Parquet planning columns

Every column of the fourteen planning files in a Parquet snapshot — the schedule as posted — with its physical type, whether it can be empty, and what it holds. It is a lookup table for data engineers and report developers who are already reading the export and need to know what a column means; for how snapshots are laid out and how to tell a finished one from a partial write, see Parquet export.

Contents

• activity.parquet
• resource.parquet
• calendar.parquet
• successor.parquet
• structure.parquet
• structureElement.parquet
• activityStructureElement.parquet
• resourceAssignment.parquet
• resourceAvailability.parquet
• calendarOperation.parquet
• weeklyRepeatingPeriod.parquet
• connectedPeriod.parquet
• profile.parquet
• profilePoint.parquet

How to read these tables

Each section lists the fixed columns of one file: the columns every snapshot of that file has, on every deployment. Metadata columns are extra, differ per customer, and are covered separately below.

Every column is physically nullable. The writer emits an optional Parquet field for all of them, so the file's schema marks nothing as required. The Empty? column therefore describes the data rather than the schema:

• never — the value is a non-optional part of the entity and every row carries one.
• possible — the value is genuinely optional, and a consumer has to handle null.

A consumer that tries to read a non-null constraint out of the schema will not find one.

Column order is not a contract. Columns are discovered by reflection over the domain model, most-derived declarations first and inherited ones last — which is why Id and Created sit at the end of almost every table rather than the start. These tables list them in the order the file actually stores them, but that order can shift when the model changes. Address columns by name.

Enum columns carry names, not numbers. Where a column is an enum its permitted values are listed under the table. This is worth checking if you are moving a query over from the REST export, which serves the same values as integers alongside a separate …Text field: compare a Parquet enum column against a number and you get an empty result with no error to tell you why.

Metadata columns

Alongside the fixed columns above, a file can carry one column per metadata field. Only three planning files do: activity.parquet, resource.parquet and resourceAssignment.parquet. Every other planning file has fixed columns only, and each section below says so.

The set is not driven by your schedule type configuration. It is driven by the values that are actually there: the export collects the distinct metadata fields that have a value on the rows in scope, so a field configured for your deployment but never populated produces no column at all. Three consequences worth planning for:

• The column set differs between customers, between two schedules of one customer, and between two cut-offs of the same schedule. A field populated for the first time appears as a new column in the next snapshot only, and earlier snapshots are not rewritten to gain it.

• The scope differs per file, so two files in one snapshot can disagree about which fields exist:

  File	Carries the fields with values on…
  activity.parquet	activities of this planning schedule
  resource.parquet	resources of the schedule's data set
  resourceAssignment.parquet	resource assignments of this schedule's activities

• Numeric types collapse. A metadata field's numeric value is stored in a single numeric column whatever its declared type, so integer, decimal and floating-point fields all arrive as double. Boolean fields arrive as boolean, date fields as timestamp, and everything else as string.

A metadata column's name is not its identity. Names are sanitised and de-duplicated against the names already taken, so a metadata field can arrive under a suffixed name — Description_2 rather than Description. No two fixed columns collide anywhere in the export, so a _2 or _3 suffix is always a metadata field that lost a name race, either against a fixed column or against a metadata field resolved before it. The suffixes are handed out in order, so adding a field upstream can move an existing column from _2 to _3.

Resolve metadata columns through _manifest.json, which maps each physical column back to its IlapId, before combining snapshots or comparing deployments. Parquet export covers the manifest in full.

activity.parquet

The activities of the schedule as posted — dates, durations, work hours and progress. One row per activity.

Column	Type	Empty?	Meaning
ScheduleId	int32	never	The planning schedule this activity belongs to.
CalendarId	int32	possible	The working calendar the activity is scheduled against, into calendar.parquet.
ActivityType	string	never	What kind of activity this is — see the values below the table.
Description	string	possible	Free-text description of the activity.
FinishAsEarlyAsPossible	boolean	never	Whether the activity is constrained to finish as early as the logic allows.
FinishNoLaterThan	timestamp	possible	Constraint date: the activity must not finish after this.
MustFinishOn	timestamp	possible	Constraint date: the activity is pinned to finish on this date.
MustStartOn	timestamp	possible	Constraint date: the activity is pinned to start on this date.
StartAsEarlyAsPossible	boolean	never	Whether the activity is constrained to start as early as the logic allows.
StartAsLateAsPossible	boolean	never	Whether the activity is