This is the second post in a series looking at predictions for the storage industry in 2023. Find the first post here.
Storage Systems have evolved significantly since EMC introduced the first ICDA (Integrated Cached Disk Array) in 1990. The market today is one of relative stagnation with little advancement in products and features. Is this really a bad thing?
Background
Storage systems, appliances, or arrays (call them what you will) have evolved significantly since the Symmetrix was released in 1990. The first EMC arrays broke the mould of legacy storage by introducing low-cost SCSI disks with failure protection using RAID (initially just mirroring, then RAID-5 and RAID-6). DRAM caching was used to improve performance and optimise reading and writing to relatively slow drives.
One important aspect of early systems compared to today’s appliances was the degree of bespoke hardware and software in the solutions. In 1990 it wasn’t possible to build a reliable storage system using off-the-shelf components. Modern systems today take advantage of commodity components, including system memory, processors, and NAND flash. However, as we will discuss, custom designs still exist and have a place in the current market.
Standardisation
The industry has standardised on two designs, scale-up and scale-out. In scale-up architectures, performance is increased through the addition of faster controllers (compute) and/or more drives (capacity). Scale-out systems introduce extra controller nodes and generally add both storage and compute at the same time. Some implementations blur these two designs together, scaling horizontally and vertically.
In terms of protocols, storage systems have evolved from block-based connectivity to embrace file and object-based protocols. Appliances can support one or more (or even all) of these connectivity options at the same time.
Gains
What have we gained over the last 30 years?
Growth and Density – in parallel with the improvements in storage media, storage systems capacity has increased massively. It’s now easy (excluding the cost) to build petabyte systems, even as all-flash configurations. Scale-out technology allows almost infinite scaling, with geo-dispersed deployments that are efficient across multiple locations.
Features – The increase in processor performance has enabled storage systems vendors to add many new features in software (rather than hardware), including compression, de-duplication, and encryption. Features that previously required dedicated hardware can now all be implemented in software. Note: some vendors are making tactical use of hardware, for example, NetApp with Pensando cards for data reduction, but this approach isn’t widespread. The use of commodity hardware means that current upgrade cycles are now dictated by new processor architecture releases. These generally follow the Intel x86 family, such as the imminent Sapphire Rapids platform (although Arm solutions are starting to appear).
Ease of Use – storage systems from 20 years ago required considerable technical skills and understanding. Storage administrators were required to understand disk and RAID group layouts, spindle IOPS counts, the details of fibre channel, and have good debugging skills. As storage software has evolved, most of these skills have been replaced by simplified management tools and SaaS-based analytics. Storage has become very much an administrative role and much less technical.
Consolidation
In the past year, we’ve seen Pavilion Data Systems shut down, StorCentric enter Chapter 11,, yet Pure Storage and Infinidat have both reported good growth figures. Before the current period, Tintri had shut down and been acquired by DDN (which also acquired IntelliFlash from Western Digital and was previously known as Tegile Systems). Vexata was acquired by StorCentric in 2019. Throughout all of this time, Dell Technologies decided to revamp the Unity/VNX platform with the release of PowerStore and has been adding new features ever since. In parallel, the XtremIO solution, once pitched as the great replacement for EMC storage, is being quietly decommissioned.
Winners & Losers
In a relatively competitive and stable market, why are some vendors winners and others losers? We need to look outside of enterprise storage to see how the market has evolved. The public cloud has dominated growth over the last decade and opened up the concept of consumption on demand. This has replaced the forklift upgrade process of endless technology refreshes because the cloud vendors obfuscate this process into a background task.
Modern IT organisations want the same plug-and-play functionality offered in the public cloud. However, many legacy solutions aren’t built with this requirement in mind. Pure Storage solves the problem with interchangeable components and flexible purchasing models. Infinidat uses the cost advantage of HDDs to deploy fully populated hardware on day one. Neither Dell nor HPE products have been engineered for the service-based model of delivery, so we see longer commit times that make the APEX and GreenLake offerings look more like leases. However, the refresh process still involves the equivalent of a forklift upgrade.
The winners in the current market are those that have moved to flexible financial models and platform efficiency to deliver just-in-time and in-place upgrades. It’s much less now about features and functionality but rather about the experience and financial flexibility.
Predictions
With little in the way of technical innovation, what can we expect for the future of on-premises appliances?
