolutionary techniques

    Riccardo Poli · University of Essex


    I wrote a survey of the PSO Applications literature some years ago entitled "Analysis of the Publications on the Applications of Particle Swarm Optimisation" where I categorised PSO applications into groups. One of these groups (distribution networks) lists 7 or 8 publications relevant to your question. I'm not sure the article is in researchgate, but it was open access and so you can download it from the publisher's web site. I'm attaching a link.

    Hope this is a useful starting point.


  • John Machell added an answer:
    What are the effects of water storage and its impact on urban distribution system?

    There are severe water rationing practices that have resulted in consumers opting for storage than waiting for the next supply. This storage in turn has its effects on the water distribution system of a community.

  • Wed Kadhim asked a question:
    Energy systems automation

    . Identify challenges for applying “controller in the loop” simulation in distributed automation of energy distribution systems and suggest the ways of overcoming those challenges.

  • Stefan Gruner added an answer:
    What are aspects to test in distributed systems?
    I want to start my research on testing distributed systems. What are the testing aspects of distributed systems? Can anybody give me some reference from where i can start?
    Stefan Gruner · University of Pretoria

    -Deadlock, -Liveness, -Bottlenecks or single points of failure, -Balance of load, -Optimal routing of messages (shortest path problem), -Logical consistency of distributed or replicated sets of data, -Package loss rate or reliability, -Timing problems (real time versus logical time, Lamport's clocks), -Security aspects (interception, authentication, denial of service), -Software interoperability across different platforms, -Correctness of addressing and routing, etc...

  • Manuel De la Sen added an answer:
    Why is distributed control necessary in automation of energy distribution systems?

    Why is distributed control necessary in automation of energy distribution systems? How would you distribute the automation logic across many controllers?

    Manuel De la Sen · Universidad del País Vasco / Euskal Herriko Unibertsitatea

    An intuitively described  example of decentralized control related to electrical power  distribution is given in

    -Decentralized Control (North-Holland systems and control series)-
    , Madan G. Singh (Author)

  • Wanderson Paim de Jesus added an answer:
    How can we use different roles of controller in openflow protocol?
    1.2 gives 3 roles for controllers: master, equal, slave. The most acceptable for me is one master controller and several slave controllers. How about the following situations: several equal controllers and several slave controllers; a master controller, several equal controllers and several slave controllers? What are the applications of these scenarios?
    Wanderson Paim de Jesus · Universidade Federal do Rio Grande do Sul

    Chandler Wang,

    It is clear that there exists no one single and global solution for organizing a cluster of OpenFlow controllers. It will depend in your primary goal. If you're pursuing High Availability, maybe one Master and several Slaves is a good choice. This way, in case of failure, a Slave controller assume Master's position. Or else, in order to have a load balanced cluster, setting all controllers with Equal role may be part of an alternative solution. In this case, you must handle all synchronization issues and request distribution among controllers.

    After OpenFlow/SDN boom of interest, many traditional network equipment vendors and emerging start ups developed their own answers to SDN challenges. Certainly, a high available, balanced, secure and low cost solution for control plane issues competes for economic success in the global networking marketplace.


  • Muhammad Iqbal Hossain asked a question:
    What is functional abstraction?

    Can anyone discuss about functional abstraction.

    What is functional abstraction?

    why it is being used?

    how to create a formal definition of functional abstraction?

  • James Bryant added an answer:
    Is game theory a good tool to model interactions among tenants in a data center?
    Sometimes tenants can have competing business interests and could try to starve each other's tasks by strategically requesting resources
    James Bryant · Sheffield Hallam University

    The responses you have so far elicited (and your own question) assume that game theory offers a suitable framework for the analysis of such a situation.  Might I suggest that you consider instead the application of drama theory (for more information look perhaps at the archive of applications at www.dilemmasgalore.com) which in my view offers a much more realistic approach for practical interventions

  • Étienne Michon added an answer:
    Which simulation tool can simulate video streaming in P2P network?
    I'm trying to simulate p2p video streaming to evaluate the scalability of a network
    Étienne Michon · University of Strasbourg
    SimGrid can simulate milions of P2P nodes on a single physical machine. There is a comprehensive documentation and 101!
  • Muhammad Iqbal Hossain added an answer:
    Component interaction in Distributed System?
    We know components of Distributed System located at networked computer. Can anyone suggest me how these components interact between themselves and what can be testing aspect of that?
    Muhammad Iqbal Hossain · Kyungpook National University
    @Ricardo...can you tell me what are the aspects of interaction on distributed system? I am concern about functional properties.
  • Jonas Mellin added an answer:
    Does anyone have experience of using SensorML for home environments?
    SensorML is a standard for specifying sensor data processing, e.g., information fusion, sensor fusion, data fusion. In http://www.opengeospatial.org/, they are working on SensorML (as well as TransducerML).
    Jonas Mellin · University of Skövde
    Interesting, do you have more information such as web page or publications?
  • Peter T Breuer added an answer:
    What is the difference between forward slicing and taint analysis?
    I know that forward slicing is only for static program analysis while taint analysis can be used for both static and dynamic program analysis. Can anyone tell me the difference for static analysis? I can only see that both of them can explore all the variables influenced by the input.
    Peter T Breuer · Birmingham City University
    Well, I want your abstraction, i.e., "a summary". Examples can't tell one much when what they are an example of is not stated.

