Use Memcache to store prediction values
hunterowens opened this issue · comments
The current method, of loading a pickle file, is not best practices. We should use memcache
@hunterowens, is this still a problem or was it resolved?
Still best practices to do this, but downgraded from urgent to normal.
On Mon, Sep 2, 2013 at 2:19 PM, Juan-Pablo Velez
notifications@github.comwrote:
@hunterowens https://github.com/hunterowens, is this still a problem or
was it resolved?—
Reply to this email directly or view it on GitHubhttps://github.com//issues/64#issuecomment-23675203
.
I would say this is somewhat low priority.
I can make a change where the prediction values are always stored/used in memory, but are also saved to pickle files when there are updates.
Of course this means that if Flask was re-started, in-memory prediction values would be gone; so it would be necessary to reload the pickle files into memory.
The logic for this seems pretty straightforward. I've done something similar for another personal project I've worked on in the past.