1. Cachingon the web
@jcemer

2. jcemer.com
twitter.com/jcemer

4. Cache is a hardware or
software component that
stores data so future
requests for that data can
be served faster - Wikipedia

5. Caching and RAM
is the answer to
everything
- about Flickr Architecture

6. Caching is one of
the secrets of high
scalability and
performance

7. Where could cache be
present on the web?
• Browser
• Network
• Server
• Application

8. Browsercaching

9. Caching would be
useless if it did not
significantly improve
performance
- HTTP/1.1 specification

10. The user might
reuse assets during
navigation

11. • Documents
• Images
• Scripts and CSS
• Asynchronous Requests
What could be cached?

12. HTTP headers have the
responsibility to define
if a response could be
cached and for how long

13. GET /main.css HTTP/1.1
Host: jcemer.com

14. GET /main.css HTTP/1.1
Host: jcemer.com
HTTP/1.1 200
Date: Tue, 13 Sep 2016 13:32:50 GMT
Cache-Control: max-age=604800
<Response Data>

15. allows the response to
be stored in cache
Cache-Control: max-age=604800
Response header

16. CSS is requested once
during navigation
😄 😄

17. The browser requests it
again only if the cache
expires or if the user
force refreshes the page

18. allow to add more info
about the resource
Last-Modified: Mon, 12 Sep 2016 22:06:39 GMT
Etag: W/"337e7-8HrLmYe6UGIUDolQeGLoyw"
Response headers

19. GET /main.css HTTP/1.1
Host: jcemer.com
HTTP/1.1 200 OK
Date: Tue, 13 Sep 2016 13:32:50 GMT
Last-Modified: Mon, 12 Sep 2016 15:23:17 GMT
Cache-Control: max-age=604800
new

20. allow reuse the cached
resource if it didn’t change
If-Modified-Since: Mon, 12 Sep 2016 15:23:17 GMT
If-Match: W/"337e7-8HrLmYe6UGIUDolQeGLoyw"
Request headers

21. GET /main.css HTTP/1.1
Host: jcemer.com
If-Modified-Since: Mon, 12 Sep 2016 15:23:17 GMT

22. GET /main.css HTTP/1.1
Host: jcemer.com
If-Modified-Since: Mon, 12 Sep 2016 15:23:17 GMT
HTTP/1.1 304 Not Modified
Date: Tue, 13 Sep 2016 13:32:50 GMT
Cache-Control: max-age=604800
HTTP/1.1 200 OK
Date: Tue, 13 Sep 2016 13:32:50 GMT
Last-Modified: Mon, 12 Sep 2016 15:23:17 GMT
<Response Data>
<No Response Data>

23. A Website with the
HTTP headers wisely
defined will provide a
better experience for
the users

24. Networkcaching

25. Content Delivery Network
(CDN) is a globally
distributed network of
proxy servers
- Wikipedia

26. allows proxy servers
to cache the content
Cache-control: public

27. Servercaching

28. Finally, the control
is all on your hand,
developer!

29. Cache server

30. The cache server stay
between the user and
the application or
other servers

31. • Shared documents
• Images
• Scripts and CSS
• Asynchronous Requests
What could be cached?

32. • Varnish
• Squid
• nginx
Tools
https://varnish-cache.org
http://www.squid-cache.org
https://www.nginx.com

33. location / {
proxy_pass http://otherserver;
}
the nginx intermediates
the client request’s

34. nginx
cache server

35. proxy_cache_path /path/to/cache;
location / {
proxy_pass http://otherserver;
proxy_cache cache;
}
caching!

36. 😄
nginx
cache server

37. The proxy caches
content relying only
on the application
HTTP headers

38. t1
t2
t1
request at a
different time

39. 😓
…
t1
t1
t1
x3

40. proxy_cache_lock on;
proxy_cache_lock_timeout 180;
allow the proxy to delegate
only the first of similar
requests at a time

41. t1
t1
😄x1

42. All clients are waiting
until receive the
response when the
first request returns

43. What happen in
case of failures?
🤒

44. allows to delivery expired
content in case of failure
proxy_cache_use_stale timeout error http_500;

45. 😄 🤒

46. The proxy could
improve the fault
tolerance of the
application

47. proxy_cache_use_stale updating;
delivers expired content
for the subsequent similar
requests

48. Appcaching

49. Caching on the app
reduces the time of
specific operations

50. • Complex computations
• Data shared across
requests
What could be cached?

51. def price
@price ||= Price.new(unit_price, category)
end

52. Memoization stores
the results to avoid
future calculations

53. A global code
memoization is going to
last in-memory during
all the application
execution cycle

54. $price = Price.new(unit_price, category)
😓

55. cache.fetch("cat#{cat_id}", expires_in: 1.minute) do
CatagoryTax.new(cat_id)
end
http://api.rubyonrails.org/classes/ActiveSupport/Cache/Store.html
ActiveSupport::Cache::Store

56. Rails Caching API
allows store and reuse
data during an amount
of time across requests

57. https://github.com/ptarjan/node-cache/issues/77
😓

58. 🔥

59. Rails Memory Caching
wisely prunes the cached
data when it exceeds the
allotted memory size

60. One of the ways to
scale an application is
through adding more
application instances
(scale horizontally)
😓

62. Load balancer
App
instances

63. Never assumes that
anything cached in memory
or on disk will be available
on a future request
https://12factor.net/processes

64. Any storage could be
used to share cache data
between instances but the
Key-value Storages are
the most common

65. • Redis
• Memcached
Key-value Storages
http://redis.io
http://memcached.org

67. These tools have
different policies to
prune the amount of
cached data

68. Cache Storages might also
be fault tolerable with
replication and
persistence
http://redis.io/topics/persistence
http://redis.io/topics/replication
http://redis.io/topics/sentinel

70. Race condition happens
when different
application instances
fetch for a not cached
data at the same time
😓

71. expired data!
😁

72. cache.fetch(key, race_condition_ttl: 10.seconds) do
heavy_db_computation
end
http://api.rubyonrails.org/classes/ActiveSupport/Cache/Store.html
ActiveSupport::Cache::Store

73. 😁
😳

74. It is difficult to completely
eliminate cache updating
race conditions issues with
multiple application
instances

75. A solution for that is to
update the cache data
outside the application flow
and just consume cached
data on the application

76. 1. Obey the
HTTP headers

77. 2. Caching is
important

78. 3. Measure the
miss/hits of the
caching strategy

79. 4. Evaluate carefully
the caching strategy
options

80. Thanks!@jcemer