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
Problem description
ReadIndexCache is supposed to cache the most recently used <start offset, chunk name> tuples in memory using LRU strategy.
During eviction the current implementation scans all cached entries for given segment which is inefficient.
Problem location
io.pravega.segmentstore.storage.chunklayer.ReadIndexCache::evictChunks
Suggestions for an improvement
Use Guava cache as underlying cache implementation.
Correctly handle cleaning up auxiliary data structures when entries are evicted. (Eg ConcurrentSkipList based per segment index)
The text was updated successfully, but these errors were encountered:
Problem description
ReadIndexCache is supposed to cache the most recently used
<start offset, chunk name>
tuples in memory using LRU strategy.During eviction the current implementation scans all cached entries for given segment which is inefficient.
Problem location
io.pravega.segmentstore.storage.chunklayer.ReadIndexCache::evictChunks
Suggestions for an improvement
Use Guava cache as underlying cache implementation.
Correctly handle cleaning up auxiliary data structures when entries are evicted. (Eg ConcurrentSkipList based per segment index)
The text was updated successfully, but these errors were encountered: