Adopt-or-replace Adopt-or-replace is the choice ZopNight asks you to make when a resource already has its own scaling rules: adopt them (ZopNight only observes, changing nothing) or replace them… 4× Advisory category Advisory is the recommendation category for findings that are worth acting on but carry no dollar saving, because the item costs little or nothing to keep or the saving cannot be… 2× AI Gateway In general, a proxy between applications and AI model providers that controls keys, spending and which models can be used. ZopNight's AI Gateway gives an organisation virtual… 2× Alert fatigue What happens when people receive so many alerts that they start ignoring them, including the important ones. The usual fixes are fewer, better-targeted alerts and sending routine… 4× Allocation Assigning each cloud cost to the team, product, project or environment responsible for it, using tags, accounts or agreed rules for shared costs. It is the basis for showback,… 4× Amortized cost Cost with any upfront commitment payment spread evenly over the hours or days it covers, instead of counted on the day it was paid. It gives each covered resource a realistic… 2× Anomaly A cost change that is unusually large compared with its own recent pattern. ZopNight checks for anomalies once a day against a 7-day average and grades them by size: from 30%,… 1 API request charge A charge per request or operation rather than per hour of running time, used by services such as object storage reads and writes, serverless function calls and some databases.… 1 Apply / dismiss / snooze Apply, dismiss and snooze are the three ways to settle a recommendation. Apply means it has been acted on (through Remediate or by hand), dismiss means it does not fit your… 2× Approval gate The approval gate is the pause in a remediation where an admin must approve or reject the change before it reaches your cloud. It applies when the person who clicked Remediate is… 2× Assume-role An AWS mechanism where a trusted identity temporarily takes on an IAM role and receives short-lived credentials for it. ZopNight's recommended AWS connection works this way: the… 2× Audit log An audit log is a record of who changed what, when, and from where. ZopNight's audit log records every change made through its API, but not ordinary reads, and can be exported as a CSV. 5× Auto-escalation An incident and budget practice where an alert automatically moves to a more senior person or more urgent channel if it is not acknowledged or keeps recurring. ZopNight does not… 1 Auto-remediation Auto-remediation is ZopNight's built-in workflow that carries out a recommendation's fix when someone clicks Remediate: it checks preconditions, waits for admin approval if… 2× Avoidable spend Spend on resources that are idle, orphaned or larger than needed, which could be removed or reduced without hurting the service. It is an estimate of the upper limit on what could… 2× AWS Organizations The AWS service for managing many accounts together: grouping them into organizational units, applying policies across them and combining their bills into one. 2× Azure Hybrid Benefit An Azure licensing benefit that lets you reuse existing Windows Server or SQL Server licences (with Software Assurance or a qualifying subscription) on Azure, so you pay only for… 2× Azure Reservation Azure's commitment to a specific resource type, such as a VM size or database capacity, in a region for one or three years, in return for a discount. Reservations can be exchanged… 2×
Reference / Glossary
Every term,
in one place.
182 terms across 290 lessons, each with a definition pulled verbatim from the lesson that first uses it. Click any term to land there.
