Despite the rapid evolution and growth of public cloud usage, enterprises are finding value in on-premises IT infrastructure. As a result, some organizations are moving their workloads back, partially or entirely, to their own data centers. In fact, according to a survey conducted by IDC, over ½ of the IT spent on servers and storage is still driven by on-prem deployments, and over 70% of those surveyed said they plan to repatriate workloads back from public cloud to an On-Prem Infrastructure.
6. Thinking about your workload infrastructure in the next two years,
in what environments do you expect your workloads will run?
$63.0B
$75.2B
0%
1%
2%
3%
4%
5%
6%
7%
$56
$58
$60
$62
$64
$66
$68
$70
$72
$74
$76
$78
2021 2022 2023 2024 2025 2026
Spending
($B)
Enterprise Infrastructure, On-premises Deployment
Environments, $B
Spending ($B) Y/Y
2021-2016 CAGR: 3.6%
On-premises Compute and Storage Infrastructure
Source:
Chart on the left: Repatriation of Workloads from Shared to Dedicated Infrastructure Environments Update: More Focus on Operational Side, IDC #US47961422, March 2022
Chart on the right: IDC’s Worldwide Enterprise Infrastructure Tracker: Buyer and Cloud Deployment, June 2022
22%
23%
28%
29%
37%
Off Premises non-cloud
Public Cloud
Hosted Private Cloud
On Premises non-cloud
On Premises Private
Cloud
n=2,235
7. Thinking about your workloads which currently run in the public cloud (fully or partially),
do you expect any of them will be repatriated and moved to private cloud or non-cloud
infrastructure in the next two years?
Workload Mobility Is Becoming a Norm
Source: Repatriation of Workloads from Shared to Dedicated Infrastructure Environments Update: More Focus on Operational Side, IDC #US47961422, March 2022
Primary reasons why workloads will be repatriated from the public cloud
6%
10%
13%
71%
Don't know
Will run in hybrid
environment using public
cloud and private cloud/non-
cloud IT infrastructure
Will run fully in public cloud
Will move into private cloud
or non-cloud environment 8%
10%
13%
14%
15%
18%
26%
27%
33%
Lack of enterprise features in public cloud
offerings
Dissatisfaction with public cloud customer
service
Regulatory compliance issues
Bandwidth issues
IT consolidation efforts
Performance issues
Data privacy concerns
Pricing of public cloud services is too
high/unpredictable
Data security concerns
WEGHTED AVERAGE ACROSS ALL WORKLOADS
8. Interviews With Three Technology Companies
Operating Their Own IT Infrastructure
Efficiency and
Sustainability
Cost and Flexibility
Infrastructure Control
and Performance
Predictability
9. Intel: Efficiency and Sustainability
One of the largest technology companies in the world. Core expertise: infrastructure component design and
manufacturing. Big focus on sustainability. IT operations: 16 datacenter sites with 56 datacenter modules, more than
380,000 servers, 787 petabytes of storage capacity, and more than 725,000 network ports. About 95% of the servers are
used for chip design, 3% are used for traditional enterprise and office workloads, and the remaining 2% are used in the
manufacturing computing space, which includes fabrication and assembly test manufacturing plants.
"[Disaggregated architecture]
serves two purposes. One is
business advantage: it reduces
the amount of money you need
to upgrade; you only pay for
what you want to upgrade to
realize the value. And the second
one is significant ewaste
reduction: if you look at a 3U
chassis, nearly 82% of the weight
of the material you don’t need to
change."
Infrastructure management experience:
• Efficiency through AI-based automation
• Training IT managers for engineering approach
to infrastructure management
Benefits of operating own IT infrastructure:
Utilization of disaggregated architecture which drives
Future IT plans:
• Further cost reduction
• More automation
• More predictability
Cost Savings Reduction in ewaste Green Computing
10. Twitter: Cost and Flexibility
One of the largest social network providers. Operates three datacenters, hundreds of thousands of servers and multiple
petabytes of storage in the United States. All datacenters are hosted in colocation facilities, but the company operates
and manages its own compute and storage infrastructure. The company also utilizes public cloud resources for certain
workloads and data (e.g., cold storage) and targets to achieve close integration between its own dedicated and shared
cloud IT resources.
“…On-prem will still be the
backbone of our service… We can
tailor our hardware solutions to
fit what we need and strip out
what we don't need and make
that pretty cost effective. We can
do things like that we wouldn't
normally be able to do with just
an off-the-shelf solution or in the
cloud without paying extra
money for it."
Infrastructure management experience:
• Ongoing collaboration between technical
teams within Twitter and with datacenter
operators
• Working with internal customers to assess
technical needs
Benefits of operating own IT infrastructure:
Future IT plans:
• Better integration between
on-prem and public cloud
solutions
• Ability to move workloads
back and forth
Cost Savings Hardware optimization
for workloads
Adequate
security and data
privacy
11. Preferred Networks: Infrastructure Control and
Performance Scalability
An eight-year-old company that designs and develops one of the most powerful supercomputers in the world, based in
Japan. Primary business is developing and implementing projects requiring deep learning techniques. The company also
has expertise in robotics, pharma, and other business domains. PFN built its own supercomputer with a GPU cluster,
which currently housing 2,000 GPUs. PFN also designs and maintains its own storage tuned to needs of deep learning
workloads.
"We have a subsidiary … and
they use plenty of computing
power [in public cloud]. If we
open the door of our own
infrastructure to them, it will be
good for our business because
we have a fast accelerator, we
can provide more stable and
predictable GPU resources … we
can provide a predictable facility
to the subsidiary."
- Yusuke Doi, VP of Computing
Infrastructure at PFN
Infrastructure management experience:
• Rule-based automation
• Constant monitoring of resource utilization by
user role
Benefits of operating own IT infrastructure:
Future IT plans:
• At least, double the number
of on-premises server
deployments in the next five
years
• Better infrastructure control
Cost Savings Stable and predictable
performance
12. Key Takeaways
• Optimization for core
workloads
• Cost
• Control
• Flexibility
• Predictability
Common
reasons
organizations
keep investing
in on-premises
IT
infrastructure:
• Security/data privacy
• Sustainability
…and also:
• Workload needs
• More advanced IT automation
• Right skillset and training
• Integration of resources from
multiple IT environments
Important
considerations
when
investing in
on-premises IT
infrastructure:
• Trusted partner ecosystem
…and also: