By Sim Upadhyayula, Senior Director of Solutions Enablement at Supermicro

2020: A year where old problems will be looked at with a fresh perspective

With the recent advances in compute and storage technologies, 2020 could very well be a year where old problems in the datacenter will be looked at with a fresh perspective and recent problems tackled with renewed vigor.

1)      Failure Detection/Preventive Maintenance/Predictive Insights/ Analysis:  The concept of predicting impending failures in a datacenter infrastructure and alerting/taking corrective action is nothing new and has been attempted previously. However, at that time the additional compute/storage resources required for AI/ML training and inferencing was simply not economical to go mainstream. With the increased core count and accelerated GPU/FPGA available and increased densities in low latency NVMe storage, it is now possible to train and deploy predictive software that provides a more granular insight into the health of various data center infrastructure components and take remedial actions without huge inlays of human capital. This has huge potential of greatly simplifying the maintenance and management aspects of the modern day datacenters regardless of size and scale. We will continue to see accelerated integration of predictive/insightful software solutions into infrastructure deployments and increased rules based actionable insights.

2)      One cloud ain't enough:  Many customers are already relating that one cloud is not able to provide enough cover. It will likely take one private cloud and at least two public cloud providers to provide the reliability/availability/resiliency/security/governability & optimization demanded from their workloads. Managing varied interfaces and different cloud computing platforms is a messy business. We are likely to see a concerted effort by some cloud providers, third party ISVs and OEM infrastructure suppliers working towards a framework that abstracts & provides the ease of management of the workloads across different environments.

3)      Leading from the Edge:  The coming year is likely to see a continuation of workloads moving to the edge, where significant portion of data analysis will be localized. Increased compute cores and density in low latency NVMe allows skilled system designers & manufacturers to build systems that are smaller in form factor and importantly more tolerant to differences in operating temperatures and harsher environments.  

4)      More Computing power translates to more burden on Earth's resources:   Powerful high wattage processors and low latency storage technologies of today are increasingly power hungry and dissipate a lot of heat. Datacenters will likely have a hard time keeping things under control with existing designs unless they resort to system manufacturers dedicated to developing innovative, efficient, greener next generation technologies.

##

About the Author

Sim Upadhyayula is a Senior Director of Solutions Enablement at Supermicro. In his role, Sim collaborates with various enterprise and service providers in understanding their next-generation data center needs and identifying the right solutions to maximize their business outcomes.