Apache introduced new version Hadoop 2.x to handle huge and big data to store and process. The following are some differences among features between Hadoop 1.x and Hadoop 2.x
| Feature | Hadoop 1 (MapReduce) | Hadoop 2 (YARN) |
|---|
| Scalability | MR has limitations on nodes. It's limited to 4K nodes per cluster | Yet Another Resource Negotiator (YARN) has more scalability compared to MR. YARN's scalability increased up to 10K nodes per cluster. |
| Single-Point-of-Failure (SPOF) | Single Point of Failure in Hadoop 1.x because of having sole Namenode. If it goes down all running jobs halted. | By having standby Namenode, YARN has overcome SPOF problem. Automatic recovery mechanism there in case of Namenode failure. |
| MS-Windows | MR does not support MS-Windows. | YARN supports MS-Windows. |
| Storage | Minimum Block Size is 64 MB | Minimum Block Size is 128 MB |
| Management | Hadoop 1.x supports only one namespace for managing HDFS filesystem | Hadoop 2.x supports multiple namespaces |
| Programming Model | Hadoop 1.x supports only one Map Reduce programming model. | Hadoop 2.x supports only one multiple programming models like MR, Spark, Storm etc. |
Comments