Rebuilding Cost Models:�

How FinOps Helped APTrust Shape a Sustainable Business Strategy for Long-Term Preservation

Allison C.F. Ruffner

Lead Engineer for DevSecOps

May 20^th 2026, University of Wisconsin

The Story Line

In 2024 APTrust, a consortium-based digital preservation service, was facing an inflection point: our longstanding model, built on a flat $20,000 annual membership fee and low rates for additional storage, would soon fail to cover the full costs of cloud and organizational operations. For 10 years APTrust membership fees had not changed, but the cost to deliver had increased. This necessitated a deep dive into cloud costs and other expenditures. The power of practicing FinOps was such that even at the crawl, it helped us shape a sustainable future.

• FinOps helped us gain insight our cloud costs, identify a problem in our cost recovery model, and create a long-term viable business strategy.
• Exercising FinOps helped us find value in our cloud environment and be far more cost-efficient in operations while positioning us for future maturity.

​

​

“FinOps is an operational framework and cultural practice which maximizes the business value of cloud and technology, enables timely data-driven decision making, and creates financial accountability through collaboration between engineering, finance, and business teams.”��FinOps Foundation Technical Advisory Council�Updated: January 2025

What Is FinOps?

Fiscal Lean Management For Cloud Computing

“Every engineering decision is a buying decision.”� - Erik Peterson CTO/CISO @CloudZero

FinOps Capabilities In Today’s Presentation

• Not very granular
• Simple build patterns

• Informal review process
• Engineers do individual research

​

• Primary focus is on compute
• Focus on top resources, cost

Rate Optimization

Architecting for the Cloud

Workload Optimization

• Tagging exists
• Manual, annual
• Limited visibility

Allocation

Unit Economics

• Not very granular
• Simple build patterns

Reporting

• centered on high level cost
• Simple build patterns

Introducing APTrust

01

Digital preservation operating 100% in the cloud.

Preserving the Scholarly and Cultural Record

​

​

A consortium of 20+ universities and memory institutions within the US.

Who We Are:

How APTrust Works

• Provide a sustainable shared digital repository serving the long term goals of membership.
• Desktop and member/depositor-hosted tools to support preservation workflows.
• DART, which aides depositors in sending their content to APTrust’s repository.
• Preservation related services and activities.
• Bulk restoration, for Disaster Recovery.
• A community for shared learning, problem solving, and support for digital preservation.
• Workgroups to improve preservation workflows.

​

Extending the useful life and unique materials so they are accessible for future generations of students and researchers.

APTrust collects revenues only to meet the cost of operating APTrust and maintaining a reserve fund adequate to cover the costs of sunsetting the organization.

Fees & Cost Recovery

The Early Years

02

Forecasting

Budgeting

Planning & Estimating

​

​

2011 - 2019

​

​

• Monolithic design.
• Simple, stable cloud assets.
• Kept things in a ‘legacy’ culture.
• Focus was on development and features, not infrastructure.

Preservation Is A Long Game

APTrust began like a ‘shift and lift’ even though it began in the cloud. Founding members placed development constraints based on concerns over cloud immaturity, vendor lock, and vendor abandonment. Portability back to a member data center or local system was a must have requirement.

• Not much demand for FinOps maturity.
• Provided concise prices for technology assets.

Legacy Architecture

Challenges with Architecture.

​

• Brittle, inelastic, difficult to scale.
• Wasteful.
• 5TBs of EBS storage sitting idle waiting for ingests.
• The lack of data from this environment limited the ability to get a clear understanding of or allocate cloud costs.
• Planning and forecasting are basic.

​

​

​

​

​

​

Architecting for the Cloud

03

Going Cloud Native

Allocation

Workload Optimization

Rate Optimization

​

2020 - 2023

​

​

• Bring visibility – to costs, assets, processing, security.
• Remove waste – idle processes/assets, code, our time.
• Align the architecture to value enhance our ops, development and security.

Optimization, Value, & Culture

2020 – 2023 was a period of significant modernization for the APTrust preservation platform as it moved to a cloud native architecture and application design. The need for a more secure, cost efficient, time efficient, scalable, supportable and sustainable environment drove effort, while presenting opportunities to use FinOps strategies/capabilities from initial design.

What the APTrust Team Desired

Challenges The Team Faced

• Idle systems for long periods
• Thousands of files deposited at once
• Files’ sizes ranged from KBs to TBS

Transitioning from EBS to S3 storage for data streaming processing and data transformation was a uniquely innovative solution for APTrust with a combination of benefits.

​

• Unlimited scaling as required.
• Far less expensive than EBS.
• Removes waste by only using what is needed.
• Performance is significantly higher, due to massive footprint and api access surface area.
• Workload Optimization

​

Due to tagging we discovered our spending on storage for data processing with EBS dropped from ~250$/month to ~85$/month. This was a 66% drop in spending.

