Build infrastructure around your workload.

Websites, APIs & applications

Start with compute and storage needs, then select connectivity for your users.

Planning steps

1. Estimate memory, CPU, disk capacity and growth.
2. Check latency, loss and application requests from real user locations.
3. Plan backups, recovery and application monitoring; a cloud instance alone does not provide application high availability.

Start with NCS; compare NBM when dedicated hardware or greater resource isolation is required.

Databases & sustained workloads

Sustained I/O, the memory working set and recovery requirements matter alongside compute.

Planning steps

1. Confirm data size, read/write mix and storage requirements.
2. Define single-host performance, recovery and backup windows.
3. Confirm operating systems and licensing; availability needs additional architecture.

Compare NBM memory and storage, and plan backup and recovery separately.

Host equipment you own

Keep control of hardware while turning rack, power and network needs into a delivery checklist.

Planning steps

1. List equipment dimensions, weight, power and ports.
2. Confirm bandwidth billing, protection directions and additional power.
3. Agree receiving, installation, remote-hands scope and maintenance windows.

Start with colocation; choose a suitable cross-connect separately if required.

Facility & cloud connections

Select connections by endpoints and handoff interfaces, rather than the product name alone.

Planning steps

1. Identify A/Z locations, cloud platforms or facilities.
2. Confirm ports, bandwidth, Layer 2/3 handoff and addressing.
3. Check provisioning conditions, term, setup fees and redundancy.

Compare Cloud-VXC and cross-connect options; confirm endpoint coverage before quotation.

Availability-sensitive public services

Match protection scope to application protocols and exposure.

Planning steps

1. Record TCP/UDP, ports, source regions and baseline traffic.
2. Distinguish link congestion, protocol resource exhaustion and application request load.
3. Confirm mitigation, false-positive handling, blackholing and escalation.

Review NG Shield, then a matching NBM or colocation product; capacity alone is not a guarantee.

Hybrid cloud & staged migration

Map traffic between existing equipment and cloud resources before planning a hybrid deployment.

Planning steps

1. Identify services, data and dependencies that remain on premises.
2. Confirm endpoint connectivity, overlapping addresses, routing, encryption and access.
3. Migrate in stages with rollback criteria; validate data and connectivity during agreed windows.

Combine colocation, NCS/NBM and connectivity as a candidate architecture; confirm each public-cloud endpoint and handoff.

Backup, recovery & availability

Define acceptable data loss and downtime before choosing replication, backups and failover.

Planning steps

1. Define recovery point (RPO) and recovery time (RTO) objectives.
2. Avoid placing every copy in one failure domain; confirm available locations.
3. Exercise restoration and failover, recording failure conditions and rollback steps.

This is an architecture-planning scenario, not a claim that packages include managed backup, active-active failover or a defined SLA. Scope requires a separate design.