D3-GNN: Dynamic Distributed Dataflow for Streaming Graph Neural Networks

Summary: D3-GNN is the first fault-tolerant, hybrid-parallel distributed system for online GNN inference/training over continuously changing graphs. Its unrolled dataflow and inter/intra-layer windows tame cascading updates, skew, and neighborhood explosion, yielding 76× DGL throughput. (summarized by gpt-5.6-luna on Jul 24 2026)

Paper ID: 13685
Venue: VLDB
Year: 2024
Pagerank: 5.4226e-05
Overall Rank: 8,454 | 42.00%
DOI: 10.14778/3681954.3681961

Incoming Non-self Citations Over Time

Authors

1. Rustam Guliyev (University of Warwick)
2. Aparajita Haldar (Fujitsu Research Europe; University of Warwick)
3. Hakan Ferhatosmanoglu (Amazon; University of Warwick)

BibTeX Citation

@article{guliyev_vldb24,
        title = {{D3-GNN: Dynamic Distributed Dataflow for Streaming Graph Neural Networks}},
        author = {Guliyev, Rustam and Haldar, Aparajita and Ferhatosmanoglu, Hakan},
        journal = {PVLDB},
        series = {{VLDB} '24},
        volume = {17},
        number = {11},
        pages = {2764--2777},
        doi = {10.14778/3681954.3681961},
        url = {https://doi.org/10.14778/3681954.3681961},
        year = {2024}
}

Incoming Citations (Sorted by Pagerank)

Showing 2 of 2 citing papers.

Rank	Citing Paper	Year	Venue	Pagerank
10,309	A Comprehensive Benchmark on Spectral GNNs: The Impact on Efficiency, Memory, and Effectiveness	2026	SIGMOD	5.093636e-05
10,620	Understanding Evolving Graph Structures for Large Discrete-Time Dynamic Graph Representation	2026	VLDB	5.093636e-05

Outgoing Citations (Sorted by Pagerank)

Showing 13 of 13 cited papers.

Citations counted here include only citations to other VLDB/SIGMOD/CIDR/PODS papers in this database.

Rank	Cited Paper	Year	Venue	Pagerank
223	AliGraph: A Comprehensive Graph Neural Network Platform	2019	VLDB	0.00024182473
264	The Ubiquity of Large Graphs and Surprising Challenges of Graph Processing	2018	VLDB	0.00022980015
927	Real-time Constrained Cycle Detection in Large Dynamic Graphs	2018	VLDB	0.00013161079
1,048	AGL: A Scalable System for Industrial-purpose Graph Machine Learning	2020	VLDB	0.00012433693
1,132	SANCUS: Staleness-Aware Communication-Avoiding Full-Graph Decentralized Training in Large-Scale Graph Neural Networks	2022	VLDB	0.00012041292
1,409	TGL: A General Framework for Temporal GNN Training on Billion-Scale Graphs	2022	VLDB	0.00010854808
2,209	Regular Path Query Evaluation on Streaming Graphs	2020	SIGMOD	8.9437338e-05
2,221	Streaming Graph Partitioning: An Experimental Study	2018	VLDB	8.9260308e-05
3,109	Managing Large Dynamic Graphs Efficiently	2012	SIGMOD	7.7482727e-05
4,686	Tornado: A System For Real-Time Iterative Analysis Over Evolving Data	2016	SIGMOD	6.5620646e-05
6,095	StreamWorks - A system for Dynamic Graph Search	2013	SIGMOD	5.9782101e-05
6,680	Beyond Analytics: The Evolution of Stream Processing Systems	2020	SIGMOD	5.8049764e-05
9,414	Model-Parallel Model Selection for Deep Learning Systems	2021	SIGMOD	5.2743893e-05

Semantically Similar Papers

#	Overall Rank	Paper	Year	Venue
1	7,374	ADGNN: Towards Scalable GNN Training with Aggregation-Difference Aware Sampling	2023	SIGMOD
2	6,630	Efficient Training of Graph Neural Networks on Large Graphs	2024	VLDB
3	4,878	NeutronStream: A Dynamic GNN Training Framework with Sliding Window for Graph Streams	2024	VLDB
4	1,863	ByteGNN: Efficient Graph Neural Network Training at Large Scale	2022	VLDB
5	4,980	Decoupled Graph Neural Networks for Large Dynamic Graphs	2023	VLDB
6	4,891	DGC: Training Dynamic Graphs with Spatio-Temporal Non-Uniformity using Graph Partitioning by Chunks	2023	SIGMOD
7	9,729	Scalable Graph Convolutional Network Training on Distributed-Memory Systems	2023	VLDB
8	5,760	DynaHB: A Communication-Avoiding Asynchronous Distributed Framework with Hybrid Batches for Dynamic GNN Training	2024	VLDB
9	3,751	G3: When Graph Neural Networks Meet Parallel Graph Processing Systems on GPUs	2020	VLDB
10	2,640	Scalable and Efficient Full-Graph GNN Training for Large Graphs	2023	SIGMOD