​

Data Streaming Processing On S3

Link to AWS’ 2022 S3 Deep Dive here to see how this actually works.

PaaS provided a great solution for APTrust to find value while thinking about simplification for cost efficiency and positioning for FinOps

• ECS Fargate runs tiny containers, scaling as needed, keeping costs low for variable workloads. Scales at the service level, aligning with the business value and allocating costs through tagging.
• RDS, Elasticache, ALB, EFS were much easier staff serving Ops. RDS and Elasticache also presented low hanging fruit for RI purchases that continue to support annual discounts for compute. We use the tiniest Elasticache instances. Rate optimization & workload optimization are key.
• Spot instances are used for containers in Dev and Demo environments, while Prod uses a combination of On Demand and Spot Instances; Rate Optimization.

​

​

​

Microservices and Paas

*APTrust could now see they spent 12$/month on fixity checks in the new architecture in prod. Fixity is a critical process for preservation ops.

Lexicons, Tags, & Allocation

​

​

​

​

​

​

A critical strategy to gain visibility, identify assets and allocate costs to the relevant business process was integrating a tagging solution into the lexicon defining the environment. Tagging is an essential component of allocation for aligning cloud costs with business value.

Inflection Point

04

Unit Economics

Allocation

TOC

​

2024

​

​

Signals

In 2024, APTrust observed signals that combined to indicate the revenue stream and storage cost recovery model would no longer cover expenses.

• UVA funding subsidy was cut in half from 400k to 200k.
• Research into pricing new storage options per GB indicated underlying losses.
• Other community-based preservation platforms were sunsetting due to financial failures.
• Inflation.

APTrust had to address challenges from expenditures and budgeting based on what we were observing in the environment. We had to consider new ways of thinking about how cost recovery was done, and how it should be allocated. Because of earlier efforts with Allocation and Architecting for the Cloud would could now dig much deeper into the data and focus more on Unit Economics.

Signals

• Determine the True Cost of Ownership of the preservation platform. (IT)
• Allocate according to IT costs/preservation using a unit metric. (Unit Economics)
• Address other costs rising due to inflation in a different but parallel method.

Digging Into Cloud Costs

2024 Environment operational costs - does not include storage or AWS Support costs.

�

Overhead annual storage costs based on Oct 1, 2024. - Storage for demo,staging, and ops not included in membership billed storage amounts. 

�

A New Cost Model

05

Unit Economics

Reporting

TOC

​

​

2025 and Beyond

​

​

IT Cost Recovery Model

(OSTC + CIOC + OSC + BSC) / TMS

(Operations Support Tech Costs + Cloud Infra Overhead Costs + Overhead Storage Cost + Business Support Costs) / Total Membership Storage

The cost of preservation services per GB of storage.

Reporting

Fiscal Years 2025 - 2032

The APTrust Executive Director proposes fees for all services, with input from the Advisory Committee, and the APTrust Governing Board approves them. The additional layers of data helped APTrust’s Members understand the relationship between the business value of preservation and cloud costs.

Storage price increases, no Sustaining Membership fee change.

Storage price increase, Sustaining Membership fee increase to $25,000

Though this data was was centered on high level aggregated costs, the deepest data level isn’t required to make clear and impactful decisions. Membership was able to walk through APTrust’s option of future fee strategies.

Deep Archive: $80/TB/Year ($60 increase, 103.53% cost recovery)�

Proposed Fees For FY 2027

New Membership Fee: July 1, 2027 (May 2027 invoices, due August, FY28)

New Storage Fees: April 1, 2027 (May 2028 invoices, due August, FY 29)

* Board Approved Feb. 2026 *

​

Consulting

Glacier: $110/TB/Year ($50 increase, 100.48% cost recovery)�

High Assurance: $460/TB/Year ($40 increase, 100.53% cost recovery)�

Membership Fee: $25,000�

Proposed Timeline:

Final Thoughts

FinOps provides transparency. The more you see, the more waste gets removed, the easier to find value. Engineering in ways to support transparency makes your cloud ops that much leaner.

​

Being able to report the data and metrics around IT costs and the cost of preservation was very powerful for addressing APTrust’s cost challenges with our Membership and APTrust Board.

Don’t overcomplicate things. Keep it simple. The deepest data level isn’t required for effective decision making. Be lean with your own time.

Financial accountability is a collaborative activity. Wearing multiple hats helps, but APTrust’s team worked closely to engineer a cost-effective cloud stack capable of providing granular data for FinOps actions.

Do you have any questions?

flavia.ruffner@aptrust.org

acr7d@virginia.edu

aptrust.org

Thanks!

#finops

#cloud

#preservation

Presentation template created by slidesgo.

Icons by flaticon.