How do I install a storage accelerator on a server?
By submitting your personal information, you agree that TechTarget and its partners may contact you regarding relevant content, products and special offers.
Some I/O accelerators -- also called storage accelerators -- are server-agnostic and can be installed on any conventional x86 server that meets PCI Express (PCIe) bus requirements. For example, Fusion-io's ioDrive2 units and the P420m from Micron Technology require a PCIe 2.0 four-serial-lane slot. While PCIe slots on modern servers are common, there is little space for a bulky I/O accelerator. Also, there is rarely more than one slot available, so that presents a problem if another PCIe device is installed on the server, such as a multiport NIC adapter.
Some I/O accelerators are server-specific. For example, the HP IO Accelerator for BladeSystem c-Class units is designed for Gen8 HP ProLiant Blade Servers. These accelerators use an interface that will only function in that class of systems. This offers a mixed blessing: Vendor-specific accelerators are easier to install and proper integration is assured, but the units will typically command a premium price tag.
Regardless of the actual installation, it's important to note that storage accelerators are radically different local storage devices compared with conventional disks. Even well-refined and supported storage accelerators could experience operational problems with certain unexpected combinations of hardware and software.
Businesses that choose to adopt storage accelerators generally approach the technology only for specific mission-critical servers, and only after extensive product evaluation and proof-of-principal testing with the actual applications that will use the storage.
Dig Deeper on Enterprise data storage strategies
Related Q&A from Stephen J. Bigelow
Photon OS optimizes VMware Photon platform deployment, not only in vSphere but in GCE, EC2 and more. Follow these steps to learn how to run Photon OS...continue reading
Performance problems can be caused by a number of things, including overprovisioning and poor vCPU selection and assignment to VMs. Use these ...continue reading
Think about what types of workloads are running on a VM before assigning compute resources, and consider using vCPUs from different cores for ...continue reading
Have a question for an expert?
Please add a title for your question
Get answers from a TechTarget expert on whatever's puzzling you.