Dash underwent a network stress test, processing over double Bitcoin’s transactions without a performance decrease, validating long-term scaling plan.

Over the course of a couple of days, the Dash network underwent an impromptu network capacity test, with well over a half-million transactions over a 24-hour period. This test was done through use of a transaction tool that send out batches of transactions, and was initiated by community members without prior planning or notice to developers.

At present, Dash has 2MB blocks, compared to the roughly 1MB blocks of Bitcoin. Dash’s block interval is approximately 2.5 minutes, four times that of Bitcoin, so theoretically Dash currently possesses eight times the network capacity. Dash’s long-term scaling plan includes on-chain scaling up to 400MB blocks, and network viability has been tested extensively with university research up through 10MB blocks, which is roughly 40 times Bitcoin’s present capacity and over half of transactions that PayPal currently processes.

The Dash network continued to operate uninterrupted despite a massive influx on transactions

During the process of testing transaction capacity, the Dash network experienced no detectable disruptions. The network experienced zero orphaned blocks over the two days that the experiment ran, and the mempool backlog of unconfirmed transactions reached a maximum of 3.2MB, which was due to the irregularity of the test with thousands of transactions sent at once. This mempool size equaled under two blocks worth of data, meaning it would take the network five minutes to clear the data. By contrast, Bitcoin’s block interval is 10 minutes, meaning it would take at least twice the time to clear a backlog of any size. According to Benjamin Nitschke, development lead of the MyDashWallet project, the extra load on the network was barely noticeable:

“We didn’t notice anything [amiss], CPU went from 0.5% to maybe 3%, and if we hadn’t told anyone, others would have only noticed a few days later on transaction charts. For some reason ShapeShift stopped working for Dash and Poloniex stopped working as well, but these look unrelated. No reason to think their nodes crashed as they were not under stress at all (our Bitcoin node is at least 10% higher CPU and our Ethereum node is using 100x higher CPU for just following all transactions than what Dash does, even under stress).”

Part of this success is due to several optimizations of the code with the recent 12.3 release, which is mostly, though not completely, implemented by the network. According to Dash Core Lead Developer Udjinm6, optimizations in the new release were especially apparent during this experiment when contrasted with older 12.2 nodes:

“We noticed that 1-CPU 12.3 nodes kept running at 10-12% load, without any noticeable impact by the stress test. However, 2-CPU 12.2 node load fluctuated around 40-60%, with 1-CPUs being close to 100% most of the time.”

While the test shows the capacity of the Dash network to process significant transaction volume, it represents a mere fraction of the network’s present capacity with current infrastructure, which theoretically could handle millions of transactions right now. Nitschke noted that this is a mere fraction what the network can process:

“Dash was only “tested” to about 10% of its maximum daily capabilities. We had a bunch of almost full 2MB blocks, that was nice to see, but other than that we can only say: Dash works as advertised.”

Bitcoin, Litecoin experienced severe network disruptions at similar capacities

The experience of the Dash network’s scaling test stands in stark contrast with other major payments-focused blockchain networks, particularly Bitcoin and Litecoin. On December 14th and January 4th of 2017 and 2018 respectively, when Bitcoin’s daily transaction volume was at its highest point of between about 424,000 and 490,000, median transaction fees were above $14 and average fees were above $21, with both figures spiking as high as $34 and $55 respectively. Litecoin reached its daily transaction volume peak on the 4th of January at about 225,000, and hit a median transaction fee of about $0.27 on that day and following days, and an average fee of $0.50.

By comparison, Dash processed significantly higher transaction volumes than Bitcoin, and several times higher than Litecoin, during its test period, with no discernible change to transaction fees, which remained at 1/20th of a cent median and under two cents average. This clearly demonstrates the Dash network’s superiority at processing unexpected large numbers of transactions without disruption, solidifying the case for its use for payments over other competing networks.

Bitcoin Cash is pursuing a similar on-chain scaling strategy to Dash, and researchers in the two communities have collaborated on findings, though Bitcoin Cash’s network has yet to be tested at this level.