This document summarizes a presentation about caching at different levels of a web application stack. It discusses caching client-side assets with HTTP headers, full-page caching with tools like Varnish, fragment caching of dynamic content, object caching with memcached, and database query caching. It also covers challenges like cache invalidation, fragmentation, and complexity that caching can introduce. The presentation emphasizes measuring performance to determine if caching is worthwhile.

1. Caching Up and Down
the Stack
Long Island/Queens Django Meetup 5/20/14

2. Hi, I’m Dan Kuebrich
● Software engineer, python fan
● Web performance geek
● Founder of Tracelytics, now part of AppNeta
● Once (and future?) Queens resident

4. DJANGO

5. What is “caching”?
● Caching is avoiding doing expensive work
o by doing cheaper work
● Common examples?
o On repeat visits, your browser doesn’t download
images that haven’t changed
o Your CPU caches instructions, data so it doesn’t
have to go to RAM… or to disk!

6. What is “caching”?
Uncached
Client
Data Source

7. What is “caching”?
Client
Data Source
Uncached Cached
Cache Intermediary
Client
Data Source

8. What is “caching”?
Client
Data Source
Uncached Cached
Cache Intermediary
Client
Data Source
Fast!
Slow...

9. “Latency Numbers Every Programmer Should Know”
Systems Performance: Enterprise and the Cloud by Brendan Gregg
http://books.google.com/books?id=xQdvAQAAQBAJ&pg=PA20&lpg=PA20&source=bl&ots=hlTgyxdrnR&sig=CCjddHrY1H6muMVW9BFcbdO7DDo&hl=en&sa=X&ei=dS7oUquhOYr9oAT9oYGoDw&ved=0CCkQ6AEwAA#v=onepage
&q&f=false

10. A whole mess of caching:
● Browser cache
● CDN
● Proxy / optimizer
● Application-based
o Full-page
o Fragment
o Object cache
● Database
o Query cache
o Denormalization
Closer to the user
Closer to the data

11. Caching in Django apps: Frontend
● Client-side assets
● Full pages

12. Client-side assets

13. Client-side assets

14. Client-side assets
● Use HTTP caches!
o Browser
o CDN
o Intermediate proxies
● Set policy with cache headers
o Cache-Control / Expires
o ETag / Last-Modified

15. HTTP Cache-Control and Expires
● Stop the browser from even asking for it
● Expires
o Pick a date in the future, good til then
● Cache-control
o More flexible
o Introduced in HTTP 1.1
o Use this one

16. HTTP Cache-Control and Expires
dan@JLTM21:~$ curl -I https://login.tv.appneta.com/cache/tl-layouts_base_unauth-
compiled-162c2ceecd9a7ff1e65ab460c2b99852a49f5a43.css
HTTP/1.1 200 OK
Accept-Ranges: bytes
Cache-Control: max-age=315360000
Content-length: 5955
Content-Type: text/css
Date: Tue, 20 May 2014 23:12:16 GMT
Expires: Thu, 31 Dec 2037 23:55:55 GMT
Last-Modified: Fri, 16 May 2014 20:51:19 GMT
Server: nginx
Connection: keep-alive

17. HTTP Cache Control in Django
https://docs.djangoproject.com/en/dev/topics/cache/

18. ETag + Last-Modified

19. ETag + Last-Modified
dan@JLTM21:~$ curl -I www.appneta.com/stylesheets/styles.css
HTTP/1.1 200 OK
Last-Modified: Tue, 20 May 2014 05:52:50 GMT
ETag: "30854c-1c3d3-4f9ce7d715080"
Vary: Accept-Encoding
Content-Type: text/css
...

20. ETag + Last-Modified
dan@JLTM21:~$ curl -I www.appneta.com/stylesheets/styles.css --header 'If-None-
Match: "30854c-1c3d3-4f9ce7d715080"'
HTTP/1.1 304 Not Modified
Last-Modified: Tue, 20 May 2014 05:52:50 GMT
ETag: "30854c-1c3d3-4f9ce7d715080"
Vary: Accept-Encoding
Content-Type: text/css
Date: Tue, 20 May 2014 23:21:12 GMT
...

21. ETag vs Last-Modified
● Last-Modified is date-based
● ETag is content-based
● Most webservers generate both
● Some webservers (Apache) generate etags
that depend on local state
o If you have a load-balanced pool of servers working
here, they might not be using the same etags!

