Mirantis Training Update, June 2015: Answering Common Student Questions
Welcome to Mirantis Training’s new monthly Q&A section. During our training sessions for OpenStack, we often get some excellent questions from our students. We’ve kept track of all of our students’ questions and will start working our way through answering those questions here on this blog.
If you have a question that you would like a Mirantis technical instructor to answer, please ask it in the comments section below and we’ll try our best to answer it in next month’s post.
What does HA mean?
HA means “High Availability” and refers to efforts to minimize system downtime and loss of data. Most system designers make an effort to architect their system in a resilient way such that disruptions in their application stack are not noticed by their end users. A simple example of this is the use of load balancers in front of a cluster of web servers.
The load balancers are configured to receive traffic and transparently forward the request to one of the web servers in the cluster. In this scenario, if a particular web server goes down, the load balancer simply uses other web servers and is still able to serve the request without any indication to the end-user that one of the web servers has failed.
How does OpenStack provide High Availability to VMs?
OpenStack provides High Availability to VMs in several different ways. At the most basic level, the environment can be architected and deployed with high availability in mind by following the High Availability Guide. This can be achieved by setting up the foundation components (message queues and database servers) as redundant services such that a loss or failure of any individual component does not bring down the whole environment.
Going higher up the stack, redundant networks can be deployed so that each bare metal server is connected to multiple network switches. At the app layer, an application architected for the cloud can be built and deployed across multiple VMs such that the loss of any individual VM does not impact the overall application. Finally, for ultimate protection against loss of availability, the VMs can be geographically distributed by running app VMs in multiple datacenters. Of course, in order to support this, the application has to be built with this kind of distribution in mind. One of the most well-known examples of a large application deployed in a highly available fashion is the Netflix streaming service. You can see a really good presentation of how this was architected here.
Do you have additional questions about OpenStack? Schedule training for OpenStack with Mirantis Training. Our training is purely based on 100% open-source OpenStack, with no vendor specific versions and/or proprietary implementations. For further details, take a look at the Mirantis Training course catalog below: