Hi Dojo,
When running DataMiner on Cassandra it is recommended to have the local DB located on a different drive than DataMiner - I can see the benefits when dealing with physical drives and the advantage of following the same approach with a VM: I guess this option would also facilitate the migration from a local DB architecture to a Cassandra cluster (*) - correct?
In terms of performance, however, is there any actual difference when the DMA is on a VM and "C:" & "D:" might be just logical drives coming from the same SSD storage?
Thanks,
A.
(* = P.S.: Is the new tool for "Cassandra to Cassandra cluster" released?
Would it allow also to move from local Windows-Based Cassandra to a Linux based Cassandra cluster?).
On the OS level we keep the logical drive as c: for Dataminer and d: for Cassandra regardless of being a VM or physical server. A valid reason to keep this practice is to sector and organise the data storage on the OS level.
The recommendation is it to keep Cassandra and DataMiner in sperate drives however if you have a DM system with the Cassandra database still on the same logical drive with DataMiner it is possible to move it straight to Cassandra on remote cluster. The operation here would in practice be similar the moving Cassandra from c: to d: however instead we move it to a remote server with Cassandra installed on it and its required configuration.
Hi Alberto
Not necessarily this recommendation is to ensure optimal performance and storage capabilities of having the database on a different drive. On the other hand, it will not interfere the migration from a local DB architecture to a Cassandra cluster .
In terms of performance the VM’s need to be above the requirements to avoid issues especially for demanding maintenance tasks such as compaction and repair.
Please see the disk throughput on the dojo reference below:
Regarding virtual servers - DataMiner Dojo
The measurements can be taken with 3rd party tools.
Cassandra to cassandra cluster Migrator
Requirements :
Microsoft Windows. While the individual Cassandra nodes can be located on other (possibly Linux) machines, the tool must be run on a Windows machine.
Thanks for this first feedback, Joao
I have reviewed these links but I don’t fully understand the relevance of Passmark / CPU benchmark within the context of a virtualized system, so I’m back with even more questions 🙂
With reference to logical drives of a VM where disk throughput seems ok, can Cassandra be kept on the same logical drive as DataMiner (e.g. C:)?
Or should it be on a separate (logical) drive to facilitate migrations later on?
If to perform a migration in a Virtualized System
from
A) Cassandra+DataMiner on “C:”
to
C) Cassandra nodes on Cluster,
Is it required to go through the step
B) Cassandra on “D:” and Dataminer on “C:”
Or is it feasible to jump directly from step A to step C?
Any steer will be helpful.
Thanks Joao, this clarifies – I understand it makes sense to separate data & app also on a VM, even if the performance could be more affected by the disk troughput associated with the virtualization and the related logical drives.