Storage area networks have been a stalwart of organizations’ IT infrastructure for the past three decades. Originally consisting of various interconnected disk arrays, tape libraries, hubs and/or switches, this series of networked storage devices and software stack provides pools of block-based storage to clients and servers with their own file systems.
A decade ago, VMware introduced a virtual SAN, which was a software-based distributed storage system. A vSAN was supposed to eliminate the need for expensive, overpowered storage controllers through software installed the existing servers and use commodity server-class storage media. By virtualizing the SAN environment in such a way, organizations could lower the cost of their storage architectures by leveraging the undeniable price advantages vSANs have over other storage architectures.
Since Broadcom’s acquisition of VMware, many organizations that had deployed vSANs have looked to other alternatives that can offer the enterprise-class performance, scalability and resiliency while fulfilling the original cost reduction promise. With growing uncertainty facing VMware users, plus it historical challenge of not living up to the original promise there must be a better way to build a vSAN.
If you were to design a new, modern virtualized storage area network, what critical components would you want to be incorporated? An integrated hypervisor? Better data efficiency? Supported dedicated arrays? Protection from drive failures? Snapshots that act like clones?
Let’s explore how a properly designed vSAN eliminates the need for expensive, overpowered storage controllers and the exorbitant 5X to 10X markup that storage vendors place on storage media.
Chief Marketing Officer, VergeIO.
vSAN with an integrated hypervisor
Instead of having separate software for each infrastructure layer, including hypervisors, management GUIs, and network functionality, integrating all of these layers into a single, efficient code base results in a measurable increase in VM density and performance for both compute and storage IO. Integrating the vSAN instead of making it a subservient VM, an integrated approach enables marked performance improvements over other vSAN technologies and even dedicated all-flash arrays.
With a tightly integrated vSAN, the hypervisor can run on almost any modern X86 server. Organizations can select their preferred server and media while keeping open the possibility of utilizing different hardware in the future. This flexibility enables the use of almost any type of media from almost any vendor.
vSAN with deduplication at the core
By integrating all infrastructure software into a single code base, the vSAN can economically use CPU and RAM resources to perform low-level deduplication operations to gain capacity for free effectively. Deduplicating hundreds of remote locations into a single location also minimizes replication times and capacity requirements to store all the data.
A vSAN with FC-SAN support
When a vSAN environment is deployed in conjunction with an FC-SAN, VM data can be centralized on the FC storage array, with each server node contributing storage. These virtual drives are consolidated into a unified storage pool accessible by VMs, enhancing data storage capabilities.
The scalability of storage capacity within this environment offers multiple avenues to explore. IT can augment the existing FC array. Or it can integrate server-class NVMe SSDs and HDDS to achieve a 10x cost reduction. In addition, implementing emerging storage technologies with this new approach to vSAN will be much quicker compared to updating legacy storage vendors’ systems.
A vSAN with intelligent data failure protection
With efficiency gains from deduplication technology, vSANs become more intelligent in applying drive failure protection. Because the vSAN is integrated into the same code as deduplication, multiple copies of each unique segment are stored for ultimate protection. With data stored on different drives and different nodes, VMs continue to have access to data if a drive or even an entire node fails. When a VM needs to access data on a failed drive, the redundant copy of that segment can be accessed with no impact on application performance.
A vSAN with snapshots that act like clones
Unlike vSAN snapshot technologies offered by vendors like VMware, integrating snapshot functionality into the same code as deduplication capabilities can provide a clone-like capability.
This approach to snapshot technology essentially makes a full copy of the source, which can be a virtual machine, a virtual data center, or the entire instance. A single metadata instance is created to manage the unique status of each snapshot taken. This approach is relatively flat and far less complex than legacy snapshot methods.
Customers could retain an unlimited number of snapshots for an indefinite period without impacting performance. Because the snapshots are devised to be read-only, they are also protected against cyber threats like ransomware attacks.
Conclusion
Addressing performance and cost issues, ensuring data availability against failure, and elevating data resiliency to protect against loss or theft can help build a better vSAN than VMware offers.
Integrating the hypervisor with storage software represents a significant leap forward. Consolidating functions into a single efficient code base and deploying cutting-edge features, such as built-in deduplication and intelligent data failure protection, should become standard in the industry. These advancements, combined with the ability to utilize commodity hardware, mean that next-generation vSANs are capable of delivering the unparalleled performance, scalability, and cost savings required today by organizations of all sizes.
We feature the best cloud storage.
This article was produced as part of TechRadarPro’s Expert Insights channel where we feature the best and brightest minds in the technology industry today. The views expressed here are those of the author and are not necessarily those of TechRadarPro or Future plc. If you are interested in contributing find out more here:
