Scaling Twitter

Rails Scales.
(but not out of the box)

First, Some Facts
• 600 requests per second. Growing fast.
• 180 Rails Instances (Mongrel). Growing fast.
• 1 Database Server (MySQL) + 1 Slave.
• 30-odd Processes for Misc. Jobs
• 8 Sun X4100s
• Many users, many updates.

Joy Pain

Oct Nov Dec Jan Feb March Apr

IM IN UR RAILZ

MAKIN EM GO FAST

It’s Easy, Really.
1. Realize Your Site is Slow
2. Optimize the Database
3. Cache the Hell out of Everything
4. Scale Messaging
5. Deal With Abuse

It’s Easy, Really.
1. Realize Your Site is Slow
2. Optimize the Database
3. Cache the Hell out of Everything
4. Scale Messaging
5. Deal With Abuse
6. Proﬁt

the
more
you
know

{ Part the First }

Don’t Be Like Us

• Munin
• Nagios
• AWStats & Google Analytics
• Exception Notiﬁer / Exception Logger
• Immediately add reporting to track problems.

Test Everything

• Start Before You Start

• No Need To Be Fancy

• Tests Will Save Your Life

• Agile Becomes
Important When Your
Site Is Down

Benchmarks?
let your users do it.

The Database
{ Part the Second }

“The Next Application I Build is Going
to Be Easily Partitionable” - S. Butterﬁeld

class AddIndex < ActiveRecord::Migration
def self.up
add_index :users, :email
end

def self.down
remove_index :users, :email
end
end

Repeat for any column that appears in a WHERE clause

Rails won’t do this for you.

class DenormalizeFriendsIds < ActiveRecord::Migration
def self.up
add_column "users", "friends_ids", :text
end

def self.down
remove_column "users", "friends_ids"
end
end

class Friendship < ActiveRecord::Base
belongs_to :user
belongs_to :friend

after_create :add_to_denormalized_friends
after_destroy :remove_from_denormalized_friends

def add_to_denormalized_friends
user.friends_ids << friend.id
user.friends_ids.uniq!
user.save_without_validation
end

def remove_from_denormalized_friends
user.friends_ids.delete(friend.id)
user.save_without_validation
end
end

bob.friends.map(&:email)
Status.count()
“email like ‘%#{search}%’”

That’s where we are.
Seriously.
If your Rails application is doing anything more
complex than that, you’re doing something wrong*.

* or you observed the First Rule of Butterﬁeld.

Partitioning Comes Later.
(we’ll let you know how it goes)

class Status < ActiveRecord::Base
class << self
def count_with_memcache(*args)
return count_without_memcache unless args.empty?
count = CACHE.get(“status_count”)
if count.nil?
count = count_without_memcache
CACHE.set(“status_count”, count)
end
count
end
alias_method_chain :count, :memcache
end
after_create :increment_memcache_count
after_destroy :decrement_memcache_count
...
end

class User < ActiveRecord::Base
def friends_statuses
ids = CACHE.get(“friends_statuses:#{id}”)
Status.find(:all, :conditions => [“id IN (?)”, ids])
end
end

class Status < ActiveRecord::Base
after_create :update_caches
def update_caches
user.friends_ids.each do |friend_id|
ids = CACHE.get(“friends_statuses:#{friend_id}”)
ids.pop
ids.unshift(id)
CACHE.set(“friends_statuses:#{friend_id}”, ids)
end
end
end

The Future

ve d
ti r
co
Ac
e
R

90% API Requests
Cache Them!

“There are only two hard things in CS:
cache invalidation and naming things.”

– Phil Karlton, via Tim Bray

You Already Knew All
That Other Stuff, Right?

Producer Consumer
Message
Producer Consumer
Queue
Producer Consumer

DRb
• The Good:
• Stupid Easy
• Reasonably Fast
• The Bad:
• Kinda Flaky
• Zero Redundancy
• Tightly Coupled

ejabberd

Jabber Client
(drb)

Incoming Outgoing
Presence
Messages Messages

MySQL

Server
DRb.start_service ‘druby://localhost:10000’, myobject

Client
myobject = DRbObject.new_with_uri(‘druby://localhost:10000’)

Rinda

• Shared Queue (TupleSpace)
• Built with DRb
• RingyDingy makes it stupid easy
• See Eric Hodel’s documentation
• O(N) for take(). Sigh.

Timestamp: 12/22/06 01:53:14 (4 months ago)
Author: lattice
Message: Fugly. Seriously. Fugly.

SELECT * FROM messages WHERE
substring(truncate(id,0),-2,1) = #{@fugly_dist_idx}

It Scales.
(except it stopped on Tuesday)

Options

• ActiveMQ (Java)
• RabbitMQ (erlang)
• MySQL + Lightweight Locking
• Something Else?

erlang?

What are you doing?
Stabbing my eyes out with a fork.

Starling

• Ruby, will be ported to something faster
• 4000 transactional msgs/s
• First pass written in 4 hours
• Speaks MemCache (set, get)

Use Messages to
Invalidate Cache
(it’s really not that hard)

9000 friends in 24 hours
(doesn’t scale)

http://flickr.com/photos/heather/464504545/
http://flickr.com/photos/curiouskiwi/165229284/
http://flickr.com/photo_zoom.gne?id=42914103&size=l
http://flickr.com/photos/madstillz/354596905/
http://flickr.com/photos/laughingsquid/382242677/
http://flickr.com/photos/bng/46678227/

Scaling Twitter

More Related Content

What's hot

What's hot (20)

Similar to Scaling Twitter

Similar to Scaling Twitter (20)

More from Blaine

More from Blaine (6)

Recently uploaded

Recently uploaded (20)

Scaling Twitter