Nutanix goes all-flash, will rescue dying data from imploding cities

Converged system startup Nutanix has topped out its range with an all-flash appliance, and added protection at inter-city distances.

Its clusterable appliances combine servers, storage and networking to provide a single converged data centre that, it says, replaces racks of disparate servers, storage arrays and network gear. Nutanix claims it's far simpler to order, deploy, operate and manage than conventional separately-sourced IT gear.

The NX-9040 has six 800GB or 1.6GB SSDs, meaning 9.6TB of total raw capacity. The previous high-end system, the NX-8150, features 4 400GB, 800GB or 1.6TB SSDs and twenty 1TB HDDs, making a maximum of 20TB of disk and 6.4TB of SSD.

Processing power for the NX-9040 comes from dual Intel Ivy Bridge E5-2690v2 with 20 cores and running at 3.0 GHz. The NX-8150 uses dual Ivy Bridges too; an E5-2690v2 with 20 cores @ 3.0 GHz and an E5-2697v2 with 24 cores at 2.7 GHz.

Nutanix says the NX-9040 can support up to 115 virtual machines (VMs) but it doesn’t provide an equivalent number for the NX-8150. (We have asked them to supply this number and will update if they get back to us)

It’s designed to run “applications with large working sets, such as databases supporting online transaction processing” (OLTP), according to the marketing blurb. Deduplication and compression raise the effective capacity. The apps get fast, predictable and consistent data access latency. Performance scales out linearly as nodes are added into a cluster.

Data is said to be held local to the server whose apps are accessing it; that avoids network latency.

Nutanix NX-9000

Synchronous mirroring is the means whereby Nutanix offers protection through Metro Availability, meaning data centres up to 400km away. The functionality needed is built into Nutanix’s software and Nutanix says: “Virtualisation teams can now non-disruptively migrate virtual machines between sites during planned maintenance events, providing continuous data protection with zero recovery point objective (RPO) and a near zero recovery time objective (RTO).”

One-to-many and many-to-one topologies are supported and individual virtual machines can be mirrored.

Hyper-converged systems exemplify the rediscovery of clustered servers. Remember VAX clusters? Their popularity will increase until (or if) centralised shared storage offers countervailing advantages. Pendulums swing in data centres as well as in grandfather clocks.

The NX-9000 starts from $110,000/node and can be ordered now. Metro Availability will come with a software update, NOS v4.1, and will be included in the NOS Ultimate Edition. ®