Bearer credential A credential that grants access to whoever presents it, with no further proof of identity, much like cash. API keys and personal access tokens are bearer credentials, so they must… 2× BigQuery billing export Google Cloud's way of delivering detailed billing data: cost and usage rows are written continuously into a BigQuery dataset you own, where they can be queried with SQL. Credits… 1 BilledCost The FOCUS billing column for the amount actually charged on the invoice, after discounts and credits. It is the figure to use when checking a cost report against the bill. 1 Blast radius Blast radius is how much else can break when one thing changes or fails. ZopNight shows it before you act on a recommendation, marking each directly connected resource as safe,… 6× Break-even coverage The level at which a commitment's discount on used capacity exactly pays for the capacity left unused. Below that usage level the commitment costs more than on-demand pricing… 1 Break-glass procedure A documented emergency route for gaining access, or making a change, when the normal controls are unavailable or not enough. It usually needs extra approval and is always logged… 2× Budget A budget is a monthly spending limit set on a cloud account, resource, resource group or AI provider. In ZopNight each budget has one optional alert threshold, shown as a… 2×
Carbon intensity The amount of carbon dioxide released per unit of electricity, usually grams of CO2 per kilowatt-hour. It depends on the grid's mix of energy sources, so it varies by region and… 2× Carbon-aware scheduling Running flexible workloads, such as batch jobs, at times or in regions where electricity has lower carbon intensity. 2× CDCR CDCR stands for Continuous Detect, Continuous Remediation: ZopNight's name for treating cost work as a loop that never stops, where the system keeps finding waste and fixing each… 2× Certified rule A certified rule is a recommendation rule on ZopNight's auto-remediation allowlist, meaning its fix is wired end to end and can run from the Remediate button, either in one click… 1 Chargeback Billing each team's cloud cost against that team's own budget, so it competes with their other spending. It creates strong accountability but needs accurate, agreed allocation,… 3× Cloud account A cloud account is a connection from ZopNight to one AWS account, GCP project, Azure subscription or similar. Each is read-only (discover, report and recommend) or read and write… 0 Cluster autoscaler The cluster autoscaler is the Kubernetes component that adds nodes (servers) when workloads cannot fit and removes nodes that are underused. Scaling workloads down saves money… 1 Cold-start latency The delay before a resource that was stopped or scaled to zero can serve requests again, such as a function, container or database starting up. It is the trade-off for the savings… 2× Commitment portfolio The full set of discount commitments an organisation holds, such as Reserved Instances, Savings Plans and Committed Use Discounts, managed together for coverage, term and expiry.… 1 Committed Use Discount Google Cloud's commitment discount: you commit to a level of resources or spend for one or three years and get lower rates. Resource-based commitments cover specific machine… 3× Compromised credentials Passwords, keys or tokens that have fallen into the hands of someone they were not issued to. In cloud cost terms they often show up as a sudden spend spike in unusual regions or… 2× Concrete-or-abstain Concrete-or-abstain is the principle that a recommendation either names a specific action with a saving priced from real data, or says nothing. ZopNight shows savings as unknown,… 3× Confidence band A range around a forecast showing where the actual figure is likely to fall, rather than a single number. Wider bands mean more uncertainty. ZopNight's cost forecasts are single… 4× Cost Explorer AWS's built-in console tool and API for viewing, filtering and grouping cost over time. It keeps a limited history (about a year by default), and resource-level detail covers only… 1 Cost flow Cost flow is a diagram showing how spend splits across four groupings at once, for example provider to account to resource type to team, with band widths in proportion to cost.… 2× Cost ownership The principle that the team that runs a resource is accountable for its cost, while central FinOps and finance supply the data and the process. Ownership is usually recorded… 3× Cost-per-X A unit cost metric made by dividing cloud spend by a business count X, such as orders, requests or tenants. It is the working number behind unit economics. 2× Coverage The share of eligible usage that is paid for at a commitment rate rather than the on-demand price. It is read together with utilization. 