The market for hyper-convergence seems to know no bounds as more and more vendors move into this segment of IT. Last week EMC released their rebranding of the failed VSPEX BLUE platform as VxRAIL with EMC supplied hardware and VMware software. Looking across the market we have the two leaders (Nutanix and SimpliVity) offering hardware appliances that cater for all forms of hypervisor – including in Nutanix’ case their homegrown Acropolis product. Then there are the SDS vendors that have pivoted towards hyper-convergence. I did a little bit of research on the appliance part of the market and there’s a spread of hardware & software-based appliance solutions now available:
- Cisco – SimpliVity, StorMagic, Atlantis Computing, Maxta
- Dell – Nutanix XC, Atlantis Computing, Maxta, StarWind
- HDS – Hitachi Hyper-Scale Out Platform
- HPE – HPE StoreVirtual, Atlantis Computing, Maxta
- Lenovo – StorMagic, Atlantis Computing, Maxta, SimpliVity
- Quanta – Maxta (also EMC?)
- SuperMicro – Scale Computing, Atlantis Computing, Maxta
So what about NetApp? So far the company has developed reference architectures based around their FlexPod brand. Its been in the market for a while, but really only qualifies as being converged rather than hyper-converged. Looking deeper at the software, Data ONTAP is probably one of the most “software defined” platforms around. There has been a fully functional version of the ONTAP simulator for years and there’s the software-based ONTAP Edge for virtual environments. Clustered ONTAP scales to eight nodes with SAN/block protocols and 24 nodes with NAS protocols, however the architecture doesn’t really fit the hyper-converged model as nodes have to be deployed in pairs. There’s no “loosely coupled” or “shared nothing” scale-out design.
So how does NetApp get into hyper-convergence? I see the company as having two main options – they could purchase a startup, or use SolidFire. This is where things could get interesting. SolidFire Element OS is fully scale-out and has even previously been shipped as a software-only solution called Element X, albeit with specific hardware requirements. So, we could imagine a hyper-converged solution from NetApp that uses SolidFire and multiple hypervisor types – in fact any that support iSCSI LUNs. I don’t think this kind of solution would be that hard to build. SolidFire’s initial market was service providers, customers running scale-out storage and virtual servers. Most of the background scripting and integration that would be needed has probably already been done for customers as the SolidFire platform can be driven entirely by API.
Thinking about how hyper-converged could be implemented, I don’t have enough information to understand how Element OS currently runs on each node; whether it is a single Linux/Unix OS or has already been containerised. There may of course need to be engineering work to put both the hypervisor and O/S onto the same host – this might prove the tricky part, however I can’t imagine it being that unsurmountable.
The second alternative is to buy a start-up. Atlantis, Maxta and Springpath are the most obvious choices. This would mean NetApp spending more cash, but it wouldn’t be that expensive to pick up some good IP.
The Architect’s View
NetApp’s recent sales figures yet again paint a picture of a shrinking storage market, certainly in the legacy/traditional platform area. Hyper-convergence is taking a slice of the revenue (as has all-flash) and vendors need a hyper-converged story to compete. With the acquisition of SolidFire, NetApp has some great storage IP that could be leveraged in other ways. It would be good to see the company moving on from the permanent focus on Data ONTAP; that process of course has already started. Can they keep the momentum going?
- NetApp Integrated EVO: RAIL – A Triumph of Marketing Over Common Sense
- NetApp Drops 4%: FYQ3 Rev Misses, Outlook Misses; To Cut 12% of Staff (Barron’s Tech Trader Daily blog, retrieved 22 February 2016)
Comments are always welcome; please read our Comments Policy first. If you have any related links of interest, please feel free to add them as a comment for consideration.
Copyright (c) 2009-2016 – Post #704F – Chris M Evans, first published on https://blog.architecting.it, do not reproduce without permission.