Tier I & II: Basic Capacity

Introduction Picture your home internet connection: one cable modem, one router, one path from your house to the internet.

When something breaks, you're offline until it's fixed.

That's essentially a Tier I data center.

Now imagine you've added a second internet provider for backup, but you still only have one router and one power supply.

Better, right? That's Tier II.

The Uptime Institute established the Tier classification system in the early 1990s to create a standardized language around data center reliability.

Tier I and II represent what we call "basic capacity" facilities-they're designed to support IT operations, but they're not built to eliminate every single point of failure.

Tier I guarantees 99.671% uptime annually (roughly 28.8 hours of downtime), while Tier II bumps that to 99.741% (about 22 hours of downtime).

These percentages might sound similar, but they represent fundamentally different infrastructure philosophies.

Understanding these foundational tiers matters because not every workload demands five-nines availability.

Many organizations waste capital building Tier IV facilities for applications that could run perfectly well on Tier II infrastructure.

You'll gain practical knowledge about when basic capacity makes business sense, how to identify single-path risks, and why some of the world's largest operators still deploy Tier I and II facilities for specific use cases.

The Tier I Architecture: Single Path to Everything Think of Tier I like running a web application on a single EC2 instance with no auto-scaling group, no load balancer, and definitely no multi-AZ deployment.

Everything flows through one path.

A Tier I data center features a single distribution path serving the computer equipment.

You've got one UPS system, one set of power distribution units (PDUs), one cooling system, and one path from the utility grid to your server rack.

When maintenance happens, you shut down.

When equipment fails, you're down until repairs complete. Infrastructure Components:

• Non-redundant capacity components (think single power supply)
• Single, non-redundant distribution path
• No redundant backup systems for critical infrastructure
• Typically no raised floor (concrete slab is common)
• Basic environmental monitoring Scheduled maintenance requires full or partial facility shutdown.

Annual maintenance windows for Tier I facilities typically range from 48-72 hours, during which all or major portions of IT equipment must be powered down.

Major cloud providers don't operate Tier I facilities for customer workloads, but you'll find this architecture in smaller enterprise server rooms, edge computing locations, and development environments where downtime is acceptable.

CoreSite's earliest facilities in some secondary markets started as Tier I equivalents before being upgraded.

The economics made sense: lower capital expenditure (CapEx) by approximately 40-60% compared to Tier III builds, faster time to market, and adequate reliability for non-mission-critical applications.

The Tier II Upgrade: Adding N+1 Redundancy Where It Counts Tier II introduces the concept of n-plus-1 redundancy, but only for specific components.

Think of it like adding a second power supply to your server but keeping a single network card.

You've improved resilience, but single points of failure still exist.

The critical distinction: Tier II adds redundant capacity components but maintains a single distribution path.

Your UPS systems now have backup modules (N+1 configuration), and your cooling infrastructure includes redundant chillers or CRAC units.

But-and this matters-you still have only one path delivering power and cooling to the IT equipment. What Gets Redundancy in Tier II:

Component Type	Tier I	Tier II
UPS Modules	N	N+1
Cooling Units (CRAC/CRAH)	N	N+1
Generators	Single	N+1 typical
Power Distribution Path	Single	Single
Network Paths	Single	Single (though IT teams often add redundancy)
Utility Feeds	Single	Single

But if you need to perform maintenance on the main electrical switchgear or the primary power distribution bus, you're still taking an outage.

Digital Realty operates several Tier II facilities, particularly in markets where customer requirements don't justify Tier III economics.

Their Chicago data centers, for instance, serve regional enterprises running ERP systems, file servers, and backup infrastructure-workloads that benefit from component redundancy but can tolerate scheduled maintenance windows.

The Downtime Math: What Those Percentages Actually Mean Numbers tell stories.

Let's break down what 99.671% and 99.741% availability mean in operational reality. Annual Downtime Comparison:

Tier	Availability	Annual Downtime	Monthly Downtime	Per Week
I	99.671%	28.8 hours	2.4 hours	33.6 minutes
II	99.741%	22.0 hours	1.83 hours	25.2 minutes

If you're running a SaaS application that generates $10,000 per hour in revenue, that Tier I facility could cost you $288,000 annually in downtime-related losses.

