Just about everyone is ready for a new beginning in 2022. The past two years have been difficult to say the least; large enterprises have been rocked to the core, and the fundamentals of business have shifted.
The pandemic has forced change in every facet of business. It has pushed digital transformation to the forefront and changed the way enterprises operate, with many employees now working remotely. Yet change always brings opportunity. Volatility allows organizations to rethink long-held assumptions. In the IT realm, it’s a chance to reexamine strategies and look at things anew, including basic computing architectures.
With 2022 around the corner, here are three data storage resolutions that can help IT significantly improve efficiency, support innovation, and ultimately make substantial advances in ROI.
Resolution #1: Leave the Islands Forever
Most large enterprises have lots of islands of data storage: block storage for databases, file storage for sharing and collaboration, and object storage as a cloud gateway and for archival purposes. Managing all these islands can be a nightmare due to different vendors, different storage products, different management tools and different methods of operation.
Consolidating storage under a single paradigm can resolve management headaches while also improving performance. Costs drop as well. Software represents the majority of enterprise IT spend, and islands of storage require multiple software licenses.
Consolidating into a single, highly scalable storage system drastically reduces licensing expense. Moreover, because islands are often not filled to capacity, a single system allows organizations to better manage utilization. Over the long haul, the transition to a consolidated platform is worth the effort.
Resolution #2: Don’t Put All Your Data in the Cloud
Cloud has been transformative for modern business—but there’s a catch. In most cases financial and performance benefits occur early in the shift. Over time, storage creep can set in, not only impacting IT cost but also significantly increasing cost of revenue (COR) and cost of goods sold (COGS).
An Andreessen-Horowitz report states that for 50 of the top public software companies using cloud infrastructure, an estimated $100 billion in market value is lost simply due to the impact of cloud on margins. For the broader universe of public companies, the total impact is estimated at more than $500 billion.
Reversing decisions about the cloud isn’t easy, but storage is a reasonable place to start. Large data stores, and those for large-scale enterprise applications, can be moved to an on-premises object store without major investment. High-performance flash storage supports seamless linear scaling that rivals or beats cloud alternatives and when necessary, it can tie to the cloud for affordable, long-term archival.
In short, large organizations need on-premises storage. It provides predictable cost at a small fraction of cloud storage and more importantly, eliminates the creep that can set in from an all-cloud strategy.
Resolution #3: Get the HPC You Need—By Implementing HPIO
High Performance Computing (HPC) isn’t just for research institutions anymore. Large enterprises collect huge amounts of data from myriad sources—IoT, sensors, logs, financial, production, marketing, you name it. Somehow, all that data must be transformed into insight at the speed of modern business.
The biggest challenge lies between the creation of that data and the means of interpreting it. There are two parts to this challenge—latency and concurrency. Low latency is critical because it improves productivity; the lower the latency, the more work can be done. Concurrency, one of the benefits of low latency, is important because it allows more uses to access data at a time, thereby accelerating time-to-insight.
Michael Kagan, chief technical officer at NVIDIA, put it best when he said, “High performance computing requires high performance IO.” Modern computing systems leverage the parallel processing power of GPUs to drive faster results—but to deliver on this potential, data must be fed at rates that allow all available GPU costs to be fully utilized. In storage terms, IT must create a storage environment that writes as fast as it reads.
Some enterprises try to achieve this result by scaling out multiple storage systems on legacy fibre channel networks. Doing so creates a large, unwieldy footprint that doesn’t solve the latency problem. A better solution is Ethernet-based storage using Remote Direct Memory Access (RDMA). When combined with NVMe (Non-Volatile Memory express), a flash memory specification, IT can leverage the full benefits of parallelism, creating the high-performance computing environment that satisfies today’s competitive requirements and is ready for the demands of future workloads.
2022 may hold more hope for the future than any year in recent memory. Modern business, for its part, is reinventing itself in multiple ways to address this changing world. IT can take a fresh look at the way it operates as well. With a smart roadmap in place, it can help the enterprise finally realize that promise of a stronger, more productive and more successful new year.