Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Scaling Rails for Black Friday / Cyber Monday a...
Search
Christian Joudrey
February 18, 2015
Technology
6
5.6k
Scaling Rails for Black Friday / Cyber Monday at Shopify
Talk given at ConFoo 2015 on February 18th, 2015 and RailsConf 2015.
Christian Joudrey
February 18, 2015
Tweet
Share
More Decks by Christian Joudrey
See All by Christian Joudrey
Writing NES games! with assembly!!
cjoudrey
1
680
Developing at Scale
cjoudrey
3
460
Tips and Tricks from Shopify's codebase
cjoudrey
2
560
Scaling Shopify
cjoudrey
3
520
#pairwithme
cjoudrey
3
240
Two-factor authentication
cjoudrey
4
380
Automate your Infrastructure with Chef
cjoudrey
9
570
Other Decks in Technology
See All in Technology
Mocking in Rust Applications
taiki45
2
420
なにもしてないのにNew Relicのデータ転送量が増えていたときに確認したこと
tk3fftk
2
230
「自動テストのプラクティスを効果的に学ぶためのカードゲーム」 ( #sqip2024 )
teyamagu
PRO
2
190
自作Cコンパイラ 8時間の奮闘
soukouki
0
860
とあるOSSを継続可能にするための取り組みについて / OSS Refactoring Process
bun913
1
220
『GRANBLUE FANTASY Relink』キャラクターの魅力を支えるリグ・シミュレーション制作事例
cygames
0
170
より快適なエラーログ監視を目指して
leveragestech
4
1.5k
フルカイテン株式会社 採用資料
fullkaiten
0
32k
コンポーネントテストの手法と その効果を考える
yotahada3
3
370
言葉は感情の近似値である。その感情と言葉の誤差を最小化しよう ~コミュニケーションにおけるアナログ/デジタル変換の課題に立ち向かう~
nktamago
0
250
タイミーのレコメンドにおける ABテストの運用
ozeshun
1
220
開発者の定量・定性データを組み合わせて開発者体験を把握するための取り組み
ham0215
1
180
Featured
See All Featured
Infographics Made Easy
chrislema
239
18k
Principles of Awesome APIs and How to Build Them.
keavy
125
16k
The Pragmatic Product Professional
lauravandoore
31
6.2k
ReactJS: Keep Simple. Everything can be a component!
pedronauck
663
120k
A Modern Web Designer's Workflow
chriscoyier
692
190k
The Art of Programming - Codeland 2020
erikaheidi
48
13k
Testing 201, or: Great Expectations
jmmastey
36
7k
WebSockets: Embracing the real-time Web
robhawkes
59
7.3k
Creatively Recalculating Your Daily Design Routine
revolveconf
215
12k
Why Our Code Smells
bkeepers
PRO
334
56k
What’s in a name? Adding method to the madness
productmarketing
PRO
21
3k
The MySQL Ecosystem @ GitHub 2015
samlambert
250
12k
Transcript
scaling rails for Black Friday Cyber Monday
cjoudrey @
None
None
None
None
None
the stack
nginx unicorn • rails 4 • mysql 5.6 (percona) ruby
2.1 •
95 app servers 1,800 unicorn workers 18 job servers 1,400
job workers
scale?
None
400,000 reqs/min
3.7B$ annual GMV that’s 7,000$ per min
Black Friday Cyber Monday
Black Friday Cyber FUNday
None
~ 600,000 reqs/min
1 request 1 process =
scale++ ↓ resp. time ↑ workers
page caching
None
None
shopify/cacheable
generational caching
gzip • etag + 304 not modified
class PostsController < ApplicationController def index response_cache do @posts =
@shop.posts.paginate(params[:page]) respond_with(@posts) end end def cache_key_data { shop_id: @shop.id, path: request.path, format: request.format, params: params.slice(:page), shop_version: @shop.version } end end
md5( { shop_id: 1, path: '/posts', format: 'text/html', params: {
page: 2 }, shop_version: 123 }.to_s ) GET /posts?page=2
None
sale
query caching
shopify/identity_cache
full model caching
opt-in by design
after_commit expiry
class Product < ActiveRecord::Base include IdentityCache has_many :images cache_has_many :images,
:embed => true end product = Product.fetch(id) images = product.fetch_images
class Product < ActiveRecord::Base include IdentityCache cache_index :shop_id, :handle, :unique
=> true end Product.fetch_by_shop_id_and_handle(shop_id, handle)
None
sale
background jobs
webhooks emails • fraud detection • payment processing
None
priority queues payment • default • low realtime •
throttling
now what?
measure it! if it moves...
