Google Compute Engine is an infrastructure-as-a-service (Iaas) product announced by Google at Google IO on June 29, 2012. The system allows you to setup virtual machines running CentOS 6.2, Ubuntu 10.04 LTS and Ubuntu 12.04 LTS on the Cloud Infrastructure of Google.
You will find plenty of documentation and articles about what the IaaS does but I wanted to stay at the ergonomics level.
When Google asked me what I would a service company to focus on “a decent web hosting control panel that provides a graphical interface and automation tools designed to simplify our administrative work”.
After several days, we received the “precious”: “Google is pleased to welcome you to the Google Compute Engine Limited Preview”.
The use case
The algorithm performing this optimization relies on Genetical Algorithms and 10+ years of research in cooperation with famous manufacturers.
The algorithm is categorized as a “number cruncher”, mainly because its main task is to perform mathematical calculations; there is no database besides the in-memory data and the output file presenting the positions of the machines within a plant.
Google Compute Engine architecture
Google Compute Engine might be seen like any virtual machine you connect to using Secure Shell (SSH). As an administrator on the “dev machine” you can:
- Instantiate a standard virtual machines (instance) made available by Google
- Create custom images made with the applications you are working with (Apache, Weblogic, MySQL, AppArmor, Ruby, Python, etc.)
- Instantiate a custom image you have previously made
- etc.. see the developer’ guide https://developers.google.com/compute/docs/overview
The administrator chooses where he/she wants to run the instance and how many virtual CPUs he/she will use to run the instance.
Once the instance is made available, the administrator:
- setup a persistent disk which will outlives an instance where you could store all data you have produced during the usage of an instance
- connects persistent disk with the remote machine
- setup the firewall in order to give access to the server using a fixed IP address to the world (or a part of it)
Now, it is time for us to fine tune or source code and use efficiently the vCPU Google is giving us access to for a fraction of the cost of a new server we should manage.
At the same time, we need to learn the business model of Google and ensure that the costs are timely tracked and our infrastructure kept secured.
We have been able to master Google Compute Engine and Google Cloud Storage in 2 “long” working days.
The artifacts made available by Google are awesome: the documentation, the simple tools (gcutil and gsutil used to manage Google Compute Engine and Google Cloud Storage), the security based on oauth 2.0 and the infrastructure are made to reduce the learning curve and reduce the workload of our teams.
Today, all members of our team use the Virtual machines whether they are techies or not! A page in our Wiki describes step by step how to start an instance and run the Genetical algorithms without hassle.
The MBA Corner
What is the PESTEL analysis of the macro-environment?
Political factors: The Belgian public administration and the European Commission subsidize SMEs who do invest in Cloud technologies to innovate and produce added value to their customers.
Economic factors: those days paying in Dollars is a significant windfall; the Pay-per-use business model reduces our CAPEX and Fixed while keeping a high level of R&D (human resources)
Social factors: a company attracts more easily new employees if they show an interest to the latest trends (even if we keep the traditional infrastructure management methods learned more than 20 years ago)
Technological factors: We have access to better and diverse technologies than the one we could afford in the short and the long run (SCSI interfaces, ephemeral disks up to 1770 Gb if necessary, etc.). Technically we could run the VM through a Smartphone running a Linux/Unix shell or any $49 Android PC System like APC.IO
Environmental factors: as a green company, we do reduce drastically our CO2 consumption: no more delivery from China, mutualisation of the usage of the Google data centers, etc.
Legal factors: Google uses all security you can think of in order to rely on it and focus on your business only. The CIO should ensure that his/her internal business process enforce a strong password policy, the main raison why hackers enters your system.
Here is one case that happened in the 2010: “…Chinese infiltration of the U.S. Chamber of Commerce ended up funnelling weeks of corporate data; even after the chamber thought it had reestablished security, it discovered that an office printer and a corporate apartment thermostat were still sending data – who knows what kind? – back to China…”  can happen.
 Why David Petraeus’s Gmail account is a national security issue : http://www.washingtonpost.com/blogs/worldviews/wp/2012/11/10/why-david-petraeuss-gmail-account-is-a-national-security-issue/
Run virtual machines at Google Scale [with Google Compute Engine] : https://developers.google.com/compute/
Google Compute Engine: Computing without limits: http://googledevelopers.blogspot.be/2012/06/google-compute-engine-computing-without.html
Google Compute Engine: Documentation: https://developers.google.com/compute/
Google Cloud Storage: Documentation: https://storage.cloud.google.com/
Google APIs Console: https://code.google.com/apis/console
CentOS home page: http://www.centos.org
Ubuntu 10.04 LTS: https://help.ubuntu.com/10.04/index.html
Ubuntu 12.04 LTS: https://help.ubuntu.com/12.04/index.html