    What you say above appears to be roughly "forward slicing", except that your example does not make it clear if you think the slice consists of lines of code, or variables, or variables and line numbers. At a guess, you are thinking that a forward slice is the lines of a program that contain variables that are dependent on a certain input, but you also include in the lines variables which do not depend on that input. That's a bit strange for my mental image of slicing. But I definitely think you think a slice is an (adulterated) program (you have used the word "slices", not "slice", but I think you mis-wrote). It seems to contain the "whole lines" that contain variables that are dependent for their value on the input in question, according to some kind of analysis (unspecified).

    Is that right?

    Now you have also not described what you mean by taint analysis. Instead you have given an example .. you really will have to be more forthcoming about what you mean (and I also don't know what you mean by "similar" - do you mean "exactly the same"?)

    Am I to suppose that the result of your taint analysis will also be an adulterated program? Or this time will it only be a list of variables on each line that are dependent on the input of interest? "Variable instances" as the phrase goes. You need to tell me what your domains are.

    One thing I find interesting in your example is that you have included a line in the branch of the conditional in your program listing for forward slicing. The conditional test is dependent on the input but the variables in the line in question are not. Can you tell me your reasoning?

    Am I to assume that whatever analysis you applied for "forward slicing" was clever enough to reason that if it depends on the input whether or not the line in question is even executed or not, then what is written by that line is dependent?

    Or would your analysis be even cleverer, and deduce that the value of z in the branch is always the same whenever the branch is reached, independent of the input? Thus not include it in a slice, but instead include a line after a join of the two branches in which z appeared?

    If so, would your definition of "taint checking" agree with the analysis there?

    Do either or both of your analyses err on the conservative side? Do they underreport or overreport? Or is one or the other of them exact with respect to the real program semantics?

    And if, for example, you had something like z= x*0+y, would z be considered dependent on x by your forward slice? By your taint checking?

    And what if z was modified in the same way on both sides of the branch after the conditional? Would your forward slice include both lines (or both instances of z, if that is what your output domain is). Or only one? Or neither?

    And what would your taint checking include there?

    Really, it is very important that you describe _abstractly_ the output domains for your two techniques, as well as the kind of analysis and/or semantics that you are applying. I don't want an "example"! Abstraction is the work you are supposing to be doing :-).
  • What are the main motivations and obstacles to use a Cloud system as a platform to run hard realtime systems with a strict predictability requirement?
    What would be the advantages of employing a Cloud system as a platform to run a hard real-time system where timing predictability is important as the correctness of the system? In this case, what are the main challenges? As an example, guaranteeing timing requirements on the internet is one of the prominent challenges that is discussed in several papers but how about other challenges?
    Ra'ed (Moh'd Taisir) Masa'deh · University of Jordan
    Building on-time digital dashboards is a main concern.
  • Wei-Chih Ting added an answer:
    What techniques are used in the detection and identification of performance bottlenecks in n-tier applications from a third-party perspective?
    Deploying enterprise applications using a tiered architecture is increasingly becoming a standard both in virtualized and non-virtualised environments. A major challenge in managing such applications is the ability to be able to quickly detect performance bottleneck/anomaly/surprises and identify which of the tiers is causing the perceived constriction.
    The interest here is knowing what techniques have been used, or are being used for such tasks, especially in virtualised environment where applications are deployed in black-boxes (VM), and the infrastructure provider can only detect these changes and identify affected tier only by externally observing the performance of the applications.
    In this case, how can one use external system level performance "vital signs" of virtual machines to be able to quickly detect anomalies, and identify affected tier?
    Wei-Chih Ting · Industrial Technology Research Institute
    Let's make an example for better discussion.

    If it's a 3-tier architecture, the first tier consists of VMs dealing with client requests. All requests are then pushed into a queue like SQS or rabbitmq and then processed by the 3rd tier (scalable VMs).
    You can measure the average response time (ART) of the first tier. If the first tier is working fine, ART should be evenly distributed over time. If it's rising, add more VMs to the first tier (or fix the reverse-proxy) and keep measuring.
    For the second tier (the queue), you can measure time in queue (TIQ). If it goes up, add more VMs at the 3rd tier.

