Any idea about hardware configuration for a production SicDB


#1

Hi,

Do have any best practice suggestions for choosing hardware for a production SciDB, regarding to RAM, CPU, and Disk, and how many instances can be configured on the same physical server.

Regards,


#2

Hello,

Please take a look at this blog post for example:
paradigm4.com/2013/04/what-d … look-like/

You may find some other interesting info in the other blog articles. Cheers.
-Alex Poliakov


#3

Should we use a dedicated physical server for the cordinator instance?
And what’s the best practice for hardware configuration for the cordinator instance?


#4

No, the instances are heterogeneous. The coordinator should be about the same size as the other instances.


#5

I mean does the cordinator instance have more work to do, should we provide more phyical resources for it, will it affect other instances on the same physical machine?


#6

Best practice is to treat the server that hosts the coordinator instance the same as the ones that host the worker instances. The coordinator instance does not do a significant amount of “extra” work.