- Generated by
1.8.13
|
LineairDB
0.1.0
|
Configuration and options for LineairDB instances. More...
#include <config.h>
Public Types | |
| enum | ConcurrencyControl { Silo, SiloNWR, TwoPhaseLocking } |
| enum | Logger { ThreadLocalLogger } |
| enum | ConcurrentPointIndex { MPMCConcurrentHashSet } |
| enum | RangeIndex { EpochROWEX } |
| enum | CallbackEngine { ThreadLocal } |
Public Attributes | |
| size_t | max_thread = std::thread::hardware_concurrency() |
| The size of thread pool. More... | |
| size_t | epoch_duration_ms = 40 |
| The size of epoch duration (milliseconds). See [Tu13, Chandramouli18] to get more details of epoch-based group commit. Briefly, LineairDB concurrently processes transactions included in the same epoch duration. As you set the larger duration to this paramter, you may get a higher throughput. However, this setting results in the increase of average response time. More... | |
| ConcurrencyControl | concurrency_control_protocol = SiloNWR |
| Set a concurrency control algorithm. See LineairDB::Config::ConcurrencyControl for the enum options of this configuration. More... | |
| Logger | logger = ThreadLocalLogger |
| Set a logging algorithm. See LineairDB::Config::Logger for the enum options of this configuration. More... | |
| ConcurrentPointIndex | concurrent_point_index = MPMCConcurrentHashSet |
| Set the type of concurrent point index. See LineairDB::Config::ConcurrentPointIndex for the enum options of this configuration. More... | |
| RangeIndex | range_index = EpochROWEX |
| Set the type of range index. See LineairDB::Config::RangeIndex for the enum options of this configuration. More... | |
| CallbackEngine | callback_engine = ThreadLocal |
| Set the type of callback engine. See LineairDB::Config::CallBackEngine for the enum options of this configuration. More... | |
| bool | enable_recovery = true |
| If true, LineairDB processes recovery at the instantiation. More... | |
| bool | enable_logging = true |
| If true, LineairDB performs logging for recovery. More... | |
| bool | enable_checkpointing = true |
| If true, LineairDB performs logging for checkpoint-recovery. Checkpointing prevents the log file size from increasing monotonically. i.e, if this parameter is set to false, The disk space used by LineairDB is unbounded. You can also turn off enable_logging and use only Checkpointing. This configure gives the recoverability property called CPR-consistency [1]. CPR-consitency may volatilize the data of committed transactions at last the number of seconds specified at checkpoint_period; however, the persistence of the data before that time is guaranteed. More... | |
| size_t | checkpoint_period = 30 |
| It uses as the interval time (seconds) for checkpointing. The longer is the better for the performance but the larger interval time causes the increasing of log file size. More... | |
Configuration and options for LineairDB instances.
| CallbackEngine LineairDB::Config::callback_engine = ThreadLocal |
Set the type of callback engine. See LineairDB::Config::CallBackEngine for the enum options of this configuration.
Default: ThreadLocal
| size_t LineairDB::Config::checkpoint_period = 30 |
It uses as the interval time (seconds) for checkpointing. The longer is the better for the performance but the larger interval time causes the increasing of log file size.
Default: 30
| ConcurrencyControl LineairDB::Config::concurrency_control_protocol = SiloNWR |
Set a concurrency control algorithm. See LineairDB::Config::ConcurrencyControl for the enum options of this configuration.
Default: SiloNWR
| ConcurrentPointIndex LineairDB::Config::concurrent_point_index = MPMCConcurrentHashSet |
Set the type of concurrent point index. See LineairDB::Config::ConcurrentPointIndex for the enum options of this configuration.
Default: MPMCConcurrentHashSet
| bool LineairDB::Config::enable_checkpointing = true |
If true, LineairDB performs logging for checkpoint-recovery. Checkpointing prevents the log file size from increasing monotonically. i.e, if this parameter is set to false, The disk space used by LineairDB is unbounded. You can also turn off enable_logging and use only Checkpointing. This configure gives the recoverability property called CPR-consistency [1]. CPR-consitency may volatilize the data of committed transactions at last the number of seconds specified at checkpoint_period; however, the persistence of the data before that time is guaranteed.
Default: true [1]: https://www.microsoft.com/en-us/research/uploads/prod/2019/01/cpr-sigmod19.pdf
| bool LineairDB::Config::enable_logging = true |
If true, LineairDB performs logging for recovery.
Default: true
| bool LineairDB::Config::enable_recovery = true |
If true, LineairDB processes recovery at the instantiation.
Default: true
| size_t LineairDB::Config::epoch_duration_ms = 40 |
The size of epoch duration (milliseconds). See [Tu13, Chandramouli18] to get more details of epoch-based group commit. Briefly, LineairDB concurrently processes transactions included in the same epoch duration. As you set the larger duration to this paramter, you may get a higher throughput. However, this setting results in the increase of average response time.
Default: 40ms.
| Logger LineairDB::Config::logger = ThreadLocalLogger |
Set a logging algorithm. See LineairDB::Config::Logger for the enum options of this configuration.
Default: ThreadLocalLogger
| size_t LineairDB::Config::max_thread = std::thread::hardware_concurrency() |
The size of thread pool.
Default: LineairDB allocates threads as many the return value of std::thread::hardware_concurrency().
| RangeIndex LineairDB::Config::range_index = EpochROWEX |
Set the type of range index. See LineairDB::Config::RangeIndex for the enum options of this configuration.
Default: ROWEX (epoch-based read-optimized write-exclusive index)
1.8.13