ABOUT KOVE : SDM™
Kove : SDM™ liberates enterprises , giving them smarter artificial intelligence and machine learning , faster rendering and number crunching , more powerful analytics and the ability to put data centre performance where it was never thought possible .
The solution can be deployed in cloud , edge and on-premise environments to enhance AI / ML , high-performance computing , animation and any other high-demand applications to :
Improve Overall Performance Kove : SDM™ statistically provides local memory performance to CPUs even when memory resides across the data centre . This enables high CPU utilisation rates and even CPU saturation , when using remote memory . As a result of the unlimited dynamic memory sizing , any size computation can run completely in memory
Maximise Hardware Utilisation Get the most out of your existing hardware . While memory and CPU costs are typically 65-85 % of IT investment , most servers only achieve 20-40 % utilisation . But Kove : SDM™ ’ s provisionable memory unlocks CPU potential , so your technology infrastructure – and software engineers – can reach their full potential
Increase Return on Data Scientists Enable data scientists to spend more time analysing data rather than subdividing datasets to fit hardware constraints . With Kove : SDM™ , data scientists can run whatever size datasets they want , sizing memory to data on the fly . As a result , data scientists iterate faster , improving time-to-solution .
Achieve Sustainability Goals With increased utilisation of each server , you require fewer servers to achieve the same work , which reduces your energy needs . Imagine using less cooling , a smaller footprint , and delivering more efficient processing to your teams .
Reduce TCO Decouple memory from individual servers to achieve superior results while needing fewer servers , reducing all associated operating costs , including power and personnel . Most customers achieve an ROI in the 50-80 % range for their Kove : SDM™ investment – and it can be as high as 200 %.
datacentremagazine . com 89