Throughout years of experience, I have encountered and handled performance issues for the application, server and network layers, but storage layers were often the hardest to address. Enterprise storage used to be monolithic – unmovable, little changes encouraged and a reluctance to reconfigure or adapt to changing needs of the applications. Modern enterprise storage allows easier, more flexible management with capability to control the application facing as well as backend configuration. Managing storage in the enterprise, however, is still a full-time task since more flexibility and more capabilities mean more options. Software Defined Storage (SDS) have been abstracting and hiding daily management decisions, but since no two environments are alike, storage admins should not rely only on SDS to cater for their specific environment needs.
And so, similar to the old storage solution, managing storage still requires sizable resources as storage decisions are still key to the success of the business. Turbonomic is the only platform that takes decisions throughout the entire stack, including vendor specific storage such as EMC ScaleIO, to help workloads self-organize efficiently.
Software Defined Storage – A Partial Solution
Storage admins tend to use rules-of-thumb or subjective experience to manage storage. This method is not consistent nor is it scalable, since workloads are dynamic and ever-changing and ever growing. Even the best storage admins will make mistakes.
Other admins rely on Software Defined Storage to manage itself, but it’s a partial solution.
- Application priority – your environment has multiple tiers of application priorities that constantly change, and traditional storage does not see the application usage. While SDS like ScaleIO is very good at managing itself, it has no abstraction or control over the applications using it, whether it's a database, a web server, financial or trading application, etc’
- Storage array type - based on the vendor, storage arrays can manage themselves, but have no abstraction or control over servers, virtual machines, containers or the applications themselves
- Affinity and anti-affinity – unrelated applications utilizing the same spindles/SSD's/RAM, or applications that should be separated due to usage pattern (for example heavy random writes), or applications that need to be separated due to regulatory requirements. SDS will usually abstract the data and as a result control on IO is lost.
Being vendor agnostic Turbonomic will get the most of your storage by learning which workloads should fit which data stores and controllers. Affinity, access type and priority are all taken into consideration in real-time; continuously.
As a result, storage contentions are prevented in real time before they even become contentions. Placement of new VMs or containers is handled at the same speed - real time.
Decisions Turbonomic drives with ScaleIO
Turbonomic discovers ScaleIO Protection Domains, Storage Pools, Devices and Volumes and learns their relationships to the applications, VMs, Containers and other related resources in the supply chain. Control decisions are made for:
- Capacity - Storage Pools and Protection Domains capacity will grow or shrink as needed
- Initial Placement and movement – A volume will be chosen for VMs and they will be moved across those volumes (Storage vMotion)
For access to the ScaleIO configuration and to make changes, Turbonomic uses API calls to the Gateway end-point server using REST API.
How to start working with Turbonomic and EMC ScaleIO
Please refer to this video for an install example.
- Go to the Admin tab
- Choose "Add" (top left)
- Choose "Storage" from the list of options
- Choose type of storage - ScaleIO
- Enter the IP, user and password. The IP is your ScaleIO API gateway server IP
- Click "Apply" on the bottom left
- If entered correctly, then the ScaleIO instance will appear on the Admin tab, ready to be discovered and used
Thank you Meir Laker for providing key information for this page