3× CrashLoopBackOff CrashLoopBackOff is the Kubernetes status of a container that keeps crashing and restarting, with Kubernetes waiting longer between each attempt. The capacity reserved for it… 1 Crawl-Walk-Run The FinOps Foundation's maturity model. Crawl means basic visibility and reactive fixes, Walk means regular optimization with some processes in place, and Run means cost… 4× Credential rotation Replacing a password, key or token with a new one on a schedule or after a suspected leak. A safe order is to create the new one, switch clients to it, confirm it works, then… 2× Cron expression A cron expression is a short five-field pattern (minute, hour, day of month, month, day of week) that describes when something repeats; for example 0 9 * * 1-5 means 9am every… 2× CronJob suspend Suspending a CronJob tells Kubernetes to stop starting new runs of a scheduled job without deleting it. ZopNight suspends CronJobs when it stops a namespace, and flags CronJobs… 2× Cross-AZ traffic Data moving between availability zones (separate data centres) in the same region. AWS charges about $0.01 per GB in each direction, GCP about $0.01 per GiB, and Azure has not… 2× CUR AWS's Cost and Usage Report: the most detailed AWS billing file, with one line per resource per usage type per hour or day, delivered to an S3 bucket. CUR 2.0 is its current… 2× Custom role A custom role is a role an admin builds by choosing individual permissions, optionally limited to specific resources, for needs the three system roles do not fit. Some actions… 2×
Data residency A requirement, from law, contract or company policy, that certain data be stored and processed within a particular country or region. 1 Database protection Database protection is ZopNight's rule that a remediation never stops, deletes or resizes a database. It is enforced by what each kind of fix is allowed to touch: databases are on… 2× Deployment A deployment is a released, running version of something. In ZopCloud it is a build of your service shipped to an environment, which you can redeploy or roll back. In the AI… 7× Derived tag A derived tag is a tag value calculated from facts ZopNight already knows about a resource, such as its provider, region, type, instance type or name. It stays pending until… 3× Discovery cron The discovery cron is the timed job that re-reads each connected cloud account's inventory. In ZopNight it runs every six hours, so a resource created just after a run can take up… 2× Driver-based forecasting Forecasting cloud spend from the business drivers that cause it, such as users, orders or transactions, rather than only extending the past cost trend. It is often combined with… 4×
Effective discount The discount an organisation actually achieves, as opposed to the headline rate. It depends on how much usage the commitments cover and how fully they are used, and is often well… 6× EffectiveCost The FOCUS billing column for cost after all discounts, with upfront commitment fees spread across the period they cover. It is the column to use when reporting what each resource… 1 Egress Data leaving a cloud network, for example to the internet, to another region or to another provider. Providers charge for it per gigabyte, and the charge often appears as a… 4× Enterprise Discount Program A private agreement with AWS that gives a percentage discount on most spend in return for a committed minimum spend over several years. Its terms are confidential and usually… 1 Escalation chain A defined sequence of who is notified, and when, as a problem becomes more serious, such as budget checkpoints at 50, 75, 90 and 100% each reaching a different audience. A… 1 Event readiness Event readiness is planning capacity for a known traffic peak, such as a sale or launch: scaling up beforehand and back down afterwards. ZopNight does both automatically for… 2× Exclusive membership Exclusive membership is ZopNight's rule that a resource can belong to only one resource group at a time. Together with the rule that a resource or group can be attached to only… 2×
Fail closed A design rule that when a system cannot be sure an action is allowed, it refuses rather than permits. ZopNight's gateway refuses requests until its permission table has loaded,… 3× FinOps Foundation A vendor-neutral industry body, part of the Linux Foundation, that publishes the FinOps Framework, the maturity model and the FOCUS billing standard. Its vocabulary is widely used… 2× FinOps lifecycle The FinOps Foundation's three repeating phases: Inform (see and allocate spend), Optimize (act to reduce waste and improve rates) and Operate (build the ongoing habits and… 4× FOCUS The FinOps Open Cost and Usage Specification: an open standard, published by the FinOps Foundation, that gives every cloud's billing data the same column names and meanings. It… 1 Forecast A projection of future cloud spend based on past usage and expected changes. Unlike a budget, it is an estimate, not a limit or a commitment. 2× Forecast accuracy How close a forecast came to the actual spend, measured after the period ends, often as a percentage error. Tracking it over time shows whether forecasting is improving. 3× FX rate The exchange rate used to convert a cost from one currency to another. Good practice is to use the rate from the date of each charge, so past reports do not shift when currencies… 1
Gateway enforcement Making every access decision at the single entry point that all requests pass through, instead of in each service behind it. ZopNight's gateway checks sign-in, role permissions,… 2× Group attachment Group attachment means attaching a schedule to a resource group rather than to each resource, so every member follows it and new members pick it up automatically. A resource that… 2×
