MinIO announced the findings of its Object
Storage and AI Report, which surveyed 656 IT leaders on their storage and AI
priorities as part of a primary research initiative with User Evidence. The
survey reveals key insights into how organizations are leveraging object
storage for AI, machine learning (ML), and data-intensive workloads, as well as
the evolving demands placed on storage infrastructure to handle the scale and
speed requirements of modern AI applications.
Perhaps
the biggest takeaway from the research is the ubiquity of object storage today.
Respondents indicated that 70% of the enterprise's data was in object storage
today and it was expected to grow to 75% over the next two years.
The reason for the growth was AI. "Support AI" was the most popular reason why
IT leaders are adopting object storage at their organization. Performance and
scalability rounded out the top three.
"This
research confirms what MinIO, AWS, Azure and Google already know - that object
storage is the storage technology of the cloud - public, private or edge," said
Jonathan Symonds, Chief Marketing Officer, MinIO. "What is more interesting,
however, is the expected acceleration of adoption, driven by GenAI workloads.
Modern object stores are uniquely qualified to meet the demands of these
workloads - namely throughput performance, immutability and exascale. We are
keen to watch this research evolve as enterprises build new storage
architectures to support their AI ambitions."
While
the research shows AI to be a key adoption driver for object storage, other
workloads are equally critical, namely advanced analytics and modern datalakes
and lakehouses. IT leaders were particularly enthusiastic about modern
datalakes and lakehouses, with 92% of respondents indicating that they already
have one in place or plan to build one.
Another
interesting trend is revealed when looking at deployment models. Both the
public cloud and the private cloud are expected to grow their share of AI data
over the next 12-24 months. While the majority of respondents indicated that
their overall infrastructure was primarily in the public cloud, 68% were
concerned about the cost of running AI workloads. Looking specifically at
current approaches to running AI/ML workloads, the majority of IT leaders cite
a hybrid cloud approach as the most popular choice.
AI
deployments still come with challenges. IT leaders cited security and privacy
(44%), data governance (27%), and cloud-native storage (25%) as the three
biggest challenges to AI success at their organization. This data points to one
of the reasons enterprises repatriate data from public to private cloud-more
control over security and privacy. This, and the concerns around cloud-native
storage, reflects a larger trend of ensuring data portability.
"This
work represents one of the most comprehensive looks at the AI storage landscape
and there are some eye-opening findings", noted Ray Rhodes, co-founder at
UserEvidence. "IT leaders are clear in their preference for object storage - a
finding that transcended company size or geography. We look forward to
partnering with MinIO on more research in this key area."
To
access the full Object Storage and AI Report, please visit: https://resources.min.io/reports/object-storage-and-ai-report-12-2024.