Tier II reduces that to $220,000.

Still painful, but improving.

The Uptime Institute calculates these numbers based on planned maintenance and expected component failure rates.

Tier I facilities account for annual maintenance events lasting 1-2 days, plus unplanned outages from single-component failures.

Tier II reduces unplanned outages because redundant components absorb single failures, but planned maintenance still requires downtime. Real Failure Scenarios: In a Tier I facility, a generator failure during a utility outage means immediate shutdown.

Your UPS batteries provide 10-15 minutes (maybe 30 if you're well-provisioned), then everything goes dark.

In a Tier II facility, that same generator failure is covered by the N+1 redundancy-the backup generator kicks in automatically.

But lose both generators? You're in the same situation as Tier I.

And if you need to upgrade the automatic transfer switch (ATS) firmware, you're scheduling downtime regardless of how many generators you have.

Typical Use Cases: Where Basic Capacity Makes Sense Not every application needs Tier III or IV infrastructure.

Basic capacity facilities serve legitimate business purposes when aligned with appropriate workloads. Tier I Deployment Scenarios: Small branch offices and regional operations centers frequently use Tier I architecture.

QTS operates facilities in secondary markets where local businesses need basic colocation but can't justify premium pricing.

These customers often run:

• Development and testing environments (uptime isn't production-critical)
• Backup and disaster recovery sites (offline most of the time anyway)
• Edge computing nodes for content delivery where geographic distribution provides redundancy
• Office productivity applications with acceptable maintenance windows Meta's edge cache locations, which store copies of frequently accessed content closer to users, often use Tier I-equivalent designs.

Geographic distribution across hundreds of locations provides availability, even though individual sites have limited redundancy. Tier II Sweet Spot Applications: Tier II facilities serve enterprises that need better reliability than Tier I but can work within scheduled maintenance windows.

CyrusOne operates Tier II facilities serving:

• Regional financial services (non-trading systems)
• Healthcare providers (medical records, not life-critical monitoring)
• E-commerce operations with planned maintenance windows
• Enterprise resource planning (ERP) and customer relationship management (CRM) systems
• Video rendering and media production workflows Consider a regional hospital network.

Their electronic health records (EHR) system needs high reliability, and N+1 redundancy protects against unexpected equipment failures.

But they can schedule maintenance during low-volume periods (weekends, early mornings) when clinical operations are reduced.

Tier II provides the right balance of reliability and cost.

Switch operated several Tier II facilities in their early Las Vegas deployments before expanding into higher tiers.

Customers included gaming companies running loyalty programs, analytics platforms, and marketing systems-important applications, but not requiring 24/7/365 uptime without maintenance windows.

Practical Examples

Example 1: The ERP Migration Decision A regional manufacturing company with $500M annual revenue needs to colocate their SAP ERP system.

Their IT director is comparing Tier II space at $120/kW/month versus Tier III at $180/kW/month.

The deployment requires 50kW of power. Cost Analysis:

• Tier II: $120 × 50kW = $6,000/month = $72,000/year
• Tier III: $180 × 50kW = $9,000/month = $108,000/year
• Cost difference: $36,000/year Downtime Impact Calculation: Their ERP system downtime costs approximately $15,000/hour in lost productivity (manufacturing lines slow without real-time data, orders get delayed).

The difference between Tier II (22 hours/year) and Tier III (1.6 hours/year) is roughly 20.4 hours annually.

Expected downtime cost difference: 20.4 hours × $15,000 = $306,000/year The business case is clear: spend the extra $36,000 annually to avoid $306,000 in downtime.

But here's the nuance-if they can schedule Tier II maintenance during their annual two-week shutdown for equipment maintenance, the unplanned downtime difference shrinks dramatically.

