Planning a data storage system for openstack platform

• provide virtual resources on-demand via a self-service portal;
• scale applications as the load increases;
• increase the speed of deployment of ultimate applications;
• provide users with a more convenient way of storing data (corporate DropBox);
• ultimately, significantly reduce IT costs.

The process of designing the infrastructure of a corporate/public cloud based on OpenStack platform is always accompanied by the question — what kind of data storage system to use? We will tell you about different solutions used to create data storage systems that Services experts use in their practice.

Data storage system in OpenStack

Data storage is used in many components of OpenStack:

• OpenStack Glance — VM image storage
• OpenStack Nova — VM ephemeral disks
• OpenStack Cinder — permanent disks
• OpenStack Swift — object storage
• OpenStack Manila — shared file system

Any guest operating system uses a block device to place file systems. «Ephemeral» and permanent data storages stand out.

«Ephemeral» discs exist as long as there is a VM (if it is removed an ephemeral disk is also removed.) Those are usually root disks with an operating system, swap disks. When deploying only OpenStack Nova, users do not have access to any form of permanent storage by default. From the point of view of a user «ephemeral» disks connected to VM disappear when a virtual machine is removed. Permanent block devices are not removed together with a virtual server and can be connected to another server if necessary. Of course, you can create and remove such devices, as well as manage connections. The table below contains a comparative analysis of the data storage components available in OpenStack.

Possible scenarios

The chart below displays the results of a survey of OpenStack users (October 2015): what low-level implementations do they use to store disks permanently.

Find out more on the TAdviser website