statsd
shopify/statsd-instrument
Liquid::Template.extend StatsD::Instrument Liquid::Template.statsd_measure :render, 'Liquid.Template.render'
PaymentProcessingJob.statsd_count :perform, 'PaymentProcessingJob.processed'
None
None
load testing
genghis khan
simulate Black Friday Cyber Monday before it happens
several times per week
slow queries
# User@Host: shopify[shopify] @ [127.0.0.1] # Thread_id: 264419969 Schema: shopify
Last_errno: 0 Killed: 0 # Query_time: 0.150491 Lock_time: 0.000057 Rows_sent: 1 Rows_examined: 147841 Rows_affected: 0 Rows_read: 147841 # Bytes_sent: 1214 Tmp_tables: 0 Tmp_disk_tables: 0 Tmp_table_sizes: 0 # InnoDB_trx_id: FF7021AAA # QC_Hit: No Full_scan: No Full_join: No Tmp_table: No Tmp_table_on_disk: No # Filesort: Yes Filesort_on_disk: No Merge_passes: 0 # InnoDB_IO_r_ops: 0 InnoDB_IO_r_bytes: 0 InnoDB_IO_r_wait: 0.000000 # InnoDB_rec_lock_wait: 0.000000 InnoDB_queue_wait: 0.000000 # InnoDB_pages_distinct: 475 SET timestamp=1393385020; SELECT `discounts`.* FROM `discounts` WHERE `discounts`.`shop_id` = 1745470 AND `discounts`.`status` = 'enabled' ORDER BY ISNULL(ends_at) DESC, ends_at DESC LIMIT 1
determining root cause
https://github.com/newobj/nginx-x-rid-header nginx request_id header proxy_set_header X-Request-ID "$request_id"; log_format main '...
$request_id' step 1
https://gist.github.com/mnutt/566725 Complete 200 OK in 100ms (Views: 60ms | ActiveRecord:
40ms | request_id=bc12813bce...) log_process_action ActionController::Instrumentation step 2
https://github.com/basecamp/marginalia User Load (0.3ms) SELECT `users`.* FROM `users` WHERE `users`.`id`
= 1 LIMIT 1 /*application:Shopify, controller:users,action:show, request_id:bc12813bce...*/ basecamp/marginalia step 3
# User@Host: shopify[shopify] @ [127.0.0.1] # Thread_id: 264419969 Schema: shopify
Last_errno: 0 Killed: 0 # Query_time: 0.150491 Lock_time: 0.000057 Rows_sent: 1 Rows_examined: 147841 Rows_affected: 0 Rows_read: 147841 # Bytes_sent: 1214 Tmp_tables: 0 Tmp_disk_tables: 0 Tmp_table_sizes: 0 # InnoDB_trx_id: FF7021AAA # QC_Hit: No Full_scan: No Full_join: No Tmp_table: No Tmp_table_on_disk: No # Filesort: Yes Filesort_on_disk: No Merge_passes: 0 # InnoDB_IO_r_ops: 0 InnoDB_IO_r_bytes: 0 InnoDB_IO_r_wait: 0.000000 # InnoDB_rec_lock_wait: 0.000000 InnoDB_queue_wait: 0.000000 # InnoDB_pages_distinct: 475 SET timestamp=1393385020; SELECT `discounts`.* FROM `discounts` WHERE `discounts`.`shop_id` = 1745470 AND `discounts`.`status` = 'enabled' ORDER BY ISNULL(ends_at) DESC, ends_at DESC LIMIT 1 /*application:Shopify,controller:orders,action:pay, request_id:bc12813bce...*/ profit!
access.log rails.log slow_query.log profit! (2)
bonus! background jobs
schema migration with zero downtime
soundcloud/lhm
class NewIndexOnOrders < ActiveRecord::Migration def self.up Lhm.change_table :orders do |m|
m.add_index [:shop_id, :customer_id] end end def self.down # end end
orders 20140520_orders
insert/delete/update triggers
INSERT INTO ... SELECT ... insert/delete/update triggers
async caveat
resiliency
A resilient system is one that functions with one or
more components being unavailable or unacceptably slow. -
don’t let a take you down minor dependencies
shopify/toxiproxy ☢ redis memcached sessions your app toxiproxy
def load_customer if customer_id = session[:customer_id] @customer = Customer.find_by_id(customer_id) end
end
def load_customer if customer_id = session[:customer_id] @customer = Customer.find_by_id(customer_id) end
rescue Sessions::DataStoreUnavailable @customer = nil end
def test_storefront_resilient_to_sessions_down Toxiproxy[:sessions_data_store].down do get '/' assert_response :success end end
rinse & repeat http://www.shopify.com/technology/16906928-building-and-testing-resilient-ruby-on-rails-applications
what about resources? slow
shard 2 shard 3 shard 1 shop 4, 5, 6
shop 7, 8, 9 shop 1, 2, 3
rails request shard 2 shard 3 shard 1 shop 4,
5, 6 shop 7, 8, 9 shop 1, 2, 3
rails request shard 2 shard 3 shard 1 shop 4,
5, 6 shop 7, 8, 9 shop 1, 2, 3
how can we fail fast?
shopify/semian smart circuit-breaker
shopify/semian Semian.register(:mysql_shard_1, tickets: 5, timeout: 0.5, error_threshold: 100, error_timeout: 10,
success_threshold: 2)
shopify/semian Semian[:mysql_shard_1].acquire do # Query the resource end
rails request shard 2 shard 3 shard 1 shop 4,
5, 6 shop 7, 8, 9 shop 1, 2, 3
what else can go wrong?
Shopify Shipping rate providers Payment gateways Fulfillment services (FedEX, UPS,
USPS, etc..) (Stripe, PayPal, etc..) (Shipwire, etc…) Internal services (MySQL, Memcached, etc..)
manual circuit breakers around external dependencies
thanks! :)