Published on: 2013-11-27T20:46:50+00:00
Christian Decker is currently working on adding real-time stats to the webpage and is looking for time to implement access to the collector. He received a request from Mike Hearn to sort the snapshots by date, which he agreed to do. In response to Michael Gronager's suggestion to normalize the results with block size, Christian Decker stated that he will extend the calculations to include a size-normalized version. He also mentioned that given enough samples, they may be able to reconstruct some of the system parameters and attempted to correlate propagation speed and number of inputs/outputs.Gregory Maxwell requested Christian Decker to publish the block ids and timestamp sets for each block, which would be useful in correlating propagation information against block characteristics. To this, Christian Decker said that he's looking for a good way to publish these measurements as they are rather large in size. The conversation took place on the Bitcoin-development mailing list.A conversation between Michael Gronager and Christian Decker from 2013 about sorting snapshots by date gave rise to a discussion on real-time stats. Christian suggested extending calculations to include a size-normalized version, as well as attempting to correlate propagation speed and number of inputs/outputs to see if processing at the nodes has an influence. Michael suggested normalizing the results with block size and plotting the delay graph as e.g. normalized to the averaged blocksize or defining a standard block size to compare the plot between days. He also asked if the correlation of propagation times holds for transaction sizes as well.In a discussion on Bitcoin development, Christian Decker posted his measurement code from the Information Propagation paper and automated it to correlate propagation information against block characteristics. Gregory Maxwell requested that Decker publish the block ids and timestamp sets for each block to further analyze the data. However, Decker expressed concern about serving large files due to website capacity limitations but offered to provide the information on a per user basis if demand is not significant. Michael also contributed to the conversation, suggesting normalizing results with block size and finding correlations with transaction sizes.Christian Decker shared his measurement code on the Information Propagation paper and automated it as much as possible to create the Network Propagation page on bitcoinstats.com. This page takes a daily snapshot of the situation, calculates the time until blocks and transactions reach a certain percentile of the nodes in the network, and shows the density function describing at what times nodes learn about the existence of a block/transaction. Decker intends to add more information and plots over time, but he wanted to push this out quickly as there were some people asking for it. He mentioned that saving raw log data requires saving transaction timestamp data because whether or not a given node has a transaction already matters regarding propagation.A member of the Bitcoin-development mailing list requested graphs and raw data showing how propagation data changes over time, as well as instrumentation of bitcoind to discover where time is spent when relaying a block. Gregory Maxwell asked for block ids and timestamp sets for each block to aid in correlating propagation information against block characteristics. Christian Decker mentioned that he is trying to find a good way to publish some measurements, but doesn't want to serve them along with the website due to capacity limitations. He is currently looking for suggestions on how to publish them effectively.In November 2013, Christian Decker wrote an email regarding the measurement code from the Information Propagation paper. He automated the code as much as possible and mentioned that it would be useful to publish the block ids and timestamp sets for each block. The reason for this was to correlate propagation information against block characteristics, which came up during a discussion about the Cornell paper.Christian Decker, the author of the Information Propagation paper, has created an automated measurement code for network propagation, which takes a daily snapshot of the situation and calculates the time until blocks and transactions reach a certain percentile of nodes in the network. The result is available on the Network Propagation page on bitcoinstats.com. Christian Decker intends to add more information and plots over time, but wanted to push this out quickly as there were some people asking for it.
Updated on: 2023-08-01T06:43:42.093831+00:00