New Simplivity and Nimble Models

It’s conference season again and that means a whole bunch of announcements from HPE across their Hyper-Converged and storage platform. There is a lot to get through so let’s get started:


The first announcement is around the SimpliVity hyper-converged platform. SimpliVity was an acquisition by HPE made at the beginning of last year, the platform provides a converged solution of HPE ProLiant compute and VMware hypervisor. Newly announced is that SimpliVity will be broadening supported hypervisors beyond just VMware by also offering Hyper-V supported nodes. There is not complete feature parity in the initial release, with features such as file level restores and Rapid DR not yet supported in the Hyper-V version. But the core SimpliVity functionality of global dedupe and compression plus VM clones is all there, if you are new to SimpliVity check out my quick start guide.

Like in the VMware flavoured version management is all within the native application, so for Hyper-V this is MS Virtual Machine Manager. Currently there are 10 different configurations of VMware SimpliVity model with the Hyper-V variation at launch there will be with a smaller portfolio of 4 offerings. These will be aimed at what today is the most common requirement for Hyper-converged platforms, small and medium deployments commonly found at ROBO sites.


The storage news focuses on the Nimble platform and includes a refresh of the hardware across the range.

Store more guarantee – HPE are betting that they have pound for pound the most efficient array with their Nimble systems by offering a capacity efficiency guarantee. If you can show that another system is more capacity efficient HPE will match the difference with additional capacity. HPE’s confidence to make this bold claim comes from their architecture which has a low overhead before even considering any data reduction techniques. Any storage system will have an overhead for sparing, system use, RAID etc HPE claim that their Nimble product has less of an overhead than the competition. Once data reduction is applied this free space available from a lesser system overhead brings a significant advantage due to the multiplier effect when compression and dedupe are applied.

SCM / NVME –   Nimble storage system are future ready with a promise from HPE that in the future customers will be able to non disruptively upgrade to Storage Class Memory (SCM) and NVMe. HPE see the maximal benefit will be achieved by coupling SCM and NVMe together predicting this could achieve up to a 10x performance boost versus NVMe and flash.

New models – The new range starts with the AF20Q and AF20 which start at 6TB of useable storage and scales up to the AF80 which scales to 4PB of useable capacity, these figures assume 5:1 data reduction ratio.

The hybrid family are called HF scaling from the entry HF20H which offers 821TB of useable capacity the HF60 which has 5PB of useable capacity. The HF20C is designed for workload that require compression only and would not benefit from dedupe.

Both the Hybrid and all flash models have updated intel processors inside which is one of the key drivers to increased performance, which HPE are estimating is +220% versus the current model.

Inline always on dedupe – The Nimble hybrid models benefit from the addition of inline dedupe.  This essentially means that deduplication is performed as the data is entering the array.

You can see a summary of the Nimble announcements in this video




Pivot3 Announces Acuity

Last week Pivot3 announced a new generation of its hyper-converged models. I took interest as I had noted this company was experiencing significant growth and wanted to see what they were all about. When I dug into it I uncovered an interesting story of product acquisitions and then the evolution of product integrations, definitely interesting as we see HPE’s Simplivity and Nimble acquisition evolve.

Doing Hyper-Converged before it was cool

Let’s start at the beginning Pivot3 are arguably one of the originators of the hyper-converged scene founded back in 2003, before it was even called hyper-converged. If you’re new to hyper-converged read storage and virtualisation in a box. Similar to how other hyper-covered platforms work each node has a VM that sits on the ESXi host and acts as the gateway to the storage. The storage across the nodes is grouped together to form a single pool of storage from which datastores can be carved. Pivot3 have until now called their OS vSTAC OS and the data is written across the nodes using erasure coding.

NexGen Acquisition

NexGen had a colourful history having been co-founded by one of the founders of LeftHand networks, later being bought by Fusion-IO which itself was subsequently bought by SAN disk. It was then detached from SAN Disk and acquired by Pivot3 in January 2016, simple right.

Prioritising the important stuff

The interest from Pivot3 in NexGen was in its PCIe flash and QoS functionality. QoS in a storage system, like its better known counterpart in networking, assures performance levels to certain workloads even when a system is busy. Initially Pivot3 sold the NexGen system in two ways as a standalone unit and secondly as a package with a Pivot3 hyper-converged system, they called this second approach vSTAC SLX. Physically vSTAC SLX consisted of a Pivot3 hyper-converged cluster connected to a NexGen PCIe flash array. This essentially allowed data to be tiered between the Pivot3 nodes and the NexGen layer for higher performance. The QoS software could control the placement of data dependent on the performance requirements.

The aim here was clear to present hyper-converged as a system that can run multiple workloads concurrently by assuring service levels with QoS. This would help to broaden the appeal of hyper-converged systems which today are predominantly focused on single use cases.

Acuity Released

The latest release from Pivot3 drops the vSTAC name and becomes Acuity, it builds on the fundamental ideas laid out in the SLX release but rather than being a NexGen + Pivot3 bundle the QoS and PCIe flash functionality is available natively within the Pivot3 nodes. It would appear they have managed to pull the smarts from the NexGen box and package it within the Pivot3 nodes.

The line-up of available nodes is shown in the graphic below. As before there is the choice of all flash or hybrid nodes.  But additionally now there are the accelerator nodes which are essentially the standard flash and hybrid models with an additional NVMe layer to give a performance boost.Those with the NVMe functionality can offer up to 3.2TB of NVMe flash per node, which Pivot3 advise will be 450% faster that SATA SSD and 119% faster than SAS SSD.

The QoS targets can be set to control minimum IOPs, minimum throughput and latency. These parameters are set through polices which can also be scheduled to change for example for month end reporting.

To stay up to date with all the latest and greatest news from the storage industry make sure you are subscribed to our e-mail list.