Re: ZODB Time Series Data

On Thu, Apr 20, 2017 at 3:51 AM, Christopher Lozinski <[email protected]>
wrote:

> What do you  think of storing large amounts of time series data in the
> ZODB?
>

Depends on the application.  If we're talking about massive amounts of
data, there are specialized formats for that.  I don't see people using
ZODB for big data applications.


> All of these IoT applications collect large amounts of time-series data.
>

IMO, Non-transactional databases are better suited to collecting large
amounts of raw data.

If you use a B-tree, indexed by time, lots of rebalancing will happen as
> you add more  data.
>

Nope. BTrees don't rebalance but generally stay well  balanced through
their splitting approach.

Blob storage is good if you are storing one big blob in a transaction.
>

Blobs are best suited to storing files (e.g. media).

Although I could also see blobs being used to store common data formats.
For example, I have an idea for storing Pandas data frames in blobs using
the newish Feather format.  It could be interesting to use ZODB to share
Jupyter notebooks.  This is small data though.


> Is there some kind of storage for time series data?
>

I expect this to be highly application dependent.

Jim

-- 
Jim Fulton
http://jimfulton.info

-- 
You received this message because you are subscribed to the Google Groups "zodb" group.
To unsubscribe from this group and stop receiving emails from it, send an email to [email protected].
For more options, visit https://groups.google.com/d/optout.