Questions tagged [distributed-systems]
Questions about the challenges of solving problems with multiple cooperating but separate agents.
405
questions
0
votes
0
answers
19
views
Can we make Paxos live with a failure detector in W?
I was studying Paxos and Failure Detectors, and it was asked whether a failure detector W could be used to make Paxos live, but I can't figure it out.
Any suggestions?
- 1
4
votes
0
answers
35
views
Prior or subsequent art for Microsoft COM’s “causality IDs”
Microsoft COM, in its object RPC part, had a peculiar mechanism that could be used to prohibit concurrent calls to the same object (or another object serviced by the same event loop) but permit ...
- 171
3
votes
0
answers
28
views
Why is there no self-stabilizing determinsitic algorithm for vertex coloring in general graphs?
This question considers the design of a deterministic self-stabilizing algorithm for vertex coloring in uniform anonymous networks. Uniform anonymous networks do not have distinguished nodes and all ...
- 171
0
votes
2
answers
51
views
What is the proof in FLP for statement that "clique with majority of processes exist" if majority of the processes live?
The FLP paper provides proof in section "4. Initially Dead Processes" that consensus is possible if majority of processes live and no process dies during the consensus protocol. The proof ...
0
votes
0
answers
17
views
Algorithm to balance candies among students
Let's say N students
S1, S2, ... SN
with their respective candies count
c1, c2,... cN
The ask is to distribute the candies (...
- 101
0
votes
0
answers
19
views
Are there free/paid DISTRIBUTED solvers for a large system pf linear equations?
I've been doings some research into distributed solutions for a large system of linear equations (reading papers such as Wang et al.'s "Solving a system of linear equations: From centralized to ...
- 1
0
votes
0
answers
43
views
Algorithm for leader election in synchronous ring with a known network size $n$ with phases of length $\frac{n}{m}$
Consider the following leader election algorithm election of a synchronous ring with a known network size $n$ (simultaneous wakeup and uni-directional communication):
The leader is the node with the ...
- 315
2
votes
0
answers
87
views
Finding a connected subset of edges with minimal sum of weights for a distributed network
Let there be $n$ processes in a general connected network with no leader such that each process has a distinct UID.
Each edge has a weight , with weights being positive or negative.
Assume that each ...
- 232
0
votes
0
answers
18
views
Write-Ordering Problem, Idempotency in Distributed File Systems
I have recently finished reading the section on Distributed Systems in OSTEP. For NFS, they briefly mention the cache consistency problem and say how it is solved for reads by maintaining an attribute ...
- 171
0
votes
0
answers
21
views
Color a a general graph with maximal degree $\Delta$ using $2^{O(\Delta)}$ colors within $\log^{*}n$ rounds
Consider the following algorithm $A$ to 6-color an rooted tree within $\log^{*}n$ rounds in a distributed system:
1: Assume that initially the nodes have IDs of size $\log(n)$ bits
2: The root is ...
- 315
1
vote
0
answers
40
views
Networks and data flow - graph algorithms for propagating updates from nodes correctly
Suppose I have an acyclic directed graph of Nodes which subscribe to Events. When an Event callback is activated for some Node, the Node's internal update() method is called. Then, because the Node ...
- 837
1
vote
1
answer
127
views
How does non-DMA transfers really work?
I recently discussed DMA and non-DMA with my OS professor. Here is my current understanding:
disk controller has its own CPU, maybe own ISA, tiny program that simply handles reading from the disk (...
- 171
0
votes
1
answer
42
views
Ticket Dispenser algorithm and the size of the Ticket array
The following is a ticket Dispenser Mechanism, it's from the article:
"Closing the Complexity Gap between FCFS Mutual Exclusion and Mutual Exclusion By Robert Danek and Wojciech Golab"
http:/...
- 101
0
votes
1
answer
37
views
Universal lower bound of the multi message problem
The multi message problem is:
Let there be an undirected graph $G = (V,E)$ with $n$ vertices, and let $r \in G$. The algorithm sends a message $M_i$ of size $\Omega(\log(n))$ to each vertex $v_i$ ...
- 315
3
votes
1
answer
31
views
Logics for multi-agent and distributed systems and algorithms
The various temporal logics are particularly suitable for the specification and verification of concurrent computer programs. Are there logics that are particularly suitable for the specification and ...
- 237
0
votes
0
answers
29
views
Fault-Tolerant Distributed Work Sharing
I have a network of 1,000 computers that are sharing processing jobs. There are a few constraints:
Each computer has a maximum ratio of jobs it can perform
Can be higher than an equal share
...
