GitHub user yezhizi added a comment to the discussion: TimeSeries Proposal
Thank you very much for your valuable suggestions!
- My apologies for overlooking this point initially. I will explicitly specify
the particular Column Family (CF) to be used in the proposal.
- Given that time series data requires fundamentally different key formats
compared to other CFs, I believe introducing a dedicated CF (e.g.,
"TimeSeries") might be necessary.
- The encoding approaches used in search data implementations have provided
me with significant inspiration—thank you for pointing me in this direction! I
will revise the key format design in my proposal within the next few days.
- I haven't yet examined the compaction filter and compaction checker
implementations in detail. I will study these components thoroughly and enhance
the garbage collection section of my proposal accordingly.
- Yes. I'll keep this in mind. For the timestamp and value compression, we may
adopt the same algorithms used in Redis TimeSeries (as proposed in the [VLDB
paper](https://www.vldb.org/pvldb/vol8/p1816-teller.pdf)). Since this likely
won't be required in the initial implementation phase, we can discuss this
later. The other implementation aspects will be different from Redis TimeSeries.
GitHub link:
https://github.com/apache/kvrocks/discussions/3044#discussioncomment-13672210
----
This is an automatically sent email for [email protected].
To unsubscribe, please send an email to: [email protected]
