Row cache is a very useful feature since it is more efficient than block cache, but the key problem is how to invalidate expired rows.
Could you please share more technical details about how to make sure we do not read stale cache data? Thanks. Vladimir Rodionov <vladrodio...@gmail.com> 于2025年2月18日周二 06:16写道： > > Hello, folks, > > I’m a former *HBase contributor* who stepped away nearly five years ago. > For those unfamiliar with my work, I was behind *Incremental Backup and > Restore* and *Distributed MOB Compactions*. > > Since then, my focus has shifted to *data caching technologies*, leading to > the development of *Carrot Cache (a new Java caching library)* and *Memcarrot > (a Memcached-compatible caching server)*, both available here: > 🔗 *Carrot Data GitHub <https://www.github.com/carrotdata>* > > *Long story short*—I originally built the *HBase Row Cache* and even > presented it at an HBase conference. Recently, I gave it a *major refresh*, > updating dependencies to include *Carrot Cache* and the latest *HBase/Hadoop > 2.x*. > > 🔗 *HBase Row Cache Repository:* hbase-row-cache > <https://www.github.com/VladRodionov/hbase-row-cache> *(Please take a > look!)* > > HBase Row Cache is an implementation of the *original BigTable ScanCache*, > which is *still missing in HBase* (?). It is designed as an *HBase Region > Coprocessor*. > > *Question for the community:* > Would there be interest in bringing this into the *main source*? I don’t > want to spend time polishing a feature if there’s little demand. > > Looking forward to your thoughts! > > Best regards, > *Vladimir Rodionov*