Hallucination When an AI model states something confidently that is not true, such as inventing a cost figure instead of fetching it from a tool. It is most likely when the model answers from… 1 HPA (Horizontal Pod Autoscaler) The Horizontal Pod Autoscaler is the Kubernetes feature that adds or removes copies (replicas) of a workload based on a metric such as CPU use, within set minimum and maximum… 1
IaC drift IaC drift is the gap that opens when someone changes infrastructure directly, for example in the cloud console, instead of through its infrastructure-as-code files. The code then… 1 IaC validation An IaC validation is one check of infrastructure-as-code (Terraform or OpenTofu) against your policies, run on a planned change or on current state. ZopNight returns a pass or… 0 IAM role An AWS identity with a set of permissions that people or services take on temporarily, rather than a user with permanent passwords or keys. A third party is usually given access… 1 Idempotency The property of an operation that has the same result whether it runs once or many times, such as stopping an instance that is already stopped. It makes retries after a timeout or… 4× Identity Provider The system that holds a company's user accounts and confirms who someone is when they sign in, such as Okta, Microsoft Entra ID or Google Workspace. Other apps trust its answer… 1 Infracost An open-source tool that estimates the monthly cost of infrastructure-as-code changes, such as Terraform, and can post the estimate on a pull request. 2×
Least privilege The principle of giving each person, service or token only the access it needs for its job, and nothing more. ZopNight's cloud connections offer a read-only level for this reason:… 0 Lifecycle policy A rule that automatically moves stored data to cheaper storage tiers or deletes it after a set age, for example removing snapshots older than 90 days. Object storage and backup… 2×
Management Group An Azure container that groups subscriptions so policies and access can be applied to all of them at once. Management groups can be nested to mirror an organisation's structure. 2× Manual refresh A manual refresh is an on-demand discovery run for one cloud account, started by a person instead of waiting for the six-hourly cycle. It also re-checks permissions, so a newly… 2× MCP Model Context Protocol: an open standard that lets an AI assistant (such as Claude, Cursor or Codex) call tools on another service and get structured answers back. ZopNight runs… 8× MCP audit log ZopNight's record of every MCP tool call, reads included: who called, which tool, the request parameters and whether it succeeded. The response data is not kept, but a failed call… 3× MCP configuration The client-side settings that tell an AI app where an MCP server lives and how to sign in to it. For ZopNight this means adding the server address plus a personal access token to… 4× MCP-native Describes a product whose data and actions are reachable directly from an AI assistant over MCP, rather than only through its own web dashboard. Many tools now offer an MCP… 3× Minimum viable tag set The smallest set of required tags that still lets costs be assigned, commonly team, environment, cost centre and owner. It is kept small so that people actually apply it. 3× Model allow-list The list of AI models a given key is permitted to call; a request for any other model is refused. In ZopNight a virtual key can be limited to named models or left open to all… 1 Multi-account Running cloud workloads across many separate accounts (AWS), projects (GCP) or subscriptions (Azure) instead of one shared one, usually split by team, environment or product. It… 2×
Namespace boundary A namespace is a named section of a Kubernetes cluster that keeps one team's or service's workloads apart from others. ZopNight can stop and start a whole namespace, skipping… 1 NAT Gateway A managed service that lets resources in a private network reach the internet without being reachable from it. On AWS it bills per hour per gateway plus per gigabyte of data processed. 3× Needs Attention Needs Attention is the list in ZopNight's cloud IAM import that holds cloud roles it could not translate into ZopNight permissions with confidence. They are created with no… 2× Net of credits A cost figure with provider credits, such as promotional credits and automatic discounts, already subtracted. Google Cloud's billing export lists credits separately, so ZopNight… 1 Notification channel A notification channel is a destination where ZopNight sends alerts: Slack, Microsoft Teams, Google Chat, or any web address that accepts a webhook. Subscriptions decide which… 2×
