Our Community is getting an upgrade! To get everything ready for the relaunch, we’ll be placing the site in read-only mode starting September 21st.
We really appreciate your understanding while we get things set up behind the scenes. Catch up on all the exciting details about the move here.
Need help or have questions? Drop us a line at [email protected]
Created 10-25-2016 12:07 PM
We have two types hosts.
Type 1: 64 G memory , 11*900G disks
Type 2: 1T memory, 2*500G disks
The interconnect network is 10GE.
So is following deploy architecture reasonable? Is there any potential risk? Any recommendations? Thanks
DATANODE only on Type 1 hosts
RegionServer also only on Type 1 hosts
NODEMANGER only on Type 2 hosts.
others masters are also on Type 2 hosts.
Created 10-26-2016 08:21 AM
typically you'd deploy worker (nodemanager) and storage (datanode) service on same nodes to take advantage of having compute close to the data.
Created 10-26-2016 08:21 AM
typically you'd deploy worker (nodemanager) and storage (datanode) service on same nodes to take advantage of having compute close to the data.
Created 10-27-2016 07:34 AM
Hi Artem, Thanks for your quick response.
Currently, we consider to add some disks to Type 2 hosts(but it still less that the Type 1 hosts) and then add datanode on Type 2 hosts . Can this make the compute close to data? And the HDFS on both 2 types hosts with different size local file systems, Can this acceptable?
Created 10-27-2016 10:15 AM
that can work. What I mean compute close to data means both nodemanager and datanode should be on same node.