Data Mart > MPP > Configuration Requirements > Detailed Configuration Parameters

There are two configuration files in one distributed data mart system: local property configuration (Local Properties) and Global property configuration (Global Properties).

❖Local Properties Configuration

The local property configuration file is the necessary property file for the node of every machine. By default, it is saved at {bi.home}/bi.properties., and bi.home path is the directory relative to the installation path YH\Yonghong\bihome.

Attribute	Optional/Required	Description
dc.io.handlers=1	Optional	Defining the thread counts processing IO communication. Generally, one thread count is enough.
dc.io.channels=2	Optional	Defining the maximum Socket connections when communicating with node of other machine.
dc.io.ip=	Mandatory	Defining the IP of local machine, especially in the multiple networking interfaces. If it is not defined, the IP will be obtained by attempting from the operation system.
dc.node.types=mr	Mandatory	Define the node type of local machine, and m - Map Node, r - Reduce Node, n – Naming Node, c - Client Node. Generally it's the combination of above values.
dc.global.path=global_bi.properties	Mandatory	Define the configuration file path shared by the node of each machine.
mem.serial.mem=700	Mandatory	Define the memory size that can be distributed in batch to the memory calculation. The unit is M
mem.proc.count=2	Mandatory	Define the CPU number that can be used for memory calculation.
dc.block.units=4	Optional	Define the number of data cell in one data block. This data block is the physical file loaded or unloaded from memory. The data block is distributed to every Map node.
dc.unit.rows= 262144	Optional	Define the row count of one data cell. This data block will form a physical file to be distributed to each Map node.
dc.fs.naming.paths=	Mandatory	Define the file path of saving the metadata of Naming node. The file path can be multiple, and seperated by ';'. In this way, metadata file has higher security. Please input the absolute path. The default value is {bihome}/cloud/cloud/qry_sub.m
dc.naming.waiting=30000	Optional	Defining after Naming Node startup, for how long the available status can be switched back.
dc.naming.maps=1	Optional	Defining after the Naming Node startup, for how many alive Map node that available status can be switched back.
dc.naming.reds=1	Optional	Defining after the Naming Node startup, for how many alive Reduce node that available status can be switched back.
dc.naming.check.file=true	Optional	Defining after the Naming Node startup, whether the correct metadata should be ensured before the available status being switched back. The so-called correct metadata refers that all folders and files in the metadata are available.
dc.fs.sub.path=	Mandatory	Define the file path saving the metadata of Map node or Reduce Node. Please input the absolute path. The default value is {bihome}/cloud/cloud/qry_sub.m
dc.fs.physical.path=	Mandatory	Define the folder saving the physical data of Map node or Reduce Node. Please input the absolute path. The default value is {bihome}/cloud/cloud
dc.col.cache.count=20	Optional	Define the number of maximum memory caching of every storage type.
dc.data.debug=false	Optional	Define whether to output the debugging information of data.
dc.inverted.supported=false	Optional	Define whether to try to produce the column index to speed up the performance.
dc.inverted.ratio=3.1	Optional	Define the size of index of each row in average when trying to produce the index.
dc.buf.cache.count=10	Optional	Define the number of memory caching in the data buffering area used by communication
dc.float.frags=4	Optional	The decimal place saved in mart of single-precision floating point number.
dc.double.frags=4	Optional	The decimal place saved in mart of double-precision floating point number.
dc.mr.debug=false	Optional	When executing Map, Reduce task, type the execution progress of Map, Reduce in every 20s intervals.
dc.orderby.limit=500000	Optional	The maximum grouping number supporting sorting.
map.aggr.parallel=false	Optional	Whether to carry out parallel process to a zb file according to data fragment and Hash partition at Map end.
red.aggr.parallel=true	Optional	Whether to carry out parallel process according to Hash partition at Reduce end.
map.part.size=4	Optional	The number of hash partition at Map end.
red.part.size=32	Optional	The number of hash partition at Reduce end.
aggr.timeout=600000	Optional	The time-out period waiting for the end of relevant thread processing in the parallel processing.
parallel.min.groups=10000	Optional	The minimum grouping number that needs the parallel processing at Reduce end.

The default value is available for the system with “Optional” marked. The default value equals to the result of the first column.

❖Global Properties Configuration

The global property configuration file saves all the property files shared by all machine groups. It is saved at {bi.home}/global_bi.properties by default.and bi.home is the directory of corresponding installation path YH\Yonghong\bihome.

Attribute	Optional/Required	Description
dc.io.local=true	Optional	Standalone or multi-machine versions are marked. The default is local standalone version.
dc.cache.max=5242880	Optional	Define the maximum memory caching. If the quantity of data over the maximum is read/wrote, one time of physical read/wrote will be triggered.
dc.io.timeout=15000	Optional	Define the maximum waiting time of communication between nodes of two machines.
dc.io.block=131072	Optional	Define the caching size of Socket read and write.
dc.io.sport=5083	Optional	Define the ports of communication among nodes of each machine.
dc.io.fport=5066	Optional	Define the port of transferring files among nodes of each machine.
dc.node.naming=	Mandatory	Define the IP of Naming Node, if it is local standalone version, no need to define it.
dc.fs.dup=2	Optional	Define the number of copies of file system.
dc.update.period=15000	Optional	Define the cycle of heartbeat. For every cycle of heartbeat, Map/Reduce node shall send a report to Naming Node, to declare its survival.
dc.task.timeout=60000	Optional	Define the maximum time to complete one task. If the task is not complete over the maximum time, the system shall try to re-distribute the task.
dc.nodes.pin=	Optional	Define the Pin code needed by the communication among nodes. If Pin is null, the Pin will not be checked. The default value is null.
dc.doctor.repair=false	Optional	Define whether to restore the lost file.
dc.mismatch.remove=false	Optional	Define whether to delete the nonexistent zb file in Meta.
file.sync.interval=3600000	Optional	Define the time interval of metadata file of total quantity update.
global.data.timeout=600000	Optional	Define the time-out time of obtaining the dimension table.
zk.conn.timeout=120000	Optional	Customize the time-out of communication between client-side and ZooKeeper cluster node.
zk.conn.hosts	Optional	Customize the address from the client-side to ZooKeeper cluster. Multiple addresses are separated by commas, such as zk.conn.hosts=192.168.3.138:2181,192.168.3.138:2182,192.168.3.174:2181.
dc.use.backup=false	Optional	Customize whether the Naming backup mechanism is enabled.
dc.backup.max.bytes=1048576	Optional	Customizing after and enabling the Naming backup mechanism, maximum transportable log size from each Naming node to ZooKeeper.