OAuth 2.0 An open standard for granting an app limited access to an account without sharing its password. Strictly it handles authorization; sign-in is usually built on it through OpenID… 2× Operate cadence The regular rhythm, such as weekly, monthly or quarterly, at which teams review cost data, act on findings and check results. Data may refresh daily, but decisions happen on the… 5× Org-bound token A ZopNight credential tied to one specific organisation: a signed-in session, or a personal access token scoped to that organisation. Every MCP write tier needs one; a token not… 2× Orphan resource A resource that is still running and billing but no longer serves any purpose or has no clear owner, such as an unattached disk, an old snapshot or a forgotten test server. It… 2× Over-commitment Buying more commitment than the workload will steadily use, so part of it bills for capacity nobody consumes. It usually follows a migration, a re-architecture or a forecast that… 3× Override An override is a temporary exception to a schedule: it keeps a resource or group running, or keeps it stopped, regardless of the schedule, until an expiry time you must set. When… 6×
PAT A personal access token: a secret string a user creates to let a script or AI assistant act as them. In ZopNight a PAT can never exceed its creator's current role, can be limited… 9× Permission visibility Permission visibility is ZopNight's per-account view of which cloud permissions it has, marked Granted, Denied or Unknown, with the error for anything denied. Every discovery run… 2× Pod-level attribution Pod-level attribution means splitting a Kubernetes cluster's cost across the workloads running on it, usually by the CPU and memory each requests, then rolling that up by… 1 Policy entity A policy entity is one kind of thing ZopNight controls access to, such as resource, schedule, budget or recommendation. Each permission pairs an entity with an action (view,… 2× Pre-merge cost estimation Estimating what an infrastructure change will cost before it is merged, and showing the figure, for example as a pull request comment, alongside the code being reviewed. 1 Preset A preset is a ready-made starting configuration you can use as is or adjust. In ZopNight the schedule form offers four schedule templates (such as Business Hours, weekdays 9am to… 2× Preventive enforcement Stopping a non-compliant resource from being created in the first place, for example by a policy that blocks deployment of untagged resources, rather than finding and fixing it later. 2× PrivateLink An AWS service that connects a private network to AWS services or other networks through private endpoints, without going through the internet or a NAT Gateway. Azure Private Link… 2× Privilege escalation Gaining more access than you were given, for example by using one permission to grant yourself others. A common defence is that a delegated credential can never exceed its owner:… 1 Prompt injection An attack where someone hides instructions in content an AI agent reads (a resource tag, a ticket, a web page) to make it do something its user did not ask for. With ZopNight's… 2× Provisioned throughput Reserved AI model capacity (for example on AWS Bedrock) bought in fixed units and billed by the hour whether you use it or not. It guarantees capacity and lowers the effective… 2× PVC (PersistentVolumeClaim) A PersistentVolumeClaim is a Kubernetes request for storage that is bound to a PersistentVolume, the actual disk. The disk keeps costing money even when no workload uses it, so… 1
Rack rate The public, undiscounted price a cloud provider lists for a service, before any commitment, negotiated or usage discount. ZopNight calculates a rack-rate cost for every discovered… 3× RBAC RBAC (role-based access control) means giving people permissions through roles rather than one by one. In ZopNight a role is a set of permissions, each an action on one of 38… 1 Realized savings Savings confirmed as actually achieved, measured from real spend after a change, rather than estimated beforehand. The share of estimated savings that becomes realized is the… 4× Recommendation A recommendation is a specific suggestion to change a cloud resource, such as stopping, resizing or deleting it, produced when a rule finds waste or risk. In ZopNight each one… 0 Recommendation card A recommendation card is how ZopNight presents one finding: the resource, what was found, the evidence behind it, the current and expected cost, the potential saving, and the… 2× Reconciliation Checking that a cost figure matches an independent source, usually the provider's invoice, and explaining any gap. A report that reconciles can be trusted for budgets and chargeback. 3× Redaction Removing or masking sensitive values, such as secrets or tokens, before data is stored or shown. ZopNight's audit log replaces declared secret values with [REDACTED] before… 2× Reserved Instance An AWS commitment to pay for a specific type of capacity, such as an instance family in a region, for one or three years in return for a discount on the hourly rate. It bills… 2× Resource group A resource group is a named set of cloud resources managed together in ZopNight. A schedule, override or budget applied to the group covers every member, and adding or removing a… 3× Resource request A resource request is the CPU and memory a Kubernetes container says it needs, which the cluster reserves for it whether used or not; a limit is the most it may use. Requests… 1 Right-sizing Changing a resource's size, such as CPU, memory or instance type, to match what it actually uses. In Kubernetes it usually means adjusting pod resource requests. 2× RTO / RPO Recovery Time Objective is how quickly a service must be restored after a failure. Recovery Point Objective is how much recent data can be lost, measured in time. Together they… 2× Rule A rule is an automated check that looks at a resource's settings, usage or cost and creates a recommendation when it finds a problem, such as a disk with no server attached.… 0 Rule category A rule category is the label that groups recommendation rules by the kind of action they call for. ZopNight's 684 rules fall into 11 categories: idle, rightsizing, orphan,… 1
