I have to provide an answer on this question ever so often to customers gauging their pace with storage market.
Looking from a business perspective [Companies are at risk of missing market opportunities and or failing to adapt to market trends].
As I sense customers readiness to adapt quickly to changes as this is a key to their success, this is becoming a legitimate question for business.
In an attempt to broadly answer to the above question. I have rather decided to submit it to experts and professionals viewpoints; to elaborate further on the business benefits SMB and Enterprise businesses are gaining from Hyper-converged.
I have long been a champion of the CI and hyperCI industry attention and solutions. I have followed and watched Nutanix and Simplivity change the game against the major vendors in this space. However, as I do more research of what's out on the market place, I get confused as to what separates a CI or HCI solution from an appliance model.
What I like about the CI/HCI solution is that it's essentially a data center frame in a small footprint (i.e. compute, network, & storage). However, I feel I experience "node" sprawl even more now with having to do n+1 (or n+2 in some cases) in order to support production needs. Yes, there are tools out there to manage this. But, if I am now scaling ALL components LINEARLY, I theoretically scale my management issues 3x for every CI/HCI node I deploy.
So, this brings me back to the appliance question in my head: Why not identify the more rapidly needed items in the node (mostly compute and network functions, in my case) and centralize the SAN (NAS or FC, whichever fits your need)? Storage and it's ability to connect/deliver the performance via NAS (or even iSCSI) is definitely a great way to meet performance demands now.
The appliance approach seems to scale compute & network more smoothly of what my applications need vs. having isolated storage islands where I probably won't ever consume it all. I am still researching options, but have noticed (queue "boos" and "hiss" soundtracks here) that Oracle has an approach with their Private Cloud Appliance (PCA) that does this for me. I already like the Oracle database appliance, but it's only bare metal and no easy virtualization options. PCA seems like a rack version of a blade server with data center fabric built-in. It seems a good option to creating scaled out solutions to meet the data center needs.
Hyperconverged systems is at early stage capturing the mindshare with midsize businesses or enterprise remote or branch office running their own application and have little to no IT. They still have to design, deploy and manage 10 to 10000 servers that require resources, time and expertise that they don’t have. They still need fast access to business data and lower cost infrastructure that it is easy to stand up and manage. Some of the workloads and use cases like ROBO, VDI ,software defined storage are gaining traction with the early adopters.
The Hyperconverged market is already a crowded market with big IT vendors increasing their efforts in this market: HPE’s HC 380, EMC/VCE offerings, Cisco Hyperflex (w/ Springpath) in addition to startups like Nutanix and Simplivity.
HPE has an interesting blog on traditional vs converged infrastructure since other factors must be considered when deciding on a Hyperconverged solution.
community.hpe.com
For us, our Nutanix solution fell into the Financial Discipline category. We were replacing a few servers and our storage solution when we came across the Nutanix appliance. The cost to replace those servers and storage with similar components was greater than the cost of our Nutanix. Compared to a "traditional" infrastructure, consisting of servers, storage, and a storage network, the Nutanix combines those three components into a single appliance. The Nutanix offered us the best scalability, performance, and cost-savings over other solutions. For example, we are a R&D shop so our IT infrastructure footprint is much smaller than the company is a whole. We procured a single Nutanix 3060 with four nodes and, due to the self-management features, I am able to focus significantly more of my time to other IT and DevOps functions. More specifically, I no longer need to manage tiered storage, storage fabric, etc. so the time I am recouping is spent on more impactful projects. In conclusion, the total cost of ownership and opportunity costs were the greatest factors/benefits of the Nutanix solution.
Some great feedback from the other users here, I agree with all of it. Even though there are two clear frontrunners in the category, Nutanix and SimpliVity, there are some others to consider that are more "software defined". There's Cisco's offering recently released in cooperation with Springpath (called hyperflex I think) and there's also VMware's wich the most recent version of vSAN at least puts it in competition with all the others. Whether its Cisco's offering or VMware's, it's all about consolidating physical resources, sharing them across a cluster and automating management through more intelligent software.
Is it storage? Yes. Is it compute? Yes. But you can land on any of the solutions (Nutanix, SimpliVity, Hyperflex, VMware) through a number of conversations... could be consolidation, private/hybrid cloud, ROBO's, DR, optimization, performance enhancements, VDI, etc etc.
In the SMB, hyperconverged could be the answer to everything. A few nodes, or clusters, can be the backbone of the business. In the enterprise, it could be particular departments, workloads, or in general an area of the business that doesn't require overly complex and robust systems, or a massive rip/replace project.
Both Simplivity and Nutanix offer comprehensive hyper-converged solutions that challenge traditional stack implementations. Where rack space, power, and heat management are problematic - an all in one 4U appliance that is simple and quick to cover becomes an attractive option.
I have witnessed HC solutions being employed across retail, education, motorsport, defence and government sectors. In terms of application within the business, while vendors support HC for replacing existing virtual environments housed on traditional IT, arguably a best use of this technology is VDI for both GPU and non GPU use - given the main vendors having close technology relationships with the likes of NVIDIA.
If you don't have the internal skills to install and configure discreet compute, SAN, backup services, switches, load balancers, WAN optimisers and anything else that you might find in a traditional rack configuration - consider hyper-converged. With a sub one hour setup - to the point you would have a storage connected hypervisor - HC is worth a punt.
Great thanks for those who have already shared their experience and knowledge of HCI and moving forward, the balanced view given by Sami Ventriglia on SMB and Enterprise from this post is much appreciated as he portrayed well the benefits and possible limitations or improvements to be addressed.
Having say that, this should not restrict any beneficial view to enrich the question
cloudacademy.com
I have received this in response to my post and the first lines are inviting to more reading