- 101
0
votes
2
answers
40
views
How to interpret the concept of adjacent initial states in the FLP impossibility paper?
Link: https://dl.acm.org/doi/pdf/10.1145/3149.214121
In the proof for Lemma 2, the authors define two initial states to be adjacent if the only difference between them is the value in the input ...
- 178
0
votes
0
answers
28
views
What tree-based barrier is OpenMPI using for their barrier implementation?
I was taking a look at the barrier implementation for OpenMPI and see that they are using a sort of logN tree-based algorithm, but I'm having some trouble grokking the implementation.
Seems like a ...
- 101
1
vote
1
answer
28
views
Reducing BB to SMR and vice versa
It's often said that State Machine Replication and Byzantine broadcast are equivalent (e.g., the abstract of [1]), but it seems only obvious how to reduce SMR to BB. I.e., give BB one can construct an ...
- 191
0
votes
0
answers
31
views
Difference between Distributed File System, Cluster File System and Parallel File System
On the internet, I am unable to find concrete definitions of these three types of file-systems. Can someone clearly explain the difference between these?
1
vote
0
answers
37
views
Can a distributed system designed with logical clocks truly derive a total ordering of events?
I’m reading Leslie Lamport’s white paper, “Time, clocks, and the ordering of events in a distributed system”. His explanation for total ordering has me a bit confused in the following excerpt.
I've ...
0
votes
0
answers
25
views
$(d+1)$-coloring $(k, d)$-layered graph using a synchrounous distributed algorithm
Let $G=(V, E)$ be an undirected graph and $m$ be the maximal degree of G.
Each vertex v $\in$ V has an initial unique identifier (id) taken from the set {$1,...,n^2$} (when $|V|=n$).
Each vertex v $\...
- 181
0
votes
0
answers
24
views
Analyzing Parallel Matching Algorithm - Why it takes O(n+m) time and work?
Using the algorithm provided by this paper, they said that:
The algorithm defines a single phase of the local max algorithm. Each step of the phase takes at most O(log(m + n)) = O(logn) time and O(n +...
2
votes
1
answer
92
views
Information sharing in group
I have a group with n members. These are distributed members. It is basically a group chat. A new member is added to the group and the existing members are informed about this.
Now I want the new ...
- 123
1
vote
0
answers
64
views
For what example would Maekawa's algorithm allow out of timestamp order access to the critical section
For what example would Maekawa's algorithm allow out of timestamp order access to the critical section. It is mentioned that ordering is not satisfied in Maekawa's algorithm. But in what scenario ...
- 11
0
votes
2
answers
31
views
Chord Ring with limited table size of 3
In the normal case of a chord ring the big O notation of the look up is O(logn) because of long haul pointers of the Finger Table (or Routing Table).
In this question what if the Finger Table has a ...
- 111
3
votes
0
answers
31
views
Proof that state-based (convergent) CRDTs are causally-consistent
Conflict-Free Replicated Datatypes (CRDTs) are a class of distributed data structures where replica states are allowed to temporarily diverge, but will eventually converge (in the absence of network ...
- 263
0
votes
1
answer
30
views
Split matrix by groups of columns but *capture all combinations of X columns* for some X
Say I have a big matrix, ~50000 rows, ~80000 columns. I want to split it up and solve subproblems on different machines (horizontal scaling).
But I need to make sure every column can be combined with ...
0
votes
1
answer
29
views
Reference request: Time and proofs of a shared past
(Short discaimer: I'm a mathematician by education (category theory, algebraic geometry) and so mostly unaware of how different fields in CS relate to each other. I'd be very happy to just get some ...
- 103
0
votes
1
answer
44
views
Where is found the original library of the PBFT consensus algorithm?
In the PBFT paper written by Miguel Castro it has been mentioned:
(In Abstract:)
"BFT has been implemented as a generic program library with a simple
interface."
(Also in page 71:)
"...
- 145
1
vote
2
answers
154
views
Derivation of fraction of time when system has k requests
I was going through Distributed Systems by Maarten van Steen & Andrew S. Tanenbaum. While going through size scalability of systems, I came across this in a note.
I want to know the derivation of ...
- 33
0
votes
1
answer
49
views
How to handle Insert and Delete requests arriving out of order?
Entries need to be stored in a database. There is Insert(entry) RPC, which adds a new entry in a database, and Delete(entry) RPC,...
- 123
0
votes
0
answers
67
views
How does PBFT tolerate Byzantine Leaders?
It is said that PBFT (or Practical Byzantine Fault Tolerance consensus) is Byzantine fault tolerant (unlike Paxos or Raft that are only Crash fault tolerant.)
