16. NINJA
SPEED
9 GLOBAL
DATACENTERS
Singapore ⋅ Tokyo ⋅ Sydney ⋅
Sao Paulo ⋅ N. Virginia ⋅ Oregon ⋅
N. California ⋅ Ireland ⋅ Frankfurt
HHVM
WP BOOSTER CDN
20. AVERAGE
RESPONSE TIME
1566 ms 3108 ms
NUMBER OF
PROCESSED
REQUESTS
5387 2637
NGINX CACHE
EFFECTIVENESS
15 ms 386830
HHVM php-‐fpm
21. HHVM
although slower response,
HHVM is s*ll opera*ng
(un*l the end)
Around 150-‐200 clients,
php-‐fpm is unmeasured &
50x errors were received
HHVM vs. php-‐fpm
23. HHVM 3.6 vs. 3.9
loader.io
TEST
1
minute
0 -‐ 300
client
increase
24. HHVM 3.6 vs. 3.9
Avg response Yme: 10563 ms
Success responses: 1189
Avg response Yme: 8259 ms
Success responses: 1685
HHVM 3.9
Response *me improved by 20% & the total request
processing number increased by 50%.
32. MULTILINGUAL
NEEDS TO BE HIGH SPEED
MULTISITE
HUGE ENTERPRISE
SITE
1 AMIMOTO instance =
can have mulTple
WordPress installs
Install MulTlingual
WP plugins
63. can remain with you
even there is no associated instance
returned (and cannot be retrieved)
as soon as you stop a Virtual Machine
OpYonal
When you create a Virtual Machine,
you get a Public IP
No maUer how many Ymes
you stop / start an intance with EIP,
the same IP will conYnue to be available
as the public IP of the Virtual Machine.
If there is no EIP,
each Yme you stop and start the instance,
you get a new Public IP every Yme.
hUp://blog.digitalcube.jp/archives/1870
Elas*c IP Address vs. Public IP Address
106. DigitalCube Team & What We Do
WordPress + Amazon Web Services
AMIMOTO
Performance Tests
HHVM vs. php-‐fpm
HHVM 3.6 vs. 3.9
Sites on AMIMOTO
WordPress + Amazon Web Services
Instance SelecYon
Connect WP with tools
Autoscaling
Machine Learning (for e-‐commerce)
AWS Console
AWS Console / Dashboard
Launch Virtual Server
Configure Instance
Key Pair
ElasYc IP Address
Set up WordPress
AcYvate pre-‐installed plugins
NGINX Cache Controller Plugin
Scale up & down of instance
Route 53
Terminate Instance
Release ElasYc IP Address