Trends in Cloud Repatriation

Trends in Cloud Repatriation with Silk & ESG

Topic: Understanding Cloud Repatriation, Its Drivers, and How to Avoid It

Speakers:

Stephen Catanzano, Senior Analyst, Enterprise Strategy Group (ESG)

Dwight Wallace, Director of Go-To-Market, Silk

Summary

In this Silk + ESG session, Stephen Catanzano from Enterprise Strategy Group joins Dwight Wallace of Silk to examine the growing trend of cloud repatriation — when organizations move workloads back from public clouds to on-premises environments.

They discuss why more than half of organizations have done this, the financial and performance pitfalls driving it, and how companies can avoid repatriation by improving performance, cost efficiency, and control.

Dwight shares lessons from Silk customers like Sentara Health and Wayfair, showing how Silk’s software-defined cloud platform maintains performance and resilience — letting organizations stay confidently in the cloud.

Key Takeaways

54 % of organizations have repatriated workloads due to cost or performance issues.

Cost and performance are inseparable — optimizing one affects the other.

Over-provisioning averages 30 %, inflating cloud and licensing costs.

Silk’s architecture separates compute and storage performance to restore control.

Thin cloning, data reduction, and automation dramatically reduce cloud expense.

Customers like Sentara Health and Wayfair achieved full cloud agility and scale with Silk — avoiding repatriation altogether.

Transcript
[00:00 – 01:00] Introduction and ESG Findings

Stephen Catanzano (ESG):
Welcome, everyone. I’m Stephen Catanzano, Senior Analyst with Enterprise Strategy Group. I’m here with Dwight Wallace, Director of Go-To-Market at Silk, to talk about cloud repatriation — what it is, what’s driving it, and how to avoid it.

Recent ESG research found that 54 % of organizations have moved applications to the cloud and later brought them back on-premises.
The top reasons include:

Higher-than-expected costs

Unpredictable cloud expenses

Inability to meet performance, scalability, or availability requirements

Utility computing not living up to expectations

Stephen: Dwight, maybe start us off — how do you define cloud repatriation?

[01:00 – 02:00] What Cloud Repatriation Means

Dwight Wallace (Silk):
Cloud repatriation is the movement of applications and data back from the cloud to on-premises environments.

And the reasons you just mentioned align perfectly with what we hear at Silk. Performance drives cost in the cloud — and when performance expectations aren’t met, costs soar.

You can’t decouple performance from cost. Many customers migrate to the cloud for agility but end up returning because of these performance and expense challenges.

[02:00 – 03:00] Why Repatriation Is a Tough Decision

Stephen:
That must be a difficult decision for organizations to make.

Dwight:
Absolutely. It’s often the last resort after investing enormous time, money, and reputation into a cloud strategy.

Think about the resources required — planning, staffing, consulting, re-architecting. Reversing that is painful.
But when costs spiral and performance degrades, many organizations see no other choice.

[03:00 – 04:00] Avoiding the Pitfalls — The Role of Planning

Stephen:
Where do organizations go wrong, and how can they avoid reaching that point?

Dwight:
Planning is everything. Every migration should include an exit strategy — what happens if the cloud doesn’t deliver.

Start with data rationalization:

Place critical data on high-performance tiers.

Model how performance affects CPU and database licensing costs.

Always perform proof-of-value testing with real workloads.

Most organizations over-provision by 30 % or more, which compounds infrastructure and license costs.

[04:00 – 06:00] The Over-Provisioning Problem

Stephen:
That 30 % figure is striking. Why is over-provisioning so common?

Dwight:
Because the cloud’s performance scaling is vertical.

If you need more storage performance, you must scale up the entire VM — more vCPUs, more memory — which inflates both infrastructure and licensing costs.

Silk changes that. It sits directly on the IaaS layer, using the VM’s network bandwidth instead of the shared cloud storage network. That lets you design storage and compute independently, optimizing each for cost and performance.

