The paper reporting our extensive comparison of full-graph and mini-batch GNN training systems was accepted at VLDB 2025. We found that mini-batch training systems achieve lower time-to-accuracy in all scenarios we considered and comparable accuracy. More interesting results in the paper!