1 of 24

2 of 24

Google Compute Engine

Batch processing on demand

Julia Ferraioli, Developer Advocate

google.com/+JuliaFerraioli

​

GDG Sydney

9 October 2012

3 of 24

Agenda

  • Overview of Google Compute Engine
    • At 5000 feet / 1524 meters
    • In pictures
    • Spinning up
    • Dive into the GCE console
    • What's in a VM
    • Getting in and hopping around
    • Storing by three names
  • Putting those instances to work
  • Spinning down

4 of 24

Overview of Google Compute Engine

5 of 24

At 5000 feet / 1524 meters

Great for

  • The power of Google's infrastructure at your keyboard
  • Offers virtual machines, storage, networking
  • Allows services to scale efficiently with little work
  • Large scale analysis
  • Batch processing
  • Variable sized workloads

6 of 24

Google Compute Engine in pictures

7 of 24

Spinning up

gcutil

8 of 24

Dive into the GCE console

https://code.google.com/apis/console

9 of 24

10 of 24

Google Compute Engine in pictures

11 of 24

What's in a VM

Linux VMs

  • Root access
  • Ubuntu or CentOS
  • Modern CPU
  • 1, 2, 4, 8 CPUs
  • 3.75GB RAM per CPU

12 of 24

Getting in and hopping around

Connecting to the VMs

  • External IP assignments
  • Instance name ⇄ DNS name ⇄ hostname
  • Firewalls and configurations

Private Network

  • Isolated at the project level
  • Internal facing DNS (VM name ⇄ DNS name)

13 of 24

Storage by three names

3 Storage Options:

  • Ephemeral
  • Persistent
  • Cloud

​

Cloud Storage

  • Object storage
  • Access through API
  • Easy integration with �service accounts

​

​

14 of 24

Putting those instances to work

15 of 24

From the CLI

We will:

  • Add a firewall
  • Tunnel into our instances
  • Install Apache
  • Start serving
  • Access our instance from the outside world

16 of 24

Metadata

  • very useful for configuration & coordination of instances
  • key / value storage (strings)
  • project level values�$ gcutil setcommoninstancemetadata --metadata database_server:127.0.0.1
  • instance specific values�$ gcutil addinstance ... --metadata database_server:127.0.0.1
  • available via HTTP/JSON from the VMs:�$ curl http://metadata.google.internal/0.1/meta-data/attributes/foo ; echo

17 of 24

18 of 24

Using the metadata server

We will:

  • demonstrate querying metadata server for database_server

$ curl http://metadata.google.internal/0.1/meta-data/attributes/database_server ; echo

19 of 24

Processing batch jobs

We will:

  • Spin up 16 instances
  • Generate tiles for a fractal
  • Assemble the final, zoomable image with the Maps API

​

20 of 24

Spinning down

21 of 24

i can haz Compute Engine?

Right now:

  • Limited preview
  • Focused on batch workloads
  • Apply: http://goo.gl/edkDN
  • Talk to us! We're happy to discuss your use case

CC Image courtesy of London looks

22 of 24

Learn more

About Google Compute Engine:

​

About Google Cloud Platform:

​

About Us:

​

​

​

23 of 24

Questions?

​

Thanks!

https://developers.google.com/compute

​

google.com/+JuliaFerraioli

24 of 24