NVMe, 5G and Content Delivery Networks
As data sets grow and data-intensive applications such as Big Data analytics, artificial intelligence (AI), machine learning (ML), genomics, and IoT gain in use, the costs and time needed for data movement is becoming critically challenged. Given the terabytes, sometimes even petabytes, of data being generated each day, data movement becomes a huge issue – especially when analytics are needed in real time. The issue is that moving massive amounts of data from storage to host CPU memory to process a query is costly in terms of power consumption and time. With the impact of data movement being felt in nearly all compute applications, change is needed in 2020 to handle this growing strain.
Below are my predictions on how organizations will store and process data next year, particularly enterprises relying on hyperscale, edge and CDN environments.
NVMe is NOT Enough – Move Less and Analyze More at the Edge – As we connect more Edge and IoT devices like video surveillance cameras and autonomous cars or ‘anything’ that can churn out data 24/7 – one of the major issues that many organizations experience is bottlenecks. The challenge: out of these enormous data sets created each day, organizations are often only trying to extract only approximately 10% so, how can that be done in a timely manner? The trick is to pinpoint the value of data in real time. NVMe (Non-Volatile Memory Express) has provided a measure of relief and proven to remove existing storage protocol bottlenecks for platforms churning out terabytes and petabytes of data on a regular basis.
- But, is that enough? Even though NVMe is substantially faster, it is not fast enough by itself when petabytes of data are required to be analyzed and processed in real time.
- UPSHOT: This is where Computational Storage comes in and solves the problem of data management and movement. Computational Storage, especially the way we marry the use of NVMe SSDs and compute power, adds analytical power and speed so that results can be accomplished right away and where the data is generated.
Computational Storage Enables Better 5G Connectivity: In 2020, more edge related devices will be needed to process massive data workloads. The advent of 5G is no different.
- Here is why: We all know that 5G increases the amount of bandwidth and speed of communication but along with 5G comes the need to develop a more complex infrastructures a to support seamless connectivity.
- UPSHOT: As more cell towers are built to support 5G, there also needs to be more complex infrastructure at each bay station that can manage the data “in and out of the box” so that user data is optimally utilized. Computational Storage with its small form factors and added compute power can pack an analytical uppercut punch in the limited size and power enabled edge-datacenters that live at each of these new cell tower platforms. By providing additional compute to the confined resources that exist is paramount to successful growth of this space. Instead of requiring even more hardware and power to the server, the advent of high capacity Computational Storage provides the needed offload to the system to allow for great deployments.
Computational Storage Can Simplify the Traffic Flow of Content Delivery Networks: Streaming services have continued to dominate headlines this year, with the recent launches of Apple TV+, Disney Plus and NBC Peacock, combined with Netflix, Hulu and Amazon Prime’s increasing investments. This poses a major hurdle for the content delivery networks (CDNs) – and where Computational Storage can be a major asset.
- Customers will have no patience for buffering or interruptions in video streams.Generally speaking, video requires lots of expensive data movement, which makes it more costly and difficult to deliver. CDNs have typically relied on a traditional cloud model to support streaming customers, but they realized centralized clouds were too far on average from the end points they sent data to, racking up costs, latency and downtime.
- UPSHOT: Computational Storage can solve for these issues. While a typical CDN traffic flow involves lots of data movement and processing spread out over a variety of edge infrastructure, Computational Storage, can simplify this flow. Thanks to in-situ processing, this is all done within the storage device itself. This means more people get to watch their content faster with less overall hardware overhead. Instead of doing all the content verification and security in a centralized location, the use of Computational Storage allows for the device to authenticate and even decode thread for each user is an added value to this market.