A new study points to the fact that Moore’s Law continues to apply for cloud compute and that companies are breaking the traditional three-year data center hardware renewal cycle, instead moving to a 12-month cycle in the cloud.
The State of the Cloud 2018 from Cloudability surveyed more than 1,500 companies, accounting for $2.5 billion in cloud spend, and was conducted throughout the course of 2017.
At the beginning of the year, 42% of companies used legacy compute instances, with 58% using current generation instances. When the year concluded, only 30% of companies remained on legacy compute instances, while 70% had moved to current generation instances. This trend is in line with the fact that many businesses are continuing to take advantage of the advances in improved processor, memory, storage, and networking cycles.
From the study, three big trends emerged. For one, standard compute families are still very common, with 43% continuing to provision most of the compute workloads on standard instances.
The second is serverless computing continues to grow very quickly. The report shows quarterly breakdown growth rates throughout 2017: 100% growth in Q1; 138% in Q2; 321% in Q3; and 667% in Q4.
The final big, emerging trend is that AI (artificial intelligence) and machine learning continue to surge as well. A quarterly breakdown shows the following growth rates: 100% in Q1; 183% in Q2; 252% in Q3; and 357% in Q4.
Still, out of hundreds of cloud services available, only four accounted for the majority, or 85% of spend. These four are: Amazon Elastic Compute Cloud, Amazon Elastic Block Store, Amazon Relational Database Service, and Amazon Simple Storage Service.
Going forward, the study forecasts that both migration to the cloud and existing cloud services distribution will continue to hold steady in the near term.
Want to tweet about this article? Use hashtags #IoT #M2M #cloud #Moore #compute #hardware #Cloudability #AI #machine #learning #Amazon