← Case studiesSoftware and SaaS · United States · 2026

Pricing a complete AWS operating cost for a legal technology SaaS provider

$193,419
Annual value of the SQL Server BYOL decision
$1,053,338
Recommended annual position, three-year all-upfront
184 cores
SQL Server Standard entitlement required for full BYOL
475 TB
Backup data moved from Veeam to AWS Backup
SectorLegal practice management software, United States
Estate assessed66 in-scope instances
Composition53 Windows servers, 13 SQL Server Standard database hosts on Ubuntu, 2 non-production consolidation hosts
Storage240 TB provisioned, 475 TB of retained backup data
ScopeCompute, Microsoft licensing, storage, backup and observability
Recommended position$1,053,338 a year with SQL Server BYOL, three-year all-upfront
Headline outcomeA $193,419 annual break-even that turned a licensing debate into a procurement decision

The challenge

This engagement differed from a conventional migration assessment. The requirement was not compute and licensing in isolation, it was a complete AWS operating position covering backup and observability alongside infrastructure, so the provider could compare a whole-platform cost against what they were running today.

Modelling only Amazon EC2 and Amazon EBS would have produced a number the provider could not act on. Backup was running on an incumbent Veeam platform consuming 10 virtual machines. Monitoring and logging sat on separate tooling. Both would need replacing or retaining, and either way both belonged in the comparison.

The database estate was uniformly SQL Server Standard across all 13 hosts, which made the licensing question unusually clean: full licence included, or 184 Standard cores under active Software Assurance. That is a procurement decision, but only once someone puts a number on it.

Our approach

We right-sized against peak utilisation assumptions of 60% CPU and 80% RAM, then built three distinct cost layers into a single total cost of ownership model.

Compute and Microsoft licensing. Modelled under both full SQL Server licence included and full SQL Server BYOL, across on-demand and committed pricing, so the delta between the two was explicit rather than buried.

Storage. A tiered Amazon EBS gp3 model rather than a flat provision. Ten percent of tiered capacity was provisioned as an active tier at 10,000 IOPS, with the remaining 90% running at the gp3 baseline of 3,000 IOPS and 125 MB/s, which carries no additional charge. TempDB for all 13 SQL Server hosts was satisfied directly by instance NVMe, removing 13 separate EBS volumes from the model entirely.

Managed services. AWS Backup on a warm tier with 30-day retention across 475 TB, replacing the Veeam platform and taking 10 backup virtual machines out of scope. Amazon CloudWatch for custom metrics, centralised log ingest and 90-day log retention across all 66 workloads, covering search, alerting and compliance evidence.

Two further managed services were considered and excluded at the provider's direction, which we recorded in the model rather than silently dropping.

The results

ModelOn demandThree-year all-upfront
Shared tenancy, SQL Server licence included$1,591,936$1,246,757
Shared tenancy, SQL Server BYOL$1,398,515$1,053,338

The BYOL decision is worth $193,419 a year, and that figure is the break-even. If the provider can source 184 SQL Server Standard cores with active Software Assurance below that annual rate, BYOL wins. Stated that plainly, a licensing debate becomes a procurement decision the provider can take on their own numbers.

The committed BYOL position sits 34% below on-demand licence included.

Managed services accounted for $364,308 a year, roughly a third of the three-year BYOL total and the largest non-compute line in the model. Storage came to $240,916 a year, with the tiering strategy and NVMe TempDB placement holding IOPS charges above baseline to just $10,920 of that total.

Both managed services price linearly on consumption and sit flat across every commitment term, which meant the provider could see exactly which lines scale with the business and which do not.

Why it matters

Customers evaluating a platform move need the whole operating cost, including the services that replace incumbent tooling. A compute-only model always looks better than reality and always loses credibility at the first invoice.

Surfacing backup and observability as explicit, linearly scaling lines made the comparison against the existing platform straightforward. It also exposed the most useful insight in the engagement: managed services, not licensing or storage, were the largest non-compute cost, and the one most sensitive to retention policy. That is a lever the provider controls directly.

Tooling

Collection for this engagement ran through Cloudamize. What that tool can and cannot see: read the tool page →

Talk to us about your AWS operating cost

An AWS Optimization and Licensing Assessment can price compute, Microsoft licensing, tiered storage, backup and observability in a single model, so you are comparing whole platforms rather than line items.