This article is more than 1 year old
In 2018, how are the big storage 8 handling the industry's challenges?
Faster access, data management and HCI
Analysis All the main storage system players face three large challenges in 2018 – new technologies providing faster data access, data management suppliers aiming to wrap a commoditising abstraction layer around them, and public cloud and public cloud-like vertical stack suppliers aiming to make them irrelevant.
We look at the eight main storage players – Dell Technologies, HPE, Huawei, IBM, NetApp, Nutanix, Pure Storage and Vantara – along with Kaminario and Infinidat from this 50,000ft point of view.
The new technologies include fast networked storage access with NVMe over Fabrics, storage-class memory (SCM), and 4-bits per cell (QLC) flash. AI and machine learning is being adopted by virtually everyone as they seek to improve storage product administration and management. But storage access stack rewriting from startups like ReduxIO, WekaIO and StorOne is so new that there is no significant market adoption to give us any idea of how important it will be yet.
The commoditising abstraction layer providers look at all storage in the form of primary data, or secondary storage with Cohesity, Actifio, and others, and secondary storage protection, governance and more with suppliers like Druva, Rubrik and Veeam, plus many others. Their shtick is that you use their product to optimise storage and a corollary is the storage system becomes a commoditised lower stack component.
Containerisation is also affecting how storage is used but it's not clear yet what the overall effect will be – so we will ignore that technology here.
We can also say that, broadly speaking, shared external array storage is under attack from server SAN/hyperconverged infrastructure (HCI) systems with performance-intensive edge cases, like HPC and scale-out filers, less susceptible to the HCI onrush.
How do the main storage suppliers rate here?
Dell Technologies is well protected from one of the three identified trends because it has its own HCI and vertical stack offerings. Indeed, it is pushing HCI strongly and seeing its vertical stack offers built on its HCI base. It is also in the scale-out filer space with Isilon.
Where it has a possibly harder time competing is in data management. The growth rates for Actifio, Cohesity, Druva, Rubrik and Veeam show no reduction due to Dell competition.
Dell is also hip to the new technology noted above because of its DSSD learning experience (NVMe over Fabrics), and its server operation which will be aware of storage-class memory concepts. Its XtremIO operation should also be up to speed on both SCM and QLC flash, as it is a major prospective OEM channel outlet for suppliers of both technologies.
Where Dell might say it is challenged is in the vertical stack area. Dell EMC's Chad Sakac thinks that suppliers lacking a vertical stack strategy will face industry consolidation, as has happened with SDN and HCI.
HPE is also well aware of SCM technology because of its Machine research project and prominent server operation. Having bought SimpliVity last year, it is also active in HCI. It added Nimble to its storage array product estate in 2017 to provide an alternative to 3PAR and, arguably, the industry's best array management technology, Infosight, which is being extended to 3PAR.
HPE says HCI is a stopping point on the way to a fully composable infrastructure and it is well placed here with its Synergy product line. The idea is that the upper, application parts of a vertical stack run on hardware and system software components dynamically orchestrated and instantiated from a common resource pool.
Each vertical stack is not built from a separately defined and integrated set of components. We might see an echo here of a silo. In these terms HPE's notion is that you build and tear down virtual silos from its composable infrastructure. Dell EMC's would be is that you buy them – from Dell EMC or its channel.
The rising giant from China has a distinctive take on providing faster data access arrays based on hardware acceleration as well as NVMe over fabrics and storage-class memory.
If this comes to fruition then Huawei could have seriously faster storage arrays and a strong HCI play as well.
Our perception is that it has little data management software capability and might look to partner in that area, if it sees a need. The data management companies might see a partnering opportunity here and be proactive about it.
However, the company is more active in the public cloud area than any other storage player, apart from IBM, which has its own cloud services.
Huawei is setting up its own public cloud, and using regional telco partnerships to do so.
Big Blue grew its storage business in 2017 after several years of decline, helped by the take-up of flash array hardware by its base. It is factoring NVMe over fabrics into its POWER server systems, which also have an HCI play through a deal with Nutanix.
The FlashSystem 900 all-flash array was involved with the NVMe over Fabrics exercise, which bodes well. There is scope for IBM bundling its POWER server line with its middleware to create vertical stacks with backend linkages to its public cloud. That should appeal to that part of its customer base not fixated on x86 servers.