Declining costs and increased reliability. This is pretty obvious but does need to be called out. Cloud storage costs have plateaued and aren’t reducing further (in fact, cloud service providers are implementing more tiers instead). In contrast, the on-premises storage market is a cut-throat business with many competing companies. NAND pricing continues to fall and become more cost-competitive with HDD systems, which will have an additional impact.
Service-based pricing. We will see an increased push on buying using service-based models and consumption pricing. This means architectures need to be capable of incremental capacity growth, have remote management/monitoring and be highly commoditised. Vendors will need to become more transparent with their prices. The legacy vendors (Dell, HPE) are in an all-out push to transform everything to a service model (even if they’re still disguised leasing). Other vendors are following the same approach. A lot of innovation will come in the form of fleet management to make storage as seamless and transparent as possible.
Transition to all-flash. We’ve talked about the all-flash data centre for years. As flash costs have reduced, so have hard drive $/GB prices, so for large volumes of inactive data, flash will never be competitive (or certainly not in the next decade). However, for active data, flash will eliminate the hard drive. I expect within five years, the hard disk will be an archive-only solution and will be competing with tape. Flash will also be used tactically for performance enhancements with technologies such as CXL.
Predictable upgrades. More than ever, platform upgrades are aligned to Intel processor family releases. We will see new solutions emerge as new product families are introduced into the market. For example, Sapphire Rapids will bring PCI Express 5.0 and DDR5 memory, offering vendors an opportunity to develop faster and products.
Feature stagnation. There’s not much left to add to shared storage solutions in terms of new features. Dell Technologies believes we’ll run applications on our storage, but I think this is a corner case. We have seen some new architectures, such as Pavilion Data, but these haven’t gained traction and (as we’ve discussed) disappeared from the market. With such thin margins for storage vendors, I expect to see fewer new features and little change in storage platforms in the next ten years. We will continue to see new architectures that exploit new media better, but there aren’t many improvements to make outside of that.
Opportunities for alternative architectures. Whilst we see Intel x86 dominating the supply of storage appliances, the Arm architecture offers another route to develop more efficient storage systems. In parallel, the SmartNIC technology offloads many storage functions into peripheral hardware, which could make Arm more viable. VAST Data has developed a hardware appliance using SmartNICs that builds a scale-out data platform. SoftIron has built solutions based on Arm. GRAID has created a new “distributed” RAID card, while Nebulon has design a distributed storage solution based on custom SmartNICs.
The end of enterprise arrays. The classic “monolithic” storage appliance of the 1990s is all but dead, save for the connection to mainframe. Midrange designs are good enough for most requirements, and I expect this design will be the dominant block-based storage platform going forward. The mission-critical platforms will have a place, but that requirement will continue to shrink.
Integration. Perhaps the most significant opportunity for shared storage is greater integration. If container-based frameworks like Kubernetes become as dynamic as containers themselves, then we will need another layer to manage persistence. There’s a role here for shared storage to offer persistence, resilience, and integrated data protection. We’ve also seen other solutions like NVIDIA GPU Direct, which bypasses the need to move data via the processor. I expect we will see greater integrations in this area, where storage appliances act more autonomously from the connected server. Some of this work may come from the evolution of SmartNICs, although the use of this technology as part of a storage array has not yet gained traction.
Rise of the SoHo devices. In a classic Innovator’s Dilemma scenario, I envisage that at the lower end of the market, new entrants will emerge, building solutions that started as home systems but can deliver to an ever-wider set of requirements. These vendors, including QNAP and Synology, are building systems “good enough” to meet the entry needs of SMEs and some medium-sized businesses. The lower-capacity end of the market will be increasingly challenged by these companies and by the public cloud.
The Architect’s View®
Two decades ago, centralised shared storage was the strategic approach for the enterprise data centre. The public cloud and the rise of software-defined storage have changed that view as we gain greater diversity in how applications are designed and deployed. Consolidation in the industry will continue, driven by these factors and in part by the broader issues of the supply chain.
Future success will be dictated by the ability to move as close as possible to a cloud-based operating model. For the remaining vendors, the market is one of attrition, which will see market share gradually move to the public cloud and the top performers.
Copyright (c) 2007-2023 Brookend Limited. No reproduction without permission in part or whole. Post #8ee1.