As proof-of-work blockchains are inherently energy greedy and offer probabilistic guarantees, permissioned distributed ledgers based on the classic deterministic consensus of the distributed computing literature appear as promising solutions to record securely the information produced by a large amount of connected Internet of Things (IoT) devices. As of today, it remains unclear whether these solutions can scale to large networks. In this paper, we evaluate the performance of prominent distributed ledgers using classic consensus protocols. We select five mainstream distributed ledgers, namely Ripple, Tendermint, Corda and v0.6 and v1.0 of Hyperledger Fabric, and evaluate both the throughput and the latency of clusters with different numbers of nodes (ranging from 2 to 32) for each of them. Our results show that, while offering sometimes thousands of transactions per second throughput, their performance usually does not scale to tens of devices as it drops dramatically when the number of devices increases. This study motivates the need for alternative solutions that solve the Blockchain consensus problem, a scalable variant of the classic consensus problem dedicated to blockchains.