Scaling Django

Scaling Django Web AppsMike Malone

euro con 2009Tuesday, May 5, 2009

Hi, I’m Mike.

Tuesday, May 5, 2009



http://www.flickr.com/photos/kveton/2910536252/Tuesday, May 5, 2009

http://www.flickr.com/photos/kveton/2910536252/

http://www.flickr.com/photos/kveton/2910536252/


euro con 2009

Pownce

• Large scale

• Hundreds of requests/sec

• Thousands of DB operations/sec

• Millions of user relationships

• Millions of notes

• Terabytes of static data

7


euro con 2009

Pownce

• Encountered and eliminated many common scaling bottlenecks

• Real world example of scaling a Django app

• Django provides a lot for free

• I’ll be focusing on what you have to build yourself, and the rare places where Django got in the way

8


Scalability


euro con 2009

Scalability

10

• Speed / Performance

• Generally affected by language choice

• Achieved by adopting a particular technology

Scalability is NOT:


euro con 2009

import time

def application(environ, start_response): time.sleep(10) start_response('200 OK', [('content-type', 'text/plain')]) return ('Hello, world!',)

A Scalable Application

11


euro con 2009

def application(environ, start_response): remote_addr = environ['REMOTE_ADDR'] f = open('access-log', 'a+') f.write(remote_addr + "\n") f.flush() f.seek(0) hits = sum(1 for l in f.xreadlines()

if l.strip() == remote_addr) f.close() start_response('200 OK', [('content-type', 'text/plain')]) return (str(hits),)

A High Performance Application

12


euro con 2009

Scalability

13

A scalable system doesn’t need to change when the size of the problem changes.


euro con 2009

Scalability

• Accommodate increased usage

• Accommodate increased data

• Maintainable

14


euro con 2009

Scalability

• Two kinds of scalability

• Vertical scalability: buying more powerful hardware, replacing what you already own

• Horizontal scalability: buying additional hardware, supplementing what you already own

15


euro con 2009

Vertical Scalability

• Costs don’t scale linearly (server that’s twice is fast is more than twice as much)

• Inherently limited by current technology

• But it’s easy! If you can get away with it, good for you.

16


euro con 2009

Vertical Scalability

17

Sky scrapers are special. Normal buildings don’t need 10 floor foundations. Just build!

- Cal Henderson

“


euro con 2009

Horizontal Scalability

18

The ability to increase a system’s capacity by adding more processing units (servers)


euro con 2009


19

It’s how large apps are scaled.


euro con 2009


• A lot more work to design, build, and maintain

• Requires some planning, but you don’t have to do all the work up front

• You can scale progressively...

• Rest of the presentation is roughly in order

20


Caching


euro con 2009

Caching

• Several levels of caching available in Django

• Per-site cache: caches every page that doesn’t have GET or POST parameters

• Per-view cache: caches output of an individual view

• Template fragment cache: caches fragments of a template

• None of these are that useful if pages are heavily personalized

22


euro con 2009

Caching

• Low-level Cache API

• Much more flexible, allows you to cache at any granularity

• At Pownce we typically cached

• Individual objects

• Lists of object IDs

• Hard part is invalidation

23


euro con 2009

Caching

• Cache backends:

• Memcached

• Database caching

• Filesystem caching

24


euro con 2009

Caching

25

Use Memcache.


euro con 2009

Sessions

26

Use Memcache.


euro con 2009

Sessions

27

Or Tokyo Cabinethttp://github.com/ericflo/django-tokyo-sessions/

Thanks @ericflo


http://github.com/ericflo/django-tokyo-sessions/




euro con 2009

from django.core.cache import cache

class UserProfile(models.Model): ... def get_social_network_profiles(self): cache_key = ‘networks_for_%s’ % self.user.id profiles = cache.get(cache_key) if profiles is None: profiles = self.user.social_network_profiles.all() cache.set(cache_key, profiles) return profiles

Caching

28

Basic caching comes free with Django:


euro con 2009

from django.core.cache import cachefrom django.db.models import signals

def nuke_social_network_cache(self, instance, **kwargs): cache_key = ‘networks_for_%s’ % self.instance.user_id cache.delete(cache_key)

signals.post_save.connect(nuke_social_network_cache, sender=SocialNetworkProfile)signals.post_delete.connect(nuke_social_network_cache, sender=SocialNetworkProfile)

Caching

29

Invalidate when a model is saved or deleted:


euro con 2009

Caching

30

• Invalidate post_save, not pre_save

• Still a small race condition

• Simple solution, worked for Pownce:

