We have 7 identical physical servers (2x8 core CPU, 128GB RAM, 8x 6TB disks) that will be used for Hadoop. All of the machines are connected to 10G switch with double 10G interfaces. Since we do not have many machines we want to use one of the machines as both name node and data node. Is this configuration would cause any performance issue? The Hadoop system will be used for academic purposes, so would the data node on the name node have the same performance as the other data nodes? If this is not a good solution, what is the minimum requirements of a name node for this configuration?
Asked
Active
Viewed 1,992 times
1 Answers
2
Its not advisable to use NN and DN on the same machine as DN may get full and your NN may start having issues and in turn bring the whole cluster down!
If you are sure your DN will have enough space at all times then you may do it(not recommended).
Ani Menon
- 121
- 6