Delivers High-End Enterprise Storage Features at Midrange Price Point

IDC OPINION

Digital transformation (DX) is a hot topic in most enterprises today. DX drives a host of new business requirements that challenge legacy infrastructure, and information technology (IT) organizations are upgrading to new server and storage platforms at a rapid rate to meet those requirements. Although the enterprise storage market overall continues to grow, revenue for storage systems at midrange price points ($25,000–249,999) is growing at the fastest rate: in 2019, this market grew 10.6% to make up 60.3% of total enterprise storage revenue. Part of the reason the midrange storage segment leads the entry and high-end enterprise storage segments is that these systems are increasingly incorporating the performance, availability, scalability, and functionality of higher-end systems. For IT organizations looking to streamline costs as well as storage infrastructure, systems at midrange price points meet a broader set of requirements than does either of the other two storage classes.

For those IT organizations undergoing infrastructure modernization as part of their DX journey, IDC research indicates that access to new technologies like NVMe, scale-out designs, and artificial intelligence (AI)/machine learning (ML)–driven management is high on the wish list. These technologies are needed to meet increasing performance, availability, scalability, ease of use, and agility requirements in digitally transforming enterprises. For IT organizations looking to do more with less, infrastructure adaptability is key, and in storage, this has heightened interest in platforms that support unified storage, bare metal or virtual deployment, and different deployment modes and can non-disruptively scale both up and out. 91.1% of the enterprises traversing their DX journey deem infrastructure modernization a key determinant of success, upleveling the importance of modernizing IT infrastructure in a manner that opens up access to these and other new technologies on the horizon that will be needed by the digitized enterprise.

In May 2020, Dell EMC introduced the new PowerStore family of clustered storage systems. These systems offer high-end performance, availability, scalability, and functionality starting at midrange price points and incorporate sought-after new technologies like native NVMe, scale-out architecture, software-defined infrastructure, AI/ML-driven system self-management, and a microservices-based storage operating system (OS) design. PowerStore is an interesting and very capable new unified storage system that is scalable to over 4M IOPS and almost 11PB of storage capacity can be deployed in either a disaggregated or hypervisor mode. This system deserves to be on the short list for those enterprise customers looking for the right storage infrastructure for their own DX journey.

IN THIS WHITE PAPER

As enterprises continue to undergo DX, they are looking to improve storage performance, scalability, manageability, and agility as well as IT infrastructure efficiency. This has implications for how IT requirements are met in core, edge, and cloud-based environments and is ushering in new system architectures as well as the use of new technologies like NVMe, scale-out designs, and AI/ML. In this white paper, IDC discusses the evolving nature of IT infrastructure requirements for enterprises that are currently in the midst of their DX.

SITUATION OVERVIEW

DX, which is the move to much more data-centric business models, has become the new imperative for those enterprises looking to leverage the vast amounts of data available to help inform better business decisions. DX impacts all functional organizations within an enterprise, but IT is particularly impacted. CIOs are faced with the need to service legacy workloads to maintain business continuity while deploying next-generation applications (NGAs) that leverage newer technologies like mobile computing, social media, big data and analytics, and cloud to turn available data into strategic business assets. IT infrastructure is no longer just a core datacenter decision as CIOs consider optimal workload placement taking into account edge (or distributed), core datacenter, and public cloud–based locations.

As IT organizations work their way through these decisions, they are modernizing existing IT infrastructure at a rapid rate. According to primary research conducted by IDC in 2020, among those organizations undergoing DX, 68.2% of them are refreshing their storage infrastructure. The most popular general strategy for infrastructure modernization is to move workloads to more softwaredefined architectures that feature characteristics to meet increasingly stringent performance, availability, scalability, manageability, and agility requirements. New technologies that are important to meet these requirements include NVMe, scale-out architectures, and the use of AI/ML to help streamline administrative tasks and optimize system operation. These new technologies are discussed in detail in the sections that follow.

NVMe

While low latency and high throughput have been critical storage capabilities needed for certain legacy workloads like online transaction processing, many of the NGAs that enterprises are deploying have this high-performance storage requirement as well. These applications often have a much more realtime orientation and are, in many cases, handling high data growth environments that can easily span into the petabyte range over time. For an increasing number of these newer workloads, the SCSI protocol that has been a mainstay of enterprise storage is being challenged to meet performance and efficiency requirements, particularly for all solid state systems that are being used for the most demanding applications.

A new storage protocol standard called NVMe was first introduced in 2011 as the heir apparent to SCSI for higher-performance requirements and has become a widely deployed technology for highend enterprise storage over the past several years. NVMe was specifically developed for solid state media and makes much better use of solid state storage resources than SCSI can to drive higher performance, increased reliability and endurance, improved efficiencies, and lower cost. NVMe supports much lower latency than SCSI and significantly greater parallelism (a feature particularly important with today’s multicore CPUs).