• Instead of deleting, set the cache key to None for a short period of time

• Instead of using set to cache objects, use add, which fails if there’s already something stored for the key


euro con 2009

Advanced Caching

31

• Memcached’s atomic increment and decrement operations are useful for maintaining counts

• But they’re not available in Django 1.0

• Added in 1.1 by ticket #6464


euro con 2009

Advanced Caching

32

• You can still use them if you poke at the internals of the cache object a bit

• cache._cache is the underlying cache object

try: result = cache._cache.incr(cache_key, delta)except ValueError: # nonexistent key raises ValueError # Do it the hard way, store the result.return result


euro con 2009

Advanced Caching

33

• Other missing cache API

• delete_multi & set_multi

• append: add data to existing key after existing data

• prepend: add data to existing key before existing data

• cas: store this data, but only if no one has edited it since I fetched it


euro con 2009

Advanced Caching

34

• It’s often useful to cache objects ‘forever’ (i.e., until you explicitly invalidate them)

• User and UserProfile

• fetched almost every request

• rarely change

• But Django won’t let you

• IMO, this is a bug :(


euro con 2009

class CacheClass(BaseCache): def __init__(self, server, params): BaseCache.__init__(self, params) self._cache = memcache.Client(server.split(';'))

def add(self, key, value, timeout=0): if isinstance(value, unicode): value = value.encode('utf-8') return self._cache.add(smart_str(key), value, timeout or self.default_timeout)

The Memcache Backend

35


euro con 2009

class CacheClass(BaseCache): def __init__(self, server, params): BaseCache.__init__(self, params) self._cache = memcache.Client(server.split(';'))

def add(self, key, value, timeout=None): if isinstance(value, unicode): value = value.encode('utf-8') if timeout is None: timeout = self.default_timeout return self._cache.add(smart_str(key), value, timeout)

The Memcache Backend

36


euro con 2009

Advanced Caching

37

• Typical setup has memcached running on web servers

• Pownce web servers were I/O and memory bound, not CPU bound

• Since we had some spare CPU cycles, we compressed large objects before caching them

• The Python memcache library can do this automatically, but the API is not exposed


euro con 2009

from django.core.cache import cachefrom django.utils.encoding import smart_strimport inspect as i

if 'min_compress_len' in i.getargspec(cache._cache.set)[0]: class CacheClass(cache.__class__): def set(self, key, value, timeout=None, min_compress_len=150000): if isinstance(value, unicode): value = value.encode('utf-8') if timeout is None: timeout = self.default_timeout return self._cache.set(smart_str(key), value, timeout, min_compress_len) cache.__class__ = CacheClass

Monkey Patching core.cache

38


euro con 2009

Advanced Caching

39

• Useful tool: automagic single object cache

• Use a manager to check the cache prior to any single object get by pk

• Invalidate assets on save and delete

• Eliminated several hundred QPS at Pownce


euro con 2009

Advanced Caching

40

All this and more at:

http://github.com/mmalone/django-caching/




euro con 2009

Advanced Caching

• Consistent hashing: hashes cached objects in such a way that most objects map to the same node after a node is added or removed.

41

http://www.flickr.com/photos/deepfrozen/2191036528/




euro con 2009

Caching

42

Now you’ve made life easier for your DB server,next thing to fall over: your app server.


Load Balancing


euro con 2009

Load Balancing

• Out of the box, Django uses a shared nothing architecture

• App servers have no single point of contention

• Responsibility pushed down the stack (to DB)

• This makes scaling the app layer trivial: just add another server

44


euro con 2009

Load Balancing

45

App Servers

Database

Load Balancer

Spread work between multiple nodes in a cluster using a load balancer.

• Hardware or software• Layer 7 or Layer 4


euro con 2009

Load Balancing

46

• Hardware load balancers

• Expensive, like $35,000 each, plus maintenance contracts

• Need two for failover / high availability

• Software load balancers

• Cheap and easy, but more difficult to eliminate as a single point of failure

• Lots of options: Perlbal, Pound, HAProxy, Varnish, Nginx


euro con 2009

Load Balancing

47

• Most of these are layer 7 proxies, and some software balancers do cool things

• Caching

• Re-proxying

• Authentication

• URL rewriting


euro con 2009

Load Balancing

48

A common setup for large operations is to use redundant layer 4 hardware balancers in front of a pool of layer 7 software balancers.

