True Cost Comparison vs Hyperscalers – CPU
myresearchcloud.ca vs Hyperscalers (Apples to Apples)
Public cloud can look inexpensive on paper. In practice, once you normalize for true 1:1 physical CPU cores, storage performance, durability, support, and data transfer, the effective monthly cost is typically much higher than headline pricing.
myresearchcloud.ca is designed for academic and research workloads that need:
Predictable billing
Deterministic performance
Sovereign Canadian hosting
Non-profit, research-first mission alignment
myresearchcloud.ca Pricing Model (Predictable by Design)
myresearchcloud.ca provides fixed allocations with no variable line items.
Each vCPU includes:
1 dedicated physical Cascade Lake core (1:1)
4 GB RAM
8 GB replicated enterprise SSD
6 GB replicated HDD
Ceph replication included
No egress fees
No IOPS tiers
No surprise overages
3-Year Commitment: $17.25 per vCPU per month
Predictable monthly invoice. No usage traps.
What Hyperscalers Charge to Match Deterministic 1:1 Physical Core Allocation
To approximate deterministic 1:1 physical core performance in public cloud environments, customers typically require:
Dedicated host, sole-tenant node, or bare metal instance (to avoid shared-core overcommit)
Separate block storage volumes
Durability configuration (replication and snapshot retention)
Provisioned IOPS and throughput to avoid baseline performance caps
Enterprise or Business support plan allocation
Data transfer (egress) charges for research export workflows
Standard reserved instances alone do not provide physical-core isolation.
Realistic Effective Cost Per vCPU Per Month
(Committed, Sole-Tenant / Dedicated-Host Normalization)
When normalized to committed dedicated-host or sole-tenant equivalents with durable storage and moderate research data movement:
Compute (dedicated host / sole-tenant, committed): $35–65
Storage capacity (SSD + HDD equivalent allocation): $1–4
Provisioned IOPS & throughput (research workload typical): $1–3
Support plan allocation: $1–4
Data transfer allocation (moderate export workloads): $1–6
Realistic effective total: $40–80 per vCPU per month
Costs vary by region, contract structure, and workload behavior. Storage, performance provisioning, and data transfer charges are usage-dependent and can fluctuate monthly.
What “Apples to Apples” Means
This page compares cost under equivalent conditions:
1 vCPU = 1 physical CPU core (no oversubscription)
Sustained workloads (not burst)
Comparable durability (replication)
Comparable operational posture (support + reliability)
Reasonable allowance for data transfer
Important: Standard hyperscaler instances generally do not guarantee true 1:1 physical core allocation unless you select
dedicated host / sole-tenant / bare metal options.
True Cost Comparison
Normalized to true 1:1 physical core allocation and comparable durability, storage performance, and support.
| Feature | myresearchcloud.ca | Hyperscaler (1:1 Physical Core) |
|---|---|---|
| 1:1 Physical Core | Included | Dedicated host / sole-tenant required |
| Storage Included | Yes | Billed separately |
| Replication | Included (Ceph) | Multi-AZ / durability add-on |
| IOPS / Throughput | No artificial throttling | Provisioned performance required |
| Egress Fees | None | Metered and variable |
| Effective Cost / vCPU / Month | $20.08 | $45–85 |
Estimates reflect sustained research workloads under committed terms. Hyperscaler costs vary by region, agreement, and usage patterns. When normalized to true 1:1 physical core allocation and comparable durability, hyperscaler costs are typically 2×–4× higher for sustained research workloads.
Summary Comparison
myresearchcloud.ca is typically 2×–4× less expensive than hyperscalers when normalized to:
true 1:1 physical cores
comparable durability
comparable storage performance
and a predictable, supportable operating model
Assumptions and Methodology
1. Physical Core Equivalence
All comparisons presented on this page normalize infrastructure to a 1:1 mapping between virtual CPUs and physical CPU cores, meaning that one vCPU corresponds to one dedicated physical core without oversubscription; in hyperscaler environments, achieving this level of physical isolation typically requires the use of dedicated host, sole-tenant, or bare-metal offerings, as standard shared instance types generally do not guarantee exclusive access to underlying physical cores.
2. Sustained Research Workloads
The cost model underlying these estimates assumes continuous or near-continuous utilization over multi-month to multi-year time horizons consistent with funded academic research projects, and does not attempt to model short-lived burst workloads, highly elastic auto-scaling architectures, or ephemeral commercial consumption patterns optimized for transient usage rather than sustained allocation.
3. Storage and Durability Parity
Storage assumptions reflect parity with replicated Ceph storage, including built-in replication across failure domains, and where hyperscaler environments require multi-AZ configurations, durability add-ons, or separate configuration of storage performance parameters to achieve comparable resilience and sustained performance, the associated costs are incorporated into the effective per-vCPU estimates presented herein.
4. Provisioned Performance Requirements
Provisioned IOPS and throughput assumptions reflect typical research workloads that require sustained storage performance beyond baseline cloud volume allocations, and hyperscaler cost estimates include representative charges for performance provisioning where necessary to avoid artificial IOPS or throughput caps that would otherwise constrain research workflows.
5. Data Transfer Assumptions
Data transfer modelling reflects common academic collaboration patterns, including dataset ingestion, inter-institutional sharing, and periodic export of results, and hyperscaler cost ranges incorporate reasonable blended estimates for transfer charges; actual transfer costs may vary by region, agreement structure, and workload characteristics.
6. Enterprise Support Considerations
Enterprise-grade support allocations are included in hyperscaler estimates where appropriate in order to reflect operational parity and institutional expectations for reliability, incident response, and vendor support within research environments.
7. Pricing Variability and Predictability
Hyperscaler pricing varies materially by region, negotiated enterprise agreement, committed usage level, and configuration choices, and may fluctuate month to month based on storage performance tuning, snapshot retention, and data transfer patterns, whereas myresearchcloud.ca pricing is fixed per allocation term and does not vary based on IOPS tiers, throughput provisioning, or egress volume.
8. Informational Purpose
All estimates presented on this page are intended for informational and budgeting comparison purposes only, and actual pricing and performance will depend on specific configuration details, agreement terms, and workload characteristics.
Why This Matters for Grants and Research
Academic research operates on fixed funding envelopes, defined timelines, and clear reporting requirements. Budgets submitted in grant proposals must be understandable to reviewers, defensible to institutions, and manageable by research administrators. Variable cloud billing models—where compute, storage capacity, IOPS, throughput, support, and data transfer are all metered separately—can introduce uncertainty that is difficult to model accurately over multi-year projects.
Even when headline instance pricing appears competitive, total cost often depends on configuration decisions that evolve over time. Adjusting storage performance, increasing throughput, retaining additional snapshots, or transferring larger datasets can materially change monthly spend. For grant-funded research, that variability adds administrative overhead and financial risk.
Academic budgeting benefits from pricing that is:
Easy to justify in proposals
Consistent month to month
Not sensitive to storage performance tuning
Not exposed to transfer overages
myresearchcloud.ca provides fixed allocations with predictable monthly billing and clearly defined inclusions. Storage performance is not a separate line item, replication is built in, and data transfer is not metered. Researchers can plan for the full lifecycle of a project with confidence, without needing to continuously optimize configurations to control cost.
It is infrastructure aligned with how research is funded and executed—designed for sustained academic workloads, not variable retail cloud billing models.