NVMe storage devices first began to be deployed in servers as internal storage, but capacity utilization and scalability limitations, along with a desire to leverage enterprise-class data services (inline data reduction, thin provisioning, RAID, snapshots, encryption, replication, etc.), drove the need for a switched fabric that would allow high-performance NVMe storage to be shared. NVMe over Fabrics (NVMe-oF) fulfills that requirement, thereby enabling the full performance of enterprise-class, shared NVMe-based arrays to be directly applied to application performance. Enterprises will clearly be making the transition from SCSI to NVMe for their primary storage workloads over the next several years, and that transition will also drive the penetration of NVMe-oF (although at a lesser rate). In 2019, NVMe-based all-flash arrays (NAFAs) were already a $2 billion market, and IDC expects that by 2021, NVMe-based arrays will generate over 50% of all primary external storage revenue. NVMe-oF deployments will lag behind those of NVMe-based arrays, but most enterprises buying shared storage solutions for primary workloads will need to know that they have a simple upgrade path to NVMe-oF when they need it.

NVMe technology, both in the storage array and the storage network, will support a streamlining of IT infrastructure. Smaller systems with fewer storage devices and network ports will be able to significantly outperform SCSI-based arrays of equal size. The ability of NVMe to support increased IT infrastructure density is expected to simplify systems, reduce floor space requirements, and possibly reduce energy consumption (depending on how the technology is deployed). The use of NVMe in midrange arrays, which to date has been less widespread than with high-end enterprise arrays, will significantly up the performance capabilities of these systems, allowing them to in some cases surpass the performance capabilities of the last generation’s high-end arrays at lower cost.

Scale-Out Architectures

Most successful IT organizations undergoing DX are experiencing very high data growth. To easily accommodate this growth, storage administrators need to be able to easily expand their storage infrastructure in a nondisruptive manner. A successful start-up initially needing tens of terabytes could easily grow into needing petabytes of storage within just a few short years — certainly within the common life cycle of three to five years for enterprise storage. One of the factors driving the industry’s significant interest in and purchase of software-defined storage systems is the ability of many of these systems to scale non-disruptively by the simple addition of more nodes into a cluster. Federated clustering allows this to occur, preserving a unified management interface even as a cluster expands to more nodes to meet higher-performance and/or higher-capacity requirements.

Scale-out designs offer attractive advantages for high-growth environments:

▪ They provide low-cost configurations that are easy to purchase and deploy for edge environments. Scaling these systems is as easy as just adding another node, and with sophisticated storage management skills in short supply in these types of distributed environments, that simplicity is very attractive.

▪ The range of scalability of these systems is very broad, and they allow customers to scale both performance and capacity much more so than traditional scale-up platforms that can just scale storage capacity, providing a much more balanced growth path.

▪ The distributed nature of these platforms helps avoid “noisy neighbor” performance problems because workloads tend to have an affinity to run on a given node but can be easily moved to another node (or a newly added node) in the cluster for more efficient workload balancing.

For these and other reasons, IDC has noted the rise of scale-out platform revenue in enterprise storage, and while traditional external storage array designs still generate more revenue than scale-out platforms, there is no doubt that the industry is moving toward scale-out architectures (just as they are also moving toward software-defined designs).

Intelligent Management

Software-defined platforms provide significant configuration flexibility, thus effectively turning server, storage, and networking resources into programmable infrastructure. This feature alone can be attractive to administrators who have worked with more static IT infrastructure in the past, but when that flexibility can be combined with real-time AI/ML capabilities, enterprises take a giant step toward autonomous operations. This moves IT operations from riskier, less productive manual administration toward more efficient policy-based administration, freeing IT management resources up to perform more strategic tasks. It also meshes well with the trend IDC has noted for storage management tasks to migrate more toward IT generalists (e.g., virtual administrators, Windows/Linux systems managers) and away from more costly dedicated storage management groups. Policy-based management allows administrators to more closely tie systems performance to specific business goals, and when that management is informed by AI/ML, it becomes faster and less risky than manual storage administration.

Many enterprise storage providers offer what IDC refers to as a “cloud-based predictive analytics platform,” which has effectively replaced their older “remote monitoring” systems. The three key features that differentiate these intelligent platforms from the legacy remote monitoring approach are the scope of monitoring, how data is stored and shared, and the use of AI/ML to drive autonomous operations. These new systems collect significantly more data than before, not only capturing more indepth metrics from more components within a given storage system but also extending that data capture to other IT infrastructure components like servers, networking, and applications. These systems drive real value for end users in optimizing their installed systems to meet defined objectives in performance, availability, and other areas. Vendors that provide these systems for their enterprise storage platforms are increasingly using them to differentiate themselves from the competition.

To read full download the whitepaper:
Delivers High-End Enterprise Storage Features at Midrange Price Point

SEND ME WHITEPAPER

Previous articleOptimizing Performance with Frequent Server Replacements for Enterprises
Next articleThe Business Value of Storage Solutions from Dell Technologies