Hardware Balancers

Software Balancers

App Servers


euro con 2009

Load Balancing

49

• At Pownce, we used a single Perlbal balancer

• Easily handled all of our traffic (hundreds of simultaneous connections)

• A SPOF, but we didn’t have $100,000 for black box solutions, and weren’t worried about service guarantees beyond three or four nines

• Plus there were some neat features that we took advantage of


euro con 2009

Perlbal Reproxying

50

Perlbal reproxying is a really cool, and really poorlydocumented feature.


euro con 2009

Perlbal Reproxying

51

1. Perlbal receives request

2. Redirects to App Server

1. App server checks auth (etc.)

2. Returns HTTP 200 with X-Reproxy-URL header set to internal file server URL

3. File served from file server via Perlbal


euro con 2009

Perlbal Reproxying

• Completely transparent to end user

• Doesn’t keep large app server instance around to serve file

• Users can’t access files directly (like they could with a 302)

52


euro con 2009

def download(request, filename): # Check auth, do your thing response = HttpResponse() response[‘X-REPROXY-URL’] = ‘%s/%s’ % (FILE_SERVER, filename) return response

Perlbal Reproxying

53

Plus, it’s really easy:


euro con 2009

Load Balancing

54

Best way to reduce load on your app servers: don’t use them to do hard stuff.


Queuing


euro con 2009

Queuing

• A queue is simply a bucket that holds messages until they are removed for processing by clients

• Many expensive operations can be queued and performed asynchronously

• User experience doesn’t have to suffer

• Tell the user that you’re running the job in the background (e.g., transcoding)

• Make it look like the job was done real-time (e.g., note distribution)

56


euro con 2009

Queuing

• Lots of open source options for queuing

• Ghetto Queue (MySQL + Cron)

• this is the official name.

• Gearman

• TheSchwartz

• RabbitMQ

• Apache ActiveMQ

• ZeroMQ

57


euro con 2009

Queuing

• Lots of fancy features: brokers, exchanges, routing keys, bindings...

• Don’t let that crap get you down, this is really simple stuff

• Biggest decision: persistence

• Does your queue need to be durable and persistent, able to survive a crash?

• This requires logging to disk which slows things down, so don’t do it unless you have to

58


euro con 2009

Queuing

• Pownce used a simple ghetto queue built on MySQL / cron

• Problematic if you have multiple consumers pulling jobs from the queue

• No point in reinventing the wheel, there are dozens of battle-tested open source queues to choose from

59


euro con 2009

from django.core.management import setup_environfrom mysite import settings

setup_environ(settings)

Django Standalone Scripts

60

Consumers need to setup the Django environment


THE DATABASE!


euro con 2009

The Database

• Til now we’ve been talking about

• Shared nothing

• Pushing problems down the stack

• But we have to store a persistent and consistent view of our application’s state somewhere

• Enter, the database...

62


euro con 2009

CAP Theorem

• Three properties of a shared-data system

• Consistency: all clients see the same data

• Availability: all clients can see some version of the data

• Partition Tolerance: system properties hold even when the system is partitioned & messages are lost

• But you can only have two

63


euro con 2009

CAP Theorem

• Big long proof... here’s my version.

• Empirically, seems to make sense.

• Eric Brewer

• Professor at University of California, Berkeley

• Co-founder and Chief Scientist of Inktomi

• Probably smarter than me

64


euro con 2009

CAP Theorem

• The relational database systems we all use were built with consistency as their primary goal

• But at scale our system needs to have high availability and must be partitionable

• The RDBMS’s consistency requirements get in our way

• Most sharding / federation schemes are kludges that trade consistency for availability & partition tolerance

65


euro con 2009

The Database

• There are lots of non-relational databases coming onto the scene

• CouchDB

• Cassandra

• Tokyo Cabinet

• But they’re not that mature, and they aren’t easy to use with Django

66


euro con 2009

The Database

• Django has no support for

• Non-relational databases like CouchDB

• Multiple databases (coming soon?)

• If you’re looking for a project, plz fix this.

