@anyx

• Testing the block_log is about testing the Performance of ChainBase which claims the following but to best my knowledge not tested.

Blockchain applications depend upon a high performance database capable of millions of read/write operations per second. Additionally blockchains operate on the basis of "eventually consistent" which means that any changes made to the database are potentially reversible for an unknown amount of time depending upon the consensus protocol used.

Existing databases such as libbitcoin Database achieve high performance using similar techniques (memory mapped files), but they are heavily specialised and do not implement the logic necessary for multiple indices or undo history.

Databases such as LevelDB provide a simple Key/Value database, but suffer from poor performance relative to memory mapped file implementations.

The code is here:

https://github.com/steemit/steem/blob/master/libraries/chain/database.cpp#L112

So we already know that specialized database RocksDB which originated from LevelDB, Which originated from Bigtable has write issues on Optane.

• Do we have data to match the performance and TPS claims of ChainBase on which Bitshares, STEEM and EOS is based on ?

This is the objective. Do you have the performance data of Chainbase On Optane ?

because there is no bottleneck during runtime, the system is mostly idle

I have seen the papers that you have on Google Scholar and elsewhere - We derive at conclusions based on tests. Do you have the tests / benchmarks and results ? I have been trying to get that but I haven't found that yet. If you have the benchmark data and corresponding TPS till which points you have tested and derived on the statement there is no bottleneck during runtime, the system is mostly idle please do share.

I am looking at the following:

• the amount of transactions the Blockchain was tested (in terms of TPS but not total number of transactions per day etc)

• Did we use a test mechanism similar to the formal ones like Caliper ?