    If you can't touch any of these components (really black-boxed),
    one way is to randomly generate clients all over the world and measure ART and see if it's unbalanced.
  • Fabrice Clerot added an answer:
    How can a Petri net be used to model a distributed system?
    Can anyone recommend some papers or articles on this topic.
    Fabrice Clerot · Orange Labs
    A gentle introduction :
  • Madhavi Vaidya added an answer:
    Can I make use of R language for large data analysis?
    If yes, how is it different from MapReduce? What else can be used?
    Madhavi Vaidya · Vivekanand Education Society College of Arts, Science and Commerce
    Dear Alan, surely will check the link given. Thank you so much.
  • Frédéric Pinel added an answer:
    What are the current trends in SOA for embedded systems?
    In recent years service-orientation has become a serious architectural pattern for distributed embedded systems. Projects like DPWS have become quite popular. What do you think are the recent trends in this area and what may a research roadmap for SOA in the embedded world contain?
    Frédéric Pinel · University of Luxembourg
    Hello Marco,
    The idea of SOA is not only applicable to IP-networks. However, complying to standards makes some things easier, as in Netflix's deployment of new versions of a service in the cloud (but I cannot say that the numerous SOA standards are a good thing). The same ideas and benefits (such as separation of concern) could be applied to the embedded world. The automotive world probably has standards of it's own. This is also similar to the Erlang approach, where applications are composed of multiple services, with failover mechanisms (although Erlang is an integrated system, unlike SOA).
  • Ivan Simecek added an answer:
    How universal is the tuning of MPI programs with respect to topology and architecture?
    I'm asking those who have experiences with MPI working on various environments.
    It can be used on simple Beowulf clusters, consisting of four PCs connected to a single switch (or even hub), where all info can be broadcasted (or multicasted) to all hosts. It can also be used on large supercomputers of complicated topology, with large diameters.
    It seems various cases need different program optimization don't they?
    In particular - are the collective operations efficient on supercomputers, like BlueGene? AFAIK, modern algorithms (like SUMMA) often use MPI_Reduce, at least...

    What are your experiences? How to design programs for MPI?
    Ivan Simecek · Czech Technical University in Prague
    For best performance, you should design your application as two-level (MPI as high-level, OpenMP as low-level of paralelism). On the low-level, you can use all features of shared memory (including mutexes, critical sections and so on). You should balance carefully the number of threads per one MPI node (it is usually not equal to the number of cores!). And then you should consider the influence of mapping of data among MPI nodes:
    a) if the data are local (with small distance between MPI nodes), it usually implies that you should use point-to-point routines.
    b) If the data are not local (with high distance between MPI nodes), it usually implies that you should use collective routines.
    There are some additional optimizations: you can hide (some part of) latencies caused by MPI routines by clever usage of non-blocking version of these routines and so on.
  • Hassan Abedi added an answer:
    Concurrency Vs Parallelism
    What is the difference between Concurrent Transmission/Processing vs Parallel Transmission/Processing ?
    Hassan Abedi · Iran University of Science and Technology
    Have a look please
  • James J Coyle added an answer:
    Can anyone think of a meaningful application for a multi-writer shared storage?
    A multi-writer shared storage is the one that allows two writers to modify the same storage object concurrently without imposing any locking. Usually in cases of write collision one of the write operations is visible and the other is hidden, meaning that no read operation will later observe this write. I just wonder if such behavior is useful for any existing applications. For example if it was that the shared storage would implement a file system then would it be OK if one writer would overwrite the changes of a concurrent writer on a single file?
    James J Coyle · Iowa State University
    What you are thinking about is done in distributed e-commerce.

    If you have web hosts on two continents, you don't want a failure in communication to stop both from taking orders.

    Often, each host has an independent inventory, so they do not need to communicate. But what happens in the rare event that someone wants to buy the last seat on an aircraft and you can't communicate with all the other servers?

    The answer is that you go ahead and sell, without the lock, and arbitrate what happens later when you can again communicate.

    If you both sold the same seat, you upgrade someone to first class, or you
    bump a passenger, or something in between.

    Alternatively, you could put those people on standby until you can communicate.
    This is like not committing the write in the database.

About Distributed Systems

Distributed computing is a field of computer science that studies distributed systems. A distributed system consists of multiple autonomous computers that communicate through a computer network. The computers interact with each other in order to achieve a common goal. A computer program that runs in a distributed system is called a distributed program, and distributed programming is the process of writing such programs.

Topic Followers (14806) See all