- Some node is non scalable or it is inefficient.
- Some node has gone down and it is not available.
- How the partitioning of the resources is being done.
- It might happen that the node has become unreliable.
- How waste processing is being done.
- Management of user logs and the resources it can develop.
How these scalability and high cluster utilization will be done? It can also find out that. How high availability for the confidence when we maintain. Then, it also develops a flexible resource model. We can use multiple data processing algorithms using yarn. We can also perform log aggression and resource utilization. The resource localization activities using this cluster management technology on top of yarn in Hadoop.
Hadoop YARN Components:
YARN components are having military components. There are three components of YARN. One is the resource manager. Second is the node manager and the last one is that application master.
Hadoop YARN Resource Manager:
The resource manager is the master daemon of YARN. It manages the CPU and memory related tasks for all the applications. The Resource manager deals with. How the resources or which the sources will be allocated to each application. Which memory to be occupied by which application. Resource allocations and memory allocations to the applications will be handled by YARN resource manager. t has two main different components. One is a scheduler, another is the application manager.
1- Scheduler:
Scheduler is not doing any kind of monitoring or tracking operations. The scheduler is used to allocate resources to the running applications. It is responsible to allocate resources to the running applications. The applications can get it smooth execution. But, it is not dealing with any kind of administrative, any kind of monitoring or tracking related operations. It does not perform any tracking or monitoring related jobs.
2- Application Manager:
Application manager is used to manage the running application master in the cluster. It is responsible for starting the application masters to track and restart on different nodes if needed. Starting of the application master and the starting of node. All these things will be falling under the responsibility of application manager. So, It is doing some little kind of tracking and monitoring. But. this scheduler is only for the allocation of resources to the running applications.
Hadoop YARN Node Manager:
The slave daemon is the node manager. It manages the usage of resources and send reports to the resource manager. Resource manager is the master demon. This particular node manager is the slave daemon. It is reporting to the master demon. More ever, The resource manager regarding the usages of the resources. It also tracks the health of the node. How the note is running? What is the current condition of node? So,that will be reported to the resource manager. It also supports plugging long running auxiliary services to the node manager. So, It is merely an application specific service.