In today’s digital transformation landscape, several trends and common business challenges are fueling the increased need for real-time processing and smarter memory solutions. The key macro trends to recognize include:

• The explosive growth of data
• The need to process data in real time
• Innovation driven by software in the cloud era, where hardware must often catch up
• A common desire to get more out the data by using Artificial Intelligence (AI), Machine Learning (ML), and Deep Learning (DL)

Common business challenges also exist when adopting a digital transformation journey, such as:

• Scaling infrastructure to address data growth in all domains: Compute, memory, storage, and network
• Processing data faster, requiring more in-memory computing
• Limiting the high expense of memory
• Managing data center complexity, which leads to the need for consistent Day-0 and Day-2 operations and monitoring across the infrastructure to help reduce operating expenses (OpEX)

These trends and challenges combine with unprecedented growth in of real-time data. According to International Data Corporation (IDC) The Digitization of the World, From Edge to Core report, by 2024, 25% of global data will be real time.

Virtualization and hybrid cloud are traditional contributors, but newer demands for real-time processing are stemming from AI and business intelligence; real-time analytics, ML, and massive datasets; and predictive analytics, transactional processing, and edge processing. With this explosive data growth, by 2025, the global data sphere is expected to reach 175 zettabytes—up from 37 zettabytes in 2018, as the IDC reports.

Data-intensive workloads need hot data in memory, but Dynamic Random Access Memory (DRAM) is expensive and has limited capacity. Further, enterprise data is a key asset that must be protected, but security and encryption can compromise performance. The following figure shows how DRAM price per GB increases significantly as memory density reaches its highest values.

Figure 1.   DRAM price per GB

Note: This figure source is VMware: VMworld 2021 session VI2342: The Big Memory Transformation, available with VMworld 2021 login. Prices that are shown in this figure are provided as an example and subject to change.

PMem is a memory technology where nonvolatile memory that is built on a standard 288-pin DDR4 DIMM form factor is installed on the memory bus (a CPU memory slot). This design allows PMem to exist alongside volatile memory such as DRAM and work in conjunction to achieve higher overall system memory capacity with better performance.  PMem can maintain data even after a system power outage. It is used by critical business applications that are sensitive to downtime but require high performance, while lowering the TCO.

PMem greatly reduces storage latency in ESXi based systems. VMs that are configured with PMem and applications running inside those VMs benefit from increased speed with lower costs when compared to a purely DRAM-based ESXi host.