Tier II's N+1 redundancy handles unexpected failures, and planned maintenance aligns with existing business cycles. Decision: They chose Tier II and negotiated extended UPS battery runtime to cover longer utility outages, effectively getting Tier II+ performance at Tier II pricing.

Example 2: Content Delivery Edge Node A streaming media company operates 200+ edge locations globally.

Each location serves cached video content to reduce latency and bandwidth costs to their origin servers.

They're evaluating infrastructure requirements for a new edge site in a secondary market. Requirements Analysis: Each edge node serves 10Gbps of traffic during peak hours.

Geographic redundancy means any single location can go offline-traffic automatically routes to the next nearest node.

User experience degrades slightly (added latency), but service continues.

For this use case, Tier I makes perfect sense:

• Single path power infrastructure (no N+1 UPS)
• Basic cooling (single CRAC unit with service contract for 4-hour response)
• No backup generator (UPS batteries provide clean shutdown capability)
• Capital cost: ~$250,000 vs. $600,000+ for Tier II The key insight: Application-layer redundancy (multiple geographic nodes) provides higher availability than infrastructure redundancy at a single site.

AWS uses this same philosophy-individual edge locations aren't Tier IV facilities, but CloudFront achieves high availability through distribution across hundreds of points of presence.

Common Misconceptions Misconception #1: "Tier II means twice as reliable as Tier I" The numbers don't work that way.

Going from 99.671% to 99.741% represents a reduction of 6.8 hours of downtime annually-meaningful but not transformative.

Both tiers still have significant downtime compared to Tier III (99.982%, 1.6 hours) or Tier IV (99.995%, 26 minutes).

Many decision-makers see "Tier II" and assume they're getting dramatically better availability.

The real value of Tier II isn't twice the uptime-it's protection against unplanned outages from single-component failures.

Your CRAC unit dies at 2 AM? Tier II keeps running; Tier I starts overheating.

The fundamental limitation remains: both Tier I and II have single distribution paths.

That's the architectural constraint that defines "basic capacity." Until you eliminate single points of failure in the distribution path itself (which requires concurrent maintainability), you're working within basic capacity parameters. Misconception #2: "Tier certifications are just marketing" Some operators claim "Tier III performance" without formal certification.

Real tier-certification requires Uptime Institute assessment of design documents, facility walkthrough, and operational validation.

The difference matters.

Equinix operates multiple certified Tier III facilities and can prove concurrent maintainability through documented procedures and physical infrastructure.

A facility claiming "Tier III design" without certification might have the right equipment but lack proper procedures, spare parts inventory, or staff training to actually deliver Tier III uptime.

Basic capacity facilities face less scrutiny precisely because their limitations are understood-single path means downtime during maintenance, period.

Summary & Key Takeaways

• Tier I provides 99.671% uptime (28.8 hours downtime/year) with single-path, non-redundant infrastructure-adequate for non-critical workloads, development environments, and edge locations where geographic distribution provides redundancy
• Tier II achieves 99.741% uptime (22 hours downtime/year) by adding n-plus-1 redundancy to capacity components (UPS, cooling, generators) while maintaining single distribution paths
• Both tiers require planned maintenance downtime-the single distribution path means no concurrent maintainability; equipment upgrades and repairs force partial or full facility shutdowns
• Application architecture often matters more than facility tier-hyperscalers achieve high availability by distributing workloads across multiple lower-tier facilities rather than concentrating in single high-tier sites
• Cost optimization drives tier selection-Tier I costs 40-60% less than Tier III to build; align facility reliability with actual business requirements rather than over-provisioning for theoretical needs
• Geographic and industry context influences tier choice-secondary markets, regional operations, and workloads with scheduled maintenance windows find basic capacity facilities economically efficient

Next Steps Explore the Tier III and IV lesson to understand how concurrent maintainability and fault tolerance eliminate the planned downtime limitations inherent in basic capacity facilities.

Understanding the full tier spectrum helps you make data-driven infrastructure decisions aligned with business requirements.

Study the tier-certification standards from the Uptime Institute to learn the specific design and operational requirements that separate actual certified facilities from operators making marketing claims.