Nfina Storage Architecture

Nfina Technologies develops, manufactures, services, supports, and monitors storage and virtualization infrastructure utilizing Nfina hardware and software to provide highly reliable and redundant systems to businesses across the globe. This document aims to educate the reader on the various architectures which Nfina storage and virtualization systems can be configured along with use cases for each.
Nfina-Store
Nfina-Store is the operating system utilized in Nfina systems which can provide block and file level storage to a wide range of environments. Nfina-Store utilizes a Linux kernel and ZFS to provide pool-based storage which can optionally be made highly available by forming a storage cluster to prevent downtime. Nfina-Store can be deployed directly to physical Nfina hardware or run under ESXi in a hyperconverged (HCI) architecture. Nfina-Store provides client machines file or block based storage utilizing iSCSI, FC, NFS, or SMB while utilizing RAM and NVMe for data acceleration of the underlying storage pool disks.

Figure 1 – Nfina-Store Data Acceleration
Single Node
Single node Nfina-Store systems can be configured as a physical storage server or as a virtual storage appliance under ESXi. The following sections describe both scenarios
NAS – Physical Storage Server
Deploying Nfina-Store as a single node NAS appliance requires installing the Nfina-Store operating system onto physical hardware with the goal of providing network-based block or file storage to client machines. This is a single controller architecture which means that maintenance or hardware failure will result in downtime. This architecture is typically deployed as a file server or in virtualized environments which can be completely shut down for periods of time to perform maintenance.
Hyperconverged Infrastructure (HCI) – Virtual Storage Appliance
In an Nfina HCI system, Nfina-Store runs as a storage virtual machine and utilizes PCIe passthrough to unlock performance bottlenecks present in virtualized storage solutions. Deploying Nfina-Store as a single node HCI appliance requires installing the Nfina-Store operating system as a virtual machine on the ESXi hypervisor with the goal of providing network-based block or file storage to ESXi. This architecture adds virtualization capabilities to the system within a single chassis and the ability to create immutable array level snapshots as well as the ability to replicate these snapshots to a secondary location for DR.
Running this architecture allows Nfina-Store to serve as both the underlying storage for the virtual machines running under ESXi, but also simultaneously provide SMB, NFS, or iSCSI storage to any client device supporting these storage protocols.

Figure 2 – Single Node Nfina-Store Solution
2-Node HCI
Deploying Nfina-Store in a dual node HCI system requires two Nfina servers configured with identical hardware which will run ESXi with the Nfina-Store operating system to control the storage between the two nodes. The Nfina-Store software will combine the storage between the two nodes utilizing either 2-way or 4-way mirrored storage pools with the disks present in each mirror residing in opposite chassis. This means that if one of the machines fails, half of the disks remain online and the storage pool remains available to the remaining ESXi node. This allows virtual machines previously running on the failed node to be migrated to the remaining node in the cluster.

Figure 3 – 2 Node HCI Solution
This architecture provides a fully redundant HA solution with just two physical chassis which can provide cost savings. The two-node system can run completely switchless for the back-end network fabric which also allows for savings on expensive network switches. This system can tolerate a single host failure and perform system maintenance without any downtime.
3-Node HCI
Deploying Nfina-Store in a 3 node HCI architecture provides the most benefit when more compute, memory, or IO resources are required than can comfortably fit within the two-node solution. The 3 node HCI solution can optionally utilize the third compute node as a backup storage repository. The two primary compute nodes operate in the exact manner as the 2-Node HCI solution, while the third node receives array level snapshot replication for an offline copy of the data. It adds the benefit of also doubling as a compute node which can run virtual machines and provides the ability for long term backup storage restoration options.

Figure 4 – 3 Node HCI Solution
Converged
In the converged architecture, the storage and virtualization are separated into their own physical chassis instead of combining the storage, networking, and compute of each node into a single chassis as described in the HCI architecture. This architecture adds management flexibility for extending compute and storage resources independently as well as a higher performance than Nfina’s HCI systems. Nfina’s converged systems support 2 or more ESXi hosts in a fully redundant HA architecture. While there is no physical limitation other than the number of switchports available for compute nodes to connect into the storage fabric, Nfina practically recommends that no more than 12 compute nodes be connected to a single converged storage array before adding an additional storage array with two more storage controllers. Nfina recommends that any system utilizing 4 or more compute nodes deploy a converged architecture.
Failure and maintenance of compute and storage nodes in this architecture occur independently which provides more resiliency to the system. A single storage controller could be lost without downtime, and all compute nodes could fail up to the required number of compute nodes needed to run your workloads.

Figure 5 – Nfina SAN with 4x Hypervisors – Converged
Conclusion
Storage architecture is the foundation that sets the priorities of an organization’s data management, performance, content protection, and value of their data strategy. Storage management architecture is one of the more important areas that defines both the ease of operations and the success of the operation from a delivery perspective. Additionally, unprotected or improperly structured architectures can end in misfortune. It’s not unusual to find one or both elements in some of today’s small and mid-size enterprise IT environments.
Storage technologies have advanced over the decades. Unfortunately, there are many storage architectures in place that evolved by combining different solutions that may have been designed by numerous providers and consultants. With today’s storage complexities, home-brewed solutions for mid-size enterprise organizations are no longer realistic. Choosing an appropriate foundation for your storage architecture becomes key to its performance and the continued success of your operations.
There are considerations for storage architectures that involve cost, workload requirements, and providing the flexibility to easily adapt to your changing storage needs. Sometimes legacy and existing components that still have a financial life expectation must be considered. Software applications, workflows and data structures all become key components in determining the full storage solution. It is essential for small and mid-size enterprise organizations to work with a storage vendor such as Nfina or your Nfina MSP Partner to develop your storage architecture.

At Nfina, our Eco-Friendly Solutions make it easy for our customers to achieve a lower carbon footprint and play a positive role in bringing about a sustainable future. We design technologies and products to help people understand their impact and actions better.
Nfina’s Hybrid Cloud and Hyperconverged solutions provide energy efficiency by using high-density, lower-power VMs enabling our customers to scale their digital transformations sustainably by optimizing space, reducing power consumption, and lowering cooling and maintenance costs. Nfina is taking a leadership role in doing what it takes to tackle climate change.
Nfina has been carbon neutral for our operations since opening in 2012.