Hello! We recommend using SSD for persistent clusters which will make what they call "Logical" fragmentation a non-issue: on SSD reading any block is roughly equally fast.
I guess that "Internal" fragmentation is still a thing. You should expect that Index pages are on average half full, and this is expected. However, in most of use cases indexes do not dominate memory usage so that's not that much important. And for data pages we try to avoid segmentation. Regards, -- Ilya Kasnacheev вт, 5 февр. 2019 г. в 13:42, Davalb <[email protected]>: > Hello, > > we use Ignite as a cache in front of our databases (SQL Server, Oracle) and > asked ourselves if the problem of index fragmentation, which is known from > SQL databases, could also occur here. (for example see > https://blog.devart.com/sql-server-index-fragmentation-in-depth.html). > > Is there this danger with in memory or persistent caches? How does Ignite > deal with this problem? > > This is especially interesting for tables that are constantly updated. > > Do we have to take appropriate measures for index maintenance such as Index > Reorg/Rebuild? > > Regards > David > > > > -- > Sent from: http://apache-ignite-users.70518.x6.nabble.com/ >