It’s about regaining control — of both agility and expense.

[06:00 – 08:00] Data Security, Control, and Efficiency

Stephen:
Our research also shows data security and availability remain top concerns. How does that factor into repatriation?

Dwight:
Control is everything. Customers want authority over patching, availability, and data protection.

Silk helps by providing:

Cross-zone replication and data redundancy

Compression and deduplication to cut cost

Thin cloning to eliminate full-size dev/test copies

Instant cloning and cross-region mobility improve resilience and reduce expense.
Cloud gives agility — Silk adds control and predictability to it.

[08:00 – 10:00] Data Management and Thin Cloning

Stephen:
How are Silk customers approaching data management for both cost and agility?

Dwight:
They’re bringing on-prem efficiency to the cloud.

Historically, dev teams used full, production-sized copies. In the cloud, they’ve had to compromise — smaller data or lower performance.

Silk ends that compromise with thin, instant cloning.
You can spin up unlimited, full-featured database copies at near-zero cost, enabling:

Faster dev cycles

Parallel testing

No production impact

Lower total spend

[10:00 – 12:00] Strategies for Cloud Success

Stephen:
If an organization is struggling, what should they do before repatriating?

Dwight:
First, assess size and growth rate — monolith or microservices?

Ensure data sits on the right tier for its SLA.
Don’t be afraid to use non-native platforms if they deliver the performance you need.

Avoid one-vendor lock-in — be flexible and align platforms with business goals.

And remember data gravity: large datasets dictate where workloads can realistically live.

[12:00 – 15:00] How Silk Prevents Repatriation

Stephen:
So how does Silk help customers stay in the cloud?

Dwight:
About 70 % of workloads run fine on native services; Silk focuses on the top 30 % — the mission-critical tier.

Running on the IaaS layer and leveraging the VM network, Silk delivers near-maximum performance with lower cost.

We give customers:

Full control over patching, replication, and mobility

10× faster performance than native tiers

Integrated data services for cloning, snapshots, and replication

In short: performance, cost control, and availability — without leaving the cloud.

[15:00 – 18:00] Resilience, Scalability, and Automation

Dwight:
Silk enhances availability through erasure coding and resilient data placement.

It’s fully scriptable, supporting dynamic scaling for:

Seasonal peaks (e.g., Black Friday)

Heavy ETL workloads

AI training or testing cycles

Scale up or down automatically — pay only for what you use.

That’s flexibility native services often can’t match.

[18:00 – 23:00] Customer Example #1 — Sentara Health

Stephen:
Let’s talk customer success stories.

Dwight:
First, Sentara Health, a major non-profit health system in Virginia and North Carolina.

Their CTO, Jeff Thomas, wanted to move their Epic EHR fully to the cloud but hit performance limits.

Silk solved that.
We delivered the performance they needed while reducing cost.

Their ETL workloads that once took hours now finish in ~45 minutes, ensuring real-time data access for clinicians and staff.

[23:00 – 25:00] Customer Example #2 — Wayfair

Dwight:
Another great story is Wayfair, one of the world’s largest online retailers.

They struggled to meet performance requirements in the cloud until Silk stepped in.

We helped them move a dozen critical databases with full performance parity.
Now everything you see on wayfair.com — including AI-driven personalization — runs on Silk.

Their teams use thin instant cloning for rapid development and analytics, giving engineers production-grade environments at minimal cost.

[25:00 – 26:00] Closing Remarks

Stephen:
Those are powerful examples of avoiding repatriation with the right platform.

Dwight:
Exactly. The cloud delivers agility — Silk ensures you keep it without losing performance or cost control.

Before choosing to repatriate, explore solutions that give you that control back.

Stephen:
Excellent insights, Dwight. Thanks for joining, and thanks to everyone who tuned in.

To learn more, visit silk.us or connect with Dwight Wallace on LinkedIn to explore how Silk can optimize your workloads.