search

tango-controls / tango-doc

The official documentation for Tango controls
http://tango-controls.readthedocs.io
8 stars 27 forks source link

archiving #304

Open PhilLAL opened 5 years ago

PhilLAL commented 5 years ago

Reading the archiving overview (https://github.com/tango-controls/tango-doc/blob/master/source/tools-and-extensions/archiving/overview.rst), I wonder if it will be convenient to precise that HDB++ do not integrate a snapshot functionnality.

Also, some tips about the needed space and ressources depending on the number and frequency of archiving could be useful to understand the differences and needs.

Do you think it is something to integrate?

bourtemb commented 5 years ago

Reading the archiving overview (https://github.com/tango-controls/tango-doc/blob/master/source/tools-and-extensions/archiving/overview.rst), I wonder if it will be convenient to precise that HDB++ do not integrate a snapshot functionnality.

I think the Time Machine developed by Elettra provides a feature quite similar as the snapshot feature.

Also, some tips about the needed space and ressources depending on the number and frequency of archigin could be useful to understand the differences and needs.

Do you think it is something to integrate?

We don't have these numbers yet. It is foreseen to do some benchmark tests to get such numbers.

PhilLAL commented 5 years ago

Reading the archiving overview (https://github.com/tango-controls/tango-doc/blob/master/source/tools-and-extensions/archiving/overview.rst), I wonder if it will be convenient to precise that HDB++ do not integrate a snapshot functionnality.

I think the Time Machine developed by Elettra provides a feature quite similar as the snapshot feature.

Oh, intersting! Do you have a public link? :)

Also, some tips about the needed space and ressources depending on the number and frequency of archiving could be useful to understand the differences and needs. Do you think it is something to integrate?

We don't have these numbers yet. It is foreseen to do some benchmark tests to get such numbers.

From my point, I would say number of attributes, frequency, number of device servers will be the decisive characteristics which will constraint the needed ressources, Am I right? Do you see other characteristics?

bourtemb commented 5 years ago

From my point, I would say number of attributes, frequency, number of device servers will be the decisive characteristics which will constraint the needed ressources, Am I right? Do you see other characteristics?

The type of attributes to be stored is important too because if you are trying to store big spectrum attributes at high frequency, the DB will grow fast.