• Only advice: don’t get too caught up in trying to duplicate the existing ORM API

67


euro con 2009

I Want a Pony

• Save always saves every field of a model

• Causes unnecessary contention and more data transfer

• A better way:

• Use descriptors to determine what’s dirty

• Only update dirty fields when an object is saved

68


Denormalization


euro con 2009

Denormalization

• Django encourages normalized data, which is usually good

• But at scale you need to denormalize

• Corollary: joins are evil

• Django makes it really easy to do joins using the ORM, so pay attention

70


euro con 2009

Denormalization

• Start with a normalized database

• Selectively denormalize things as they become bottlenecks

• Denormalized counts, copied fields, etc. can be updated in signal handlers

71


Replication


euro con 2009

Replication

• Typical web app is 80 to 90% reads

• Adding read capacity will get you a long way

• MySQL Master-Slave replication

73

Read & Write

Read only


euro con 2009

Replication

• Django doesn’t make it easy to use multiple database connections, but it is possible

• Some caveats

• Slave lag interacts with caching in weird ways

• You can only save to your primary DB (the one you configure in settings.py)

• Unless you get really clever...

74


euro con 2009

class SlaveDatabaseWrapper(DatabaseWrapper): def _cursor(self, settings): if not self._valid_connection(): kwargs = { 'conv': django_conversions, 'charset': 'utf8', 'use_unicode': True, } kwargs = pick_random_slave(settings.SLAVE_DATABASES) self.connection = Database.connect(**kwargs) ... cursor = CursorWrapper(self.connection.cursor()) return cursor

Replication

75

1. Create a custom database wrapper by subclassing DatabaseWrapper


euro con 2009

class MultiDBQuerySet(QuerySet): ... def update(self, **kwargs): slave_conn = self.query.connection self.query.connection = default_connection super(MultiDBQuerySet, self).update(**kwargs) self.query.connection = slave_conn

Replication

76

2. Custom QuerySet that uses primary DB for writes


euro con 2009

class SlaveDatabaseManager(db.models.Manager): def get_query_set(self): return MultiDBQuerySet(self.model, query=self.create_query())

def create_query(self): return db.models.sql.Query(self.model, connection)

Replication

77

3. Custom Manager that uses your custom QuerySet


euro con 2009

Replication

78

http://github.com/mmalone/django-multidb/

Example on github:




euro con 2009

Replication

• Goal:

• Read-what-you-write consistency for writer

• Eventual consistency for everyone else

• Slave lag screws things up

79


euro con 2009

Replication

80

What happens when you become write saturated?


Federation


euro con 2009

Federation

82

• Start with Vertical Partitioning: split tables that aren’t joined across database servers

• Actually pretty easy

• Except not with Django


euro con 2009

Federation

83

django.db.models.base

FAIL!


euro con 2009

Federation

84

If the Django pony gets kicked every time someonuses {% endifnotequal %} I don’t want to know what

happens every time django.db.connection is imported.

http://www.flickr.com/photos/captainmidnight/811458621/




euro con 2009

Federation

• At some point you’ll need to split a single table across databases (e.g., user table)

• Now auto-increment won’t work

• But Django uses auto-increment for PKs

• ugh

• Pluggable UUID backend?

85


Profiling, Monitoring & Measuring


euro con 2009

>>> Article.objects.filter(pk=3).query.as_sql()('SELECT "app_article"."id", "app_article"."name", "app_article"."author_id" FROM "app_article" WHERE "app_article"."id" = %s ', (3,))

Know your SQL

87


euro con 2009

>>> import sqlparse>>> def pp_query(qs):... t = qs.query.as_sql()... sql = t[0] % t[1]... print sqlparse.format(sql, reindent=True, keyword_case='upper')... >>> pp_query(Article.objects.filter(pk=3))SELECT "app_article"."id", "app_article"."name", "app_article"."author_id"FROM "app_article"WHERE "app_article"."id" = 3

Know your SQL

88


euro con 2009

>>> from django.db import connection>>> connection.queries[{'time': '0.001', 'sql': u'SELECT "app_article"."id", "app_article"."name", "app_article"."author_id" FROM "app_article"'}]

Know your SQL

89


euro con 2009

Know your SQL

• It’d be nice if a lightweight stacktrace could be done in QuerySet.__init__

• Stick the result in connection.queries

• Now we know where the query originated

90


euro con 2009

Measuring

91

Django Debug Toolbar

http://github.com/robhudson/django-debug-toolbar/




euro con 2009

Monitoring

• Ganglia

• Munin

92

You can’t improve what you don’t measure.


euro con 2009

Measuring & Monitoring

• Measure

• Server load, CPU usage, I/O

• Database QPS

• Memcache QPS, hit rate, evictions

• Queue lengths

• Anything else interesting

93


All done... Questions?


euro con 2009

Contact Me

95

Mike [email protected]

twitter.com/mjmalone


mailto:[email protected]

mailto:[email protected]

Scaling Django

Technology