In the last quarter of a century, CPU performance has increased 2,000,000 times, while traditional storage array performance has increased only by an order of magnitude. This radical improvement in CPU performance, coupled with memory density increases and cost decreases, has made large-scale server virtualization possible, for companies of all sizes.
But the failure of traditional enterprise storage to keep pace with performance improvements in other areas of basic system architecture creates a bottleneck in server virtualization: at the level of the shared storage pool on which hypervisors depend, for their operational performance and for their economic benefit.
Cost-effective, high-performance server virtualization depends on an underlying storage infrastructure that can:
- Respond effectively, with predictable performance, to radically different workloads that are competing for underlying shared storage resources
- Provide guaranteed (and perhaps SLA’d) quantities of IOPS for every 100 GBs of storage allocated to a each virtual machine.
- Deliver required aggregate IOPS to the hypervisor without radically overprovisioning capacity (and increasing expense)
- Prevent poorly-configured or resource-hungry virtual machines from damaging the performance and availability profiles of “neighboring” virtual machines
- Integrate effectively with the storage management features and commands of the hypervisor, without requiring separate configuration and management, allowing hypervisor administrators to manage storage as an integral part of their domain without becoming storage experts, while providing the detailed telemetry required to diagnose performance problems on a per-VM basis.
What our customers say:
BroadAspect – Cloud Service Provider
“Being an end user, cloud provider and a reseller puts BroadAspect in a unique position and as a result we need a reliable storage solution as our foundation. We were looking for a partner that would provide us with usable features, ultimately contributing positively to how we do business. X-IO’s unique ‘Zero Touch’ approach is just one of these features saving us time and money. Not to mention the five-year warranty, which has significantly reduced our storage costs. This means that we can continue to grow and service our growing customer base, safe in the knowledge that we are fully supported by a high performance storage solution.”
- CTO and VP Cloud Services
Since the X-IO ISE is, in essence and by design, a virtualized storage engine in a 3u form-factor, it is ideally suited for the “I/O blender” environment produced by large-scale server virtualization environments. ISEs, operating as a single pool of storage under the control of the hypervisor, ensure that proper mix of flash and hard disk storage are allocated to the VMs supported by the hypervisor, and ensure that high-demand virtual machines get their data via the fastest-available media class regardless of the workload profiles of other virtual machines.
The high availability and reliability features of the ISE ensure that the hypervisor is based on a rock-solid, continuously available storage layer. High-performance, high-reliability virtualized server environments depend on the ability of operations and management (O&M) personnel to view and manage all virtual resources, including the storage supporting the hypervisor, from the viewpoint of that hypervisor and the virtual machines it supports.
Unlike traditional enterprise storage arrays, which overprovision capacity to provide server virtualization environments with the minimum required IOPS, and unlike upstart hybrid storage arrays, which cannot run at rated performance when their used capacity exceeds 60-70%, the X-IO ISE comes with a rated-performance-at-rated-capacity guarantee, which means that our server virtualization customers get all the IOPS they pay for, and can use 100% of each ISE’s rated capacity without losing any of the ISE’s rated performance.
These unique attributes make the ISE the storage industry’s lowest-cost option for large-scale server virtualization.
VMware Stretched Cluster Support with ISE Active-Active
For the ultimate in enterprise, stretched cluster availability take a look at this solution. ISE with Active-Active Mirroring is the only array on the market with the in-built capability of performing synchronous, active-active mirroring for a truly disaster tolerant architecture. Many enterprises requiring high availability cannot afford to implement a disaster tolerant solution that does not require manual intervention or a need to set read-only access on the passive side of the mirror.
« Get your copy of the reference architecture now!