#28977: Change Local Memory Cache to Use LRU
-------------------------------------+-------------------------------------
Reporter: Grant Jenks | Owner: Grant
| Jenks
Type: New feature | Status: assigned
Component: Core (Cache system) | Version: master
Severity: Normal | Resolution:
Keywords: | Triage Stage: Accepted
Has patch: 1 | Needs documentation: 0
Needs tests: 0 | Patch needs improvement: 0
Easy pickings: 0 | UI/UX: 0
-------------------------------------+-------------------------------------
Changes (by Grant Jenks):
* has_patch: 0 => 1
* stage: Someday/Maybe => Accepted
Comment:
Pull request created at https://github.com/django/django/pull/9555
Django developers mailing list thread at
https://groups.google.com/forum/#!topic/django-developers/Gz2XqtoYmNk
Summarizing the two responses: Josh Smeaton (author of django-lrucache-
backend, project cited in the initial post) was in favor of providing a
better default option. Adam Johnson was also +1. Josh and Adam were also
interested in providing a way to disable cache key validation. I'm +0 on
that change so long as the default is to validate the key. I'd rather not
add those changes myself.
I did some very simple benchmarking locally:
Here's the performance of cache.get for the current implementation:
{{{
$ python manage.py shell
Python 3.6.3 (default, Oct 5 2017, 22:47:21)
Type 'copyright', 'credits' or 'license' for more information
IPython 6.2.1 -- An enhanced Interactive Python. Type '?' for help.
In [1]: from django.core.cache import cache
In [2]: cache.set(b'foo', b'bar')
In [3]: %timeit cache.get(b'foo')
14.2 µs ± 109 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
}}}
And here's the performance of cache.get for the new implementation:
{{{
$ python manage.py shell
Python 3.6.3 (default, Oct 5 2017, 22:47:21)
Type 'copyright', 'credits' or 'license' for more information
IPython 6.2.1 -- An enhanced Interactive Python. Type '?' for help.
In [1]: from django.core.cache import cache
In [2]: cache.set(b'foo', b'bar')
In [3]: %timeit cache.get(b'foo')
6.29 µs ± 140 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
}}}
I also tried Josh's django-lrucache-backend implementation:
{{{
$ python manage.py shell
Python 3.6.3 (default, Oct 5 2017, 22:47:21)
Type 'copyright', 'credits' or 'license' for more information
IPython 6.2.1 -- An enhanced Interactive Python. Type '?' for help.
In [3]: from django.core.cache import caches
In [4]: cache = caches['local']
In [5]: cache.set(b'foo', b'bar')
In [6]: %timeit cache.get(b'foo')
10.1 µs ± 135 ns per loop (mean ± std. dev. of 7 runs, 100000 loops each)
}}}
It's not a great benchmark but it's encouraging. The new implementation
appears to be faster than both the current implementation and the django-
lrucache-backend. I haven't profiled but I would guess
collections.OrderedDict is very fast, as is threading.RLock both of which
are introduced by these changes.
--
Ticket URL: <https://code.djangoproject.com/ticket/28977#comment:4>
Django <https://code.djangoproject.com/>
The Web framework for perfectionists with deadlines.
--
You received this message because you are subscribed to the Google Groups
"Django updates" group.
To unsubscribe from this group and stop receiving emails from it, send an email
to [email protected].
To post to this group, send email to [email protected].
To view this discussion on the web visit
https://groups.google.com/d/msgid/django-updates/068.62c3e55b1ce399566e18a23bced11359%40djangoproject.com.
For more options, visit https://groups.google.com/d/optout.
