[
https://issues.apache.org/jira/browse/IGNITE-18350?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
Vladislav Pyatkov updated IGNITE-18350:
---------------------------------------
Description:
We have two collections, those store lock modes:
{code}
HeapLockManager.WaiterImpl#locks
HeapLockManager.WaiterImpl#intendedLocks
{code}
That is done for convenient way to implement all lock manager requirement. But
for optimization, we can do this through _BitSet_. The redesign is possible
because we have predefined lock modes (_LockMode_) and attempts to accesses to
the modes are bounded. The amount of attempts significantly less than MaxInt
value.
The optimization allows using less memory for lock on each entry.
was:
We have two collections, those store lock modes:
{code}
HeapLockManager.WaiterImpl#locks
HeapLockManager.WaiterImpl#intendedLocks
{code}
That is done for continent way to implement all lock manager requirement. But
for optimization, we can do this through _BitSet_. The redesign is possible
because we have predefined lock modes (_LockMode_) and attempts to accesses to
the modes are bounded. The amount of attempts significantly less than MaxInt
value.
The optimization allows using less memory for lock on each entry.
> Redesign two ciollections of lock modes to a bitset into the heap lock manager
> ------------------------------------------------------------------------------
>
> Key: IGNITE-18350
> URL: https://issues.apache.org/jira/browse/IGNITE-18350
> Project: Ignite
> Issue Type: Improvement
> Reporter: Vladislav Pyatkov
> Priority: Major
> Labels: ignite-3
>
> We have two collections, those store lock modes:
> {code}
> HeapLockManager.WaiterImpl#locks
> HeapLockManager.WaiterImpl#intendedLocks
> {code}
> That is done for convenient way to implement all lock manager requirement.
> But for optimization, we can do this through _BitSet_. The redesign is
> possible because we have predefined lock modes (_LockMode_) and attempts to
> accesses to the modes are bounded. The amount of attempts significantly less
> than MaxInt value.
> The optimization allows using less memory for lock on each entry.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