Safety Gate The Safety Gate is ZopNight's per-resource check that decides whether a fix may run automatically on this particular resource, beyond whether the rule supports it at all. If it… 3× SAML Security Assertion Markup Language: an open standard that lets a company's identity provider (such as Okta or Microsoft Entra ID) vouch for a user so they can sign in to other… 2× Savings decay The gradual loss of savings after an optimization effort, as resources are re-created, schedules are removed, sizes creep back up or commitments lapse. It is the usual result of… 1 Savings Plan An AWS commitment to spend a fixed amount per hour on compute for one or three years, in return for discounted rates. Unlike a Reserved Instance, it applies across instance types… 2× Scale to zero Scaling to zero means reducing something to no running copies while keeping its configuration so it can come back, for example setting a Kubernetes Deployment to zero replicas.… 3× Scaling policy A scaling policy tells a cloud autoscaler when to add or remove capacity, based on a metric such as CPU or a time of day, within set minimum and maximum sizes. ZopNight can create… 1 Schedulable resource A schedulable resource is one whose type ZopNight knows how to stop and start, such as virtual machines, Kubernetes node pools and namespaces, or Snowflake warehouses. Only these… 3× Schedule A schedule is a set of start and stop times, written as cron expressions in a chosen timezone, that ZopNight uses to switch attached resources on and off automatically. It is… 7× Scoped token An access token limited to a narrow set of actions or resources rather than everything its owner can do. Examples in ZopNight include the scan token created for… 2× Sequenced execution Sequenced execution starts and stops a group's resources in a set order rather than all at once. By tier, ZopNight starts storage, then databases, then compute, then applications,… 4× Service account An identity for software rather than a person, used by an application or integration to call cloud APIs. In GCP, ZopNight's recommended one-click connection creates a dedicated… 3× Service Control Policy An AWS Organizations policy that sets the maximum permissions for every account under an organizational unit or the whole organization. It never grants access on its own; it only… 1 Service principal Azure's identity for an application, created from an app registration, which software uses to sign in and call Azure APIs. ZopNight's recommended Azure connection is a service… 2× Severity Severity is the urgency level attached to a finding or alert. ZopNight recommendations carry one of five levels (critical, high, medium, low, info), while notifications carry one… 2× Shared services account A cloud account that holds infrastructure used by several teams, such as networking, logging or CI, usually run by a platform team. Its cost is then reported centrally or… 2× Shared-services allocation Dividing the cost of infrastructure several teams use, such as networking, logging or a shared database, among those teams by an agreed rule: an equal split, a usage-based share… 2× Showback Reporting each team's cloud cost back to that team without moving money between budgets. It creates awareness and ownership with little administrative friction. ZopNight's Cost… 3× Skill (Claude Code) A saved, reusable set of instructions that Claude Code loads by name, stored as a folder containing a SKILL.md file, either per user or committed to a project. Teams use skills to… 3× SLA A service level agreement: a promised level of availability or performance, such as 99.9% uptime, often with penalties if missed. It sets the limit on how far a service can be cut… 2× Smart Tag A Smart Tag is a tag ZopNight works out for a resource from a tagging policy instead of someone typing it in. Once you accept it, it counts for cost reporting inside ZopNight, and… 3× SOC 2 An independent audit report, defined by the AICPA, on how well a service company controls security and related areas such as availability and confidentiality. A Type II report… 1 Spot Spare cloud capacity sold at a large discount, often 60 to 90% off on-demand, that the provider can reclaim at short notice (about two minutes on AWS). It suits work that can be… 2× SQL Warehouse Databricks' compute resource for running SQL queries and dashboards. It has its own auto-stop setting for idle time, and ZopNight schedules can also start and stop it at set times. 2× SSRF guard A check that stops a server from being tricked into calling internal addresses (server-side request forgery). Before ZopNight fetches from a customer-supplied URL, it resolves the… 2× StatefulSet A StatefulSet is a Kubernetes workload whose copies keep a fixed identity and their own storage, used for things like databases and queues. Scaling it to zero stops the copies but… 1 Stranded commitment A commitment that no longer matches any running usage, for example after a workload moves to a different instance family or region. It keeps billing until it is exchanged, sold… 1 Sustained use discount A discount Google Cloud applies automatically to certain Compute Engine VMs that run for a large part of the month. It needs no commitment and shows as a credit in the billing data. 2× System role A system role is one of ZopNight's three built-in roles, which cannot be edited: Admin (everything), Editor (create and change most things, view-only on administration) and Viewer… 1
