Skip to end of metadata
Go to start of metadata
- Size required is 4 times source data for HPCC and this is just for THOR (If we are using SALT or typical SALT type use cases).
- Roxie will be ¼ of THOR count usually.. Because the data is compressed and we don’t hold any transient data in Roxie.
- THOR – each node can hold about 2.5 TB of data (MAX). So plan for THOR nodes accordingly.
- Recommended SAS drive for both THOR and ROXIE as its almost equal to SATA now.. Else SAS for at least ROXIE
- THOR replicates data (2 copies – This needs to be verified).. Looks like the only way to do it is
- 2 management nodes to be allocated for THOR.. the DALI, SASHA, DFU etc
- Roxie would like to keep most of its data in memory. So plan memory accordingly
- We need 1 DALI for every Roxie cluster.. Apparently that starts the ROXIE server
- Roxie +DALI needs to have a mirror. This is because, when you need to update indexes, you update the mirror and make that primary and bring the other one down. This is not really a necessity except for high availability and performance requirements.