You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Is your feature request related to a problem? Please describe:
During the ingest phase, partitions block writes. Through testing, this time is approximately 30 seconds, while the rocksDB ingest time is around 200 milliseconds. The main reason is that during the ingest phase, the time interval for the meta to send RPC_BULK_LOAD is too long (once every 10 seconds).
Single - node ingest concurrency level
Table - level ingest time (ms)
Minimum partition - level ingest time (ms)
Maximum partition - level ingest time (ms)
Average partition - level ingest time (ms)
Highest Rocksdb ingest time (ms)
Average Rocksdb ingest time (ms)
8
470006
15007
30075
27244
618
459.06
16
240005
19978
30002
28652
580
482.43
100
40004
29967
29986
29978
620
509.65
Describe the feature you'd like:
reduce partition ingest time
The text was updated successfully, but these errors were encountered:
The main reason is that during the ingest phase, the time
between meta's RPC_BULK_LOAD transmissions is too long
(once every 10 seconds). Shortening this time interval
can cut down the partition write-blocking time.
Feature Request
Is your feature request related to a problem? Please describe:
During the ingest phase, partitions block writes. Through testing, this time is approximately 30 seconds, while the rocksDB ingest time is around 200 milliseconds. The main reason is that during the ingest phase, the time interval for the meta to send RPC_BULK_LOAD is too long (once every 10 seconds).
Describe the feature you'd like:
reduce partition ingest time
The text was updated successfully, but these errors were encountered: