HCSP-Cloud Migration flashcards

35 cards. Read the term and the definition here. The deck at the top of this page is optional.

Rehost (lift-and-shift)

Moving an application to cloud infrastructure with no code changes, typically using block-level replication tooling. It is the fastest migration pattern but captures the least cloud-native benefit.

Relocate

Moving workloads to a vendor-managed platform without changing code or architecture, such as moving VMware virtual machines into a native cloud VMware service. It preserves existing operations tooling.

Migration factory

A standardized, repeatable execution pipeline of people, tooling, and runbooks that migrates workloads in waves. Throughput is measured in applications per wave rather than per project.

Refactor

Redesigning an application's architecture, often into services, to gain scalability and release agility. It delivers the highest cloud benefit but carries the highest cost, time, and risk.

Retain versus retire

Retain keeps a workload in place because no business case exists yet; retire decommissions it because it is unused or redundant. Both remove work from the migration backlog.

Mobilize phase

The migration phase between assessment and full-scale execution that builds the landing zone, selects pilot workloads, and establishes the migration factory before waves begin.

Hard dependency

A synchronous, latency-sensitive connection between workloads that must remain co-located or closely connected. Hard dependencies define move groups and constrain wave sequencing.

Move group

A set of workloads that must migrate together because of hard dependencies or shared infrastructure. Waves are composed of one or more move groups ordered by risk and value.

Right-sizing from utilization

Sizing target resources from observed peak utilization over a representative window, such as four weeks, rather than from provisioned capacity, then adding headroom for growth.

Landing zone

A pre-built cloud environment defining accounts, network topology, identity, logging, guardrails, and connectivity. It must be validated before the first migration wave executes.

Wave readiness gate

A formal checklist confirming landing zone validation, connectivity tests, approved runbooks, proven rollback, and stakeholder sign-off before a migration wave starts.

Application dependency mapping

Discovering network flows, ports, protocols, and data exchanges between applications to identify coupling. It combines automated discovery data with stakeholder interviews.

Agent-based replication

Continuous block-level copying of a server's volumes to the target environment, enabling test launches and short cutover windows with minimal source downtime.

Configuration externalization

Removing hard-coded hostnames, IP addresses, and file paths by moving them into parameter stores or service discovery. It is a prerequisite for scaling and portability.

Stateless-first sequencing

Migrating stateless application tiers before stateful tiers, since stateless components need no data synchronization and can be validated independently within a move group.

Repurchase

Replacing a self-hosted application with a SaaS product, which requires data export, integration rework, and identity federation planning rather than infrastructure migration.

Replatform

Making limited changes to move onto managed services, such as a self-managed database to a managed database or file shares to object storage, without altering core application code.

Change data capture

Continuously replicating ongoing inserts, updates, and deletes from a source database after an initial full load, enabling near-zero-downtime database migrations.

Replication lag at cutover

The delay between source and target writes. Cutover proceeds only when lag reaches zero and the source is quiesced, preventing lost or divergent data.

Heterogeneous database migration

Moving between different database engines, requiring schema conversion, data type mapping, and often rewriting application SQL and stored procedures.

Data validation checksums

Comparing hash totals, row counts, and sampled field values between source and target after loading. They detect truncation, encoding errors, and silent transformation defects.

Reverse replication rollback

A rollback path where changes made on the target after cutover are replicated back to the source, allowing a fast return without losing post-cutover transactions.

Dedicated interconnect versus VPN

A private dedicated connection offers higher bandwidth and predictable latency but has long provisioning lead times; an internet VPN is cheaper and faster to deploy with variable performance.

Non-overlapping CIDR design

Planning cloud address ranges that do not conflict with on-premises or other cloud networks. Overlap forces network address translation and complicates hybrid routing.

Landing zone guardrails

Organization policies, centralized logging, key management, network segmentation, and budget alerts applied at account or subscription level to enforce governance automatically.

BYOL licensing constraint

Bring-your-own-license terms may require dedicated hosts or bare-metal instances, which affects instance choice, cost model, and migration target selection.

Cutover runbook

An ordered task list with owners, timings, verification steps, communications, and explicit go/no-go decision points used to execute and validate a cutover.

Parallel run

Operating legacy and migrated systems simultaneously and comparing outputs. It is the safest cutover approach and the most expensive in cost and effort.

Rollback trigger threshold

A numeric criterion defined before cutover, such as an error-rate or replication-lag limit, that mandates reverting to the source system to avoid subjective decisions.

DNS TTL lowering

Reducing record time-to-live values days before cutover so clients pick up new addresses quickly, while retaining old records to support rollback.

Hypercare period

A temporary post-cutover window, typically one to four weeks, with elevated support staffing and daily triage. It ends when error rates stabilize and severe issues close.

FinOps inform-optimize-operate

The cost management cycle: inform through tagging and showback, optimize through rightsizing and commitment discounts, and operate through budgets, anomaly alerts, and reviews.

Rightsizing threshold rule

Sustained CPU and memory utilization below roughly twenty to forty percent over a review period marks a downsizing candidate; stateful systems require memory assessment before CPU.

Commitment discount strategy

Cover steady-state baseline load with reserved or committed-use purchases, leave variable load on-demand, and use spot or preemptible capacity for fault-tolerant batch work.

Policy-as-code guardrail

Governance rules expressed as machine-enforced policies that block or flag noncompliant resources, such as missing tags or public storage, without manual review.