22. A whole mess of caching:
● Browser cache
● CDN
● Proxy / optimizer
● Application-based
o Full-page
o Fragment
o Object cache
● Database
o Query cache
o Denormalization

23. CDNs
● Put content closer to your end-users
o and offload HTTP requests from
your servers
● Best for static assets
● Same cache control policies apply

24. Full-page caching
Client
Data Source
Varnish
No internet
standards
necessary!

25. Full-page caching: mod_pagespeed
Client
Data Source
mod_pagespeed
● Dynamically rewrites
pages with frontend
optimizations
● Caches rewritten pages

26. A whole mess of caching:
● Browser cache
● CDN
● Proxy / optimizer
● Application-based
o Full-page
o Fragment
o Object cache
● Database
o Query cache
o Denormalization

27. Full-page caching in Django

28. Wait, where is this getting cached?
● Django makes it easy to configure
o In-memory
o File-based
o Memcached
o etc.

29. Full-page caching: dynamic pages?

30. Full-page caching: dynamic pages?

31. Fragment caching

32. Full-page caching: dynamic pages?

33. Full-page caching: the ajax solution

34. Object caching
def get_item_by_id(key):
# Look up the item in our database
return session.query(User)
.filter_by(id=key)
.first()

35. Object caching
def get_item_by_id(key):
# Check in cache
val = mc.get(key)
# If exists, return it
if val:
return val
# If not, get the val, store it in the cache
val = return session.query(User)
.filter_by(id=key)
.first()
mc.set(key, val)
return val

36. Object caching
@decorator
def cache(expensive_func, key):
# Check in cache
val = mc.get(key)
# If exists, return it
if val:
return val
# If not, get the val, store it in the cache
val = expensive_func(key)
mc.set(key, val)
return val

37. Object caching
@cache
def get_item_by_id(key):
# Look up the item in our database
return session.query(User)
.filter_by(id=key)
.first()

38. Object caching in Django

39. A whole mess of caching:
● Browser cache
● CDN
● Proxy / optimizer
● Application-based
o Full-page
o Fragment
o Object cache
● Database
o Query cache
o Denormalization

40. Query caching
Client
Actual tables
Database
Query
Cache
Cached?

41. Query caching
mysql> select SQL_CACHE count(*) from traces;
+----------+
| count(*) |
+----------+
| 3135623 |
+----------+
1 row in set (0.56 sec)
mysql> select SQL_CACHE count(*) from traces;
+----------+
| count(*) |
+----------+
| 3135623 |
+----------+
1 row in set (0.00 sec)

42. Query caching

43. Query caching
Uncached
Cached

44. Denormalization
mysql> select table1.x, table2.y from table1 join table2 on table1.z = table2.q
where table1.z > 100;
mysql> select table1.x, table1.y from table1 where table1.z > 100;

45. A whole mess of caching:
● Browser cache
● CDN
● Proxy / optimizer
● Application-based
o Full-page
o Fragment
o Object cache
● Database
o Query cache
o Denormalization

46. Caching: what can go wrong?
● Invalidation
● Fragmentation
● Stampedes
● Complexity

47. Invalidation
Client
Data Source
Cache Intermediary
Update!
Write
Invalidate

48. Invalidation on page-scale
● Browser cache
● CDN
● Proxy / optimizer
● Application-based
o Full-page
o Fragment
o Object cache
● Database
o Query cache
o Denormalization
More savings,
generally more invalidation...
Smaller savings,
generally less invalidation

49. Fragmentation
● What if I have a lot of different things to
cache?
o More misses
o Potential cache eviction

50. Fragmentation
Your pages / objects
FrequencyofAccess

51. Fragmentation
Your pages / objects
FrequencyofAccess

52. Stampedes
● On a cache miss extra work is done
● The result is stored in the cache
● What if multiple simultaneous misses?

53. Stampedes
http://allthingsd.com/20080521/stampede-facebook-opens-its-profile-doors/

54. Complexity
● How much caching do I need, and where?
● What is the invalidation process
o on data update? on release?
● What happens if the caches fall over?
● How do I debug it?

55. Takeaways
● The ‘how’ of caching:
o What are you caching?
o Where are you caching it?
o How bad is a cache miss?
o How and when are you invalidating?

56. Takeaways
● The ‘why’ of caching:
o Did it actually get faster?
o Is speed worth extra complexity?
o Don’t guess – measure!
o Always use real-world conditions.

57. Questions?
?

58. Thanks!
● Interested in measuring your Django app’s
performance?
o Free trial of TraceView:
www.appneta.com/products/traceview
● See you at Velocity NYC this fall?
● Twitter: @appneta / @dankosaur

59. Resources
● Django documentation on caching: https://docs.djangoproject.com/en/dev/topics/cache/
● Varnish caching, via Disqus: http://blog.disqus.com/post/62187806135/scaling-django-to-8-
billion-page-views
● Django cache option comparisons: http://codysoyland.com/2010/jan/17/evaluating-django-
caching-options/
● More Django-specific tips: http://www.slideshare.net/csky/where-django-caching-bust-at-the-
seams
● Guide to cache-related HTTP headers: http://www.mobify.com/blog/beginners-guide-to-http-
cache-headers/
● Google PageSpeed: https://developers.google.com/speed/pagespeed/module