Traditional storage was not designed to manage the high volume of I/O workloads characteristic of today’s multi-tenant VMware virtual machine (VM) environments. Consequently, many organizations fall into the trap of over provisioning storage infrastructure to help ensure good virtual application quality of service (QoS).
The challenge is that the capital and operational expense of continuously re-architecting storage can derail the ROI of the virtualization project. The solution is storage that is specifically designed to work with the unique attributes of a VMware server infrastructure that drives enhanced efficiencies into the data center.
Purpose Built for Virtualized Storage
To solve this issue, VM-Aware storage systems are beginning to emerge. These are systems that have a granular understanding of the virtualized environment including the specific I/O requirements of particular VMs.
Tintri is a storage system based on the VM-Aware Storage architecture. By integrating directly into the virtualization operating stack and including the virtualization tool chain such as the vCenter management console, Tintri VMstore granularly provisions storage capacity and manages VM I/O workloads to ensure virtual application QoS. For example, their patented “FlashFirst” technology utilizes a combination of flash capacity and conventional disk to efficiently manage active and inactive data. Through this architecture, Tintri platforms are able to serve up 99% of all I/O from flash and provide sub-millisecond latency response times. Other forms of hybrid storage systems which are designed based on a “disk-first” architecture and use a bolted-on approach to flash typically serve only 40% of I/O from flash.
The over 50% increase in flash utilization provides Tintri with a price point advantage over All-Flash Systems and a performance advantage over traditional hybrid storage systems. To achieve similar levels of flash hit rate, IT planners would generally have to adopt either a hybrid flash array with a larger than normal compliment of SSD capacity or an All-Flash array system.
Guaranteed VM QoS
A good illustration of how VMstore strikes the right balance with allocating storage resources is its ability to manage individual VM QoS assurance profiles. Tintri storage systems assign QoS assurance profiles to ensure that each individual VM receives its required allotment of flash, I/O and storage compute resources; no more and no less. Between their FlashFirst and QoS management capabilities, Tintri can help organizations avoid the expense associated with over provisioned storage deployments.
For instance, an online transactional business application VM can live directly alongside a virtual desktop VM instance, without the risk of resource contention taking place between the two. This helps to prevent the “noisy neighbor” (one VM consuming all of the available resources) condition that can sometimes plague multi-workload environments. Traditional storage systems, on the other hand, typically need to over provision storage resources or rely exclusively on flash to help ensure that resources will not be siphoned away by rogue VMs. And even then, there’s no guarantee that storage controller I/O resources won’t suddenly be monopolized by a single VM.
Less Storage Management is More
Another major challenge is that conventional storage systems are designed to configure disk LUNs and volumes. VM performance management and protection tasks are left entirely up to the storage or virtualization administrator. This places undue burden on operations personnel to constantly analyze and remedy performance and protection issues in highly dynamic virtualized environments. Administrators may have to hop between different tools such as vCenter, their switch fabric management console and their storage system GUI to determine where performance bottlenecks are occurring. In some environments, this could take days or even weeks to resolve.
Tintri instead takes a “Zero Storage Management” approach. Through its virtualization integration, Tintri VMstore provides a global view of all VMs and displays a consolidated view of the real-time health of critical business systems. At a glance, virtual administrators can ascertain any performance or latency issues and quickly determine where bottlenecks occur. With the recently announced “Tintri Global Center” software management product, multiple Tintri storage units can be managed as a single entity; making it easier for administrators to scale and manage their virtualization storage capacity as their virtualized application server and end-user desktop environments grow.
Tintri also simplifies data protection and disaster recovery tasks with VM-level snapshot and efficient and affordable replication function. This means there is no need to worry about decoding the mapping of VMs to LUNs or volumes and deploying expensive disaster recovery solutions that are virtualization-aware.
George Crump from Storage Switzerland and Geoff Stedman from Tintri hosted a live webinar to discuss the need for storage that is better aligned with virtualization and the advantages of Tintri Zero Management Storage. You can view the webinar here: