Nice approach, also good medium article.
I just want to mention, that the performance characterstics of TarMK are different from the characteristics of a clustered instance (based on Mongo). Not because Mongo is slower, but also there are optimizations which do not necessarily follow the same pattern as with Tar. For example you might get best performance when you do your 5000 changes in 10 batches (disclaimer: I haven't checked.)
Anyway, would love to see the outcome of this benchmark on a CS authoring instance.
Nice approach, also good medium article.
I just want to mention, that the performance characterstics of TarMK are different from the characteristics of a clustered instance (based on Mongo). Not because Mongo is slower, but also there are optimizations which do not necessarily follow the same pattern as with Tar. For example you might get best performance when you do your 5000 changes in 10 batches (disclaimer: I haven't checked.)
Anyway, would love to see the outcome of this benchmark on a CS authoring instance.