On 10/04/2019 10:38, Konstantin Knizhnik wrote:
I also a little bit confused about UNDO records and MVCC support in Zedstore. Actually columnar store is mostly needed for analytic for read-only or append-only data. One of the disadvantages of Postgres is quite larger per-record space overhead caused by MVCC. It may be critical if you want to store huge timeseries with relatively small number of columns (like measurements of some sensor). It will be nice to have storage format which reduce this overhead when it is not needed (data is not updated).
Sure. Definitely something we need to optimize.
Right now, even without UNFO pages, size of zedstore is larger than size of original Postgres table. It seems to be very strange.
If you have a table with a lot of columns, but each column is small, e.g. lots of boolean columns, the item headers that zedstore currently stores for each datum take up a lot of space. We'll need to squeeze those harder to make this competitive. Instead of storing a header for each datum, if a group of consecutive tuples have the same visibility information, we could store the header just once, with an array of the datums, for example.
- Heikki