Based on this answer : LINK , in PBFT &...
- 145
0
votes
1
answer
52
views
How does caching increases scalability in distributed file system?
There are clusters of LAN connected computer systems. California cluster has cached "viral video" at first stage, when client near California requested the file firstly.
Now, another client ...
- 3
1
vote
1
answer
32
views
A lower bound for the makespan of heterogenous fog nodes
Why there is a sigma in the denominator of equation (8) in the picture? suppose we have n tasks and m fog nodes.
- 25
1
vote
0
answers
22
views
How to select a random node when nodes are untrusted in a distributed network? [closed]
Assume in a distributed network, where the nodes are not trusted and are identified by their public keys, we intend to select one of them in a random process. In such a situation, all of the nodes ...
- 145
1
vote
1
answer
78
views
can UTC replace Lamport logical clocks?
if UTC is universal, what drawbacks does it have in using it in a distributed system to coordinate events or use it as a global objective reference clock?
isn't it right to just use UTC to timestamp ...
- 111
3
votes
1
answer
328
views
Lamport Timestamps and Causality
I'm having trouble understanding lamport timestamps in practice and how they guarantee causal ordering.
Definitions
Lamport defines the "happens before" relationship in his paper. He states ...
- 133
1
vote
0
answers
32
views
Dot symbol in set-builder expressions in CRDT literature?
In Conflict-free Replicated Data Types: An Overview and other CRDT-related publications by Preguiça's group, a dot symbol often appears without definition. What does it indicate, how standard is it as ...
- 11
0
votes
1
answer
139
views
Consistent Hashing Algorithm Without Distribution / Load Balancing
I need some help finding or creating a consistent hashing algorithm with the following properties:
Given N buckets, only distributes keys to bucket N.
When number of buckets are increased from N to N+...
- 61
0
votes
1
answer
30
views
Difficulty understanding how federation increases cache performance for databases?
I am studying system design for distributed systems and in this page (https://github.com/donnemartin/system-design-primer), one of the following advantages was mentioned for federation for databases ...
- 125
0
votes
1
answer
41
views
a question related to consensus number of a concurrent object
a) We know that solving consensus in asynchronous distributed systems is impossible.
b) Let the consensus number of a concurrent object be cn = 2, which means that two processes can solve consensus ...
1
vote
0
answers
49
views
Is there a distributed streaming algorithm to verify set cover?
I have $k$ sets of similar sizes, that cover a universe $U$.
e.g. for $k=3$ and $U = \{1, 2, 3, 4, 5, 6\}$:
$S_0 = \{1, 2, 4\}$
$S_1 = \{2, 3, 4\}$
$S_2 = \{4, 5, 6\}$
I have another larger set $C$ ...
- 123
2
votes
1
answer
137
views
How do distributed joins work in a distributed relational database system?
I have been looking around for a few days trying to find a clear and concise description of how, at a technical/implementation level, how distributed joins work, but haven't found much. The best so ...
- 2,163
1
vote
1
answer
30
views
Term for the total communicated data over a distributed system
I am looking for a publication-friendly term to describe the total quantity of data communicated between nodes of a distributed system, as invoked by some discrete operation. This term would describe ...
- 111
0
votes
0
answers
40
views
Why can't we solve the consensus problem by just enforcing a new leader?
I am going through this lecture series by Martin Kleppman. In this video at around 1:25, he says you can manually configure the distributed nodes to chose a leader.
If that's the case can't we just ...
- 101
3
votes
1
answer
42
views
How does a single-node system get Availability in CAP theorem?
I have a question. I read many blogs, websites about CAP theorem. They say that single-node systems are CA, but how can it be "A" if that single node goes down ? Because if it does, the ...
- 31
1
vote
2
answers
47
views
What are good algorithms for distributed purchasing?
I'm looking for some algorithms/patterns to solve a distributed systems problem.
Lets say I have created a trading platform buying stock (I haven't), and I have a distributed system of 50 nodes that ...
- 13
0
votes
1
answer
18
views
Does pattern matching require having the data on one machine?
Is it possible to use pattern matching algorithms such as FP-Trees and others on data that resides in multiple machines?
Basically when doing pattern matching do you have to have all your data on the ...
- 251
0
votes
0
answers
74
views
Does DHT (Distributed Hash Table) allow to list ALL (not only nearest) nodes?
As I know any local DHT keeps only some of nodes - if the network is small/very small then all nodes, otherwise only some of them (nearest?).
So, when I checked different Golang implementations of DHT ...
- 101