Allowed modifications to existing SciDB cluster?


What modifications can be made safely to an existing SciDB cluster? Can I change the number of workers? Add disks? Will this impact existing arrays?

Many thanks,


Currently, all changes to the SciDB cluster – such as adding or removing instances – require stopping and re-initializing the cluster. Thus, all array data would then be lost.


At the moment, SciDB doesn’t support the kind of elasticity you have in mind. It’s on the roadmap. We’re just very busy.

If you need to increase the physical resources you’re using in your SciDB installation, you’ll need to:

  1. Unload your data (use the OPAQUE unload mode, as it’s the fastest thing for this).
  2. Re-configure your SciDB installation, increasing the instance count,or whatever.
  3. Reload the data.


Suspected that this was the case, but wanted a definitive answer. Many thanks!



I have tested the latest version of SCIDB,

add online worker instance,
remove worker instance online.
reballance data among the cluster nodes.

but now not supported these basic cluster manage features.

when restart the whole cluster , you will initialize all the cluster configuration.