Tabletop exercise A discussion-based rehearsal where a team walks through a realistic incident scenario, such as a sudden cost spike, step by step to test who does what and find gaps in the plan,… 2× Tag A key and value pair attached to a cloud resource, such as team=payments, used to record ownership, environment or purpose. GCP and Kubernetes call them labels. Cost tools use… 1 Tag coverage The share of resources, or of spend, that carries the required tags. Measuring it per tag key shows which tag is missing. ZopNight's Cost Reports show the share of cost on tagged… 4× Tag drift When tags on live resources change, disappear or no longer match the agreed standard, often because someone edited them outside the normal deployment process. The rate of drift… 2× Tag inheritance Tags set at a higher level, such as an account, project, resource group or stack, being applied automatically to the resources beneath it. Support varies by provider and service. 1 Tag taxonomy The written standard listing which tag keys are allowed and what values each may take, for example env must be one of prod, staging or dev. It prevents the same meaning being… 3× Tagging policy A tagging policy is a rule that says which tag key resources should carry and how to work out its value from the resource's own details, for example setting Application from a… 2× Tags-as-code Defining tags and tag rules in version-controlled infrastructure code, for example as default tags applied by a Terraform provider, so every new resource gets them automatically… 3× Team A team in ZopNight is a group of people with resources assigned to it. Teams are used to split costs between owners and to limit what members can see and change to their own resources. 0 Team scope Team scope limits a person's role to the resources assigned to their team. The role says what they can do; the team scope says which resources they can do it to. 2× Theoretical savings The total of every savings estimate before overlaps are removed and before checking which fixes will actually be applied. It is always larger than what will be achieved. 1 Tool call One request an AI assistant makes to a tool, such as asking an MCP server to list resources or start a schedule, with the tool's name and its inputs. In ZopNight each tool call is… 1
Unattributed spend Cloud charges that cannot be tied to a specific resource or owner, such as data transfer, taxes, support fees or untagged resources. ZopNight includes billing-only charges in… 4× Unblended cost The cost of each usage line at the rate actually charged when it happened, with no averaging across accounts and no spreading of upfront fees. By default, ZopNight's AWS billing… 3× Unit economics Cloud cost expressed per unit of business output, such as cost per order, per active user or per customer. ZopNight can draw a cost-per-unit line on the Cost Reports trend chart… 3× Unit metric A unit metric is a business number, such as orders, active users or API requests, that cloud cost is divided by to give a cost per unit. In ZopNight you supply the daily values by… 2× Utilization How much of a purchased commitment is actually used during its term, as a percentage. Unused commitment is still billed. In capacity planning, utilization also means how much of a… 2×
Validation Validation is the last step of a ZopNight remediation: after the cloud change, it reads the resource's live state and checks that it matches what the fix intended. If it does not… 1 Variance analysis Comparing actual spend with the budget or forecast for a period and explaining the difference: which teams, services or events caused it and whether it will happen again. 3× Virtual key An API key issued by an AI gateway instead of by the model vendor, so the gateway can enforce a spending cap and a list of allowed models. In ZopNight virtual keys belong to the… 3× Virtual warehouse Snowflake's unit of compute: a cluster that runs queries and uses credits only while running. It can suspend automatically when idle. ZopNight can schedule Snowflake warehouses to… 2×
Watch policy A watch policy is a recommendation rule you write yourself in ZopNight: choose which resources it covers, the metric conditions to watch (for example average CPU below 5% over a… 1 Workload identity federation A way for software in one system to prove its identity to another cloud and receive short-lived credentials, instead of storing a long-lived key. ZopNight uses it to assume… 3× Write tier A per-organisation setting in ZopNight that caps which kinds of changes AI agents may make over MCP: none (the default), 1 metadata-only, 2 reversible, or 3 irreversible. It is a… 4×
·
No term matches that filter.