Like Huawei, IBM appears not to have responded much to the rise of the newer data management players. We speculate that 2018 might bring an acquisition here to enable IBM to get its mitts on an example of the technology; the software equivalent of a Diligent, XIV or Storwize purchase.
Newly confident NetApp enters 2018 in a stronger competitive position than it has been in for several years. It has charged into server SAN/HCI with its SolidFire-powered product and has a good handle on and pathway to NVMe and SCM technology.
The whole Data Fabric scheme positions it nicely for the public cloud and the Microsoft Azure ONTAP win in 2017 was a spectacular coup.
We wonder if the company might introduce a Data Fabric product in 2018 which orchestrates ONTAP, SolidFire, StorageGRID and E-Series arrays, and provides some kind of overall management function.
Nutanix has charged to the front of the HCI suppliers and is rapidly transforming itself into a vertical stack supplier. Dell EMC might expect to surpass Nutanix with its own HCI products – ScaleIO, VxRail and VxRack supplier though.
Chad Sakac, president of converged platforms and solutions at Dell EMC, says: "VxRail and vSAN not only have more customers, more revenue, but also growth rates that are multiple times faster than the number 2, number 3 players. Most recently, we've publicly stated the 149 per cent year-on-year growth rate for Dell EMC VxRail and 180 per cent year-on-year for VMware vSAN.
"It seems like it's a two-horse race (using anyone's data) – with Dell EMC and VMware-powered HCI in the lead, our partner Nutanix close, and everyone else in a different grouping."
Were Dell EMC to overtake Nutanix with its ScaleIO/VxRail/VxRack product set by the end of 2018 then Nutanix would still, we assume, be growing strongly (just not as much as Dell EMC) and leaving the other HCI suppliers – HPE, Cisco and, presumably, NetApp – eating its dust.
A note of caution: Nutanix is transforming into a software supplier, and thus absenting itself from direct insolvent in hardware technologies such as NVMe and SCM. It's going to rely on hardware partners to do that heavy lifting.
Pure Storage is a shared, all-flash, external array supplier. It has no HCI presence. But it does have an actively pursued path to NVMe over Fabrics and a new CEO. The FlashBlade development could well use QLC flash if and when that emerges in the shape of usable chips.
It has no direct involvement in data management software, but it does have data management partnerships, like one with Cohesity.
Pure is basically a two-product company with great ambition. We would not be at all startled to learn that it has an HCI research project.
Last year Hitachi Data Systems was sucked back into the overall Hitachi organisation and merged with Pentaho and another unit to become the IoT and analytics-focused Vantara.
It is far too early to tell if this sidestep out of being a pure-play storage supplier, geographically excluded from the Japanese market, will be successful.
Vantara storage product sales could accelerate because of a rising IoT tide or equally, slow down because storage is a small part of IoT edge systems and Vantara does not shift much storage into IoT core data centres.
We wait and we will see.
Infinidat is ramping through the on-premises high-end array space with its mix of memory caching, software and disk drive capacity storage.
It has no HCI or all-flash play and no data management software presence.
All this could change. If QLC flash drives with enterprise-class reliability appear, Infinidat might use them as a storage tier.
HPE might decide its top-end array strategy is flawed with both the XP (OEM'd from Hitachi) and 3PAR systems ill-suited for the future, and decide to look to buy better tech. Who else but Infinidat is there?
The next 12 months could be very interesting for InfiniDat.
Kaminario has an excellent and well-communicated path to NVMe drives and fabrics. It is a supplier of shared external storage flash arrays, not present in the HCI or server SAN area, and neither having a vertical stack offering, nor a public cloud play.
It is also facing strong competition in what is becoming a declining and consolidating market.
Other similarly situated storage startups, like Nimble and Tegile, have fled, as it were, into the arms of bigger suppliers. Tintri is struggling. Newer startups, such as Apeiron, E8, Excelero, WekaIO and others are nipping at Kaminario's heels, and it has to show that it can outgrow them and establish itself as a standalone player like Pure Storage and Nutanix.
Its NVMe tech looks distinctive and realistic, and is well-funded for further development. ®