openstack-intro
Cloud computing with OpenStack
An introduction
This presentation consists of two parts:
- First, it deals with the question: What is Cloud Computing? Which challenges is it supposed to address?
- The second part gives a rough overview over the OpenStack cloud solution -- how does it work?
So what is Cloud?
Flickr/Daniel Spiess, CC-BY-SA
Everybody's talking about Cloud Computing. But there's no general definition of it! There is no general definition of "the Cloud", and the term's used in an inflationary manner.
IaaS
SaaS
PaaS
Maybe you've also heard of these acronyms: Infrastructure as a Service, Software as a Service, Platform as a service. When using these terms, everybody's somehow thinking of the cloud, but nobody knows what the cloud really has to do with all of this.
Flickr/Mark Norman Francis, CC-BY-NC
"So you want Cloud Computing ..."
and that tends to leave some of us slightly confused, at times. The interesting question to answer first thus is: What challenge is Cloud Computing supposed to address? And after all, what is it? To solve this problem, first take a look at the typical relation between an infrastructure service provider, the customer and the technical environment.
Flickr/Leonardo Rizzi, CC-BY-SA
First of all, there's the infrastructure provider.
He wants:
- to satisfy his customers with his service -- because they'll remain customers only that way
- to optimize his own infrastructure so that running his platform causes him as little extra work as possible
- to run his infrastructure as cheap as possible because that saves him money.
Automation
Long story short: The infrastructure provider wants automation!
Then, here we have our customers (typically, these are Hosting customers).
They want:
- a perfect service to be offered without any outages
- to be as flexible as possible (without long-running contracts)
- to pay as little as possible.
Automation
That's possible with Automation, too!
Now let's take a look at the technical environment involved in this. This typically connects service providers and customers. And in the conventional data center, it suffers from some design problems:
For the infrastructure provider:
- Lots of iron in his DC
- High cost for acquiring and maintaining hardware
- High power consumption
- Low utilization, little efficiency
For the customers:
- Complicated, non-automatic setup process
- Long-running contracts
- This leads to high overall cost
?
The interesting question is: How can we modify the technical environment to make the service provider and his customers a perfect match?
The first step on the road was virtualization. Typical virtualization setups look like this:
- There's centralized storage, in this case, it's a typical SAN
- We also have a bunch of virtualization nodes, accessing that storage
- These might be cluster controlled for HA purposes
- They start virtual machines via a hypervisor
Automation?
Automation
Not much automation going on here:
- Manually provision storage
- Manually provision network resources
- Manually install an OS
- Manually add it to the management infrastructure
No self-service capability at all.
So let's go back to the customers and see what they would really need.
a GUI. An easy and intuitive one, too! Whatever they want to do, they need an interface that allows them to do it without being a rocket-science thing.
In addition to a GUI, we also want CLIs and client libraries so using the cloud from a script or application is as easy as manipulating it from a graphical interface.
Authentication, authorization, and access control. The service provider also has to provide a framework for access restrictions. Different people in a company are allowed to do different things, and the setup has to resemble this.
The service provider also has to provide prepared images containing an operating system, preferably as simple as getting soda from a soda machine.
A solution must exist to integrate new VMs automatically into the eventually existing network infrastructure. This includes VLAN integration, this also included assigning freshly created VMs a new IP address automatically.
And last but not least, we need a central controlling instance to keep all the things I mentioned earlier together.
July 2010
Austin
October 2010
Race through these slides quickly.
Bexar
February 2011
Cactus
April 2011
Diablo
September 2011
Essex
April 2012
Folsom
September 2012
Grizzly
April 2013
Havana
September 2013
Icehouse
April 2014
Juno
October 2014
Kilo
April 2015
Liberty
October 2015
Mitaka
April 2016
Newton
October 2016
> 300
Sponsors & supporting companies
As per https://www.openstack.org/foundation/companies/, the OpenStack Foundation currently (November 2016) recognizes 381 companies as Members, Corporate Sponsors, or Supporting Companies.
> 50
Subprojects
54 projects in the Big Tent (2016)
Horizon
Keystone
Glance
Neutron
Nova
Automation?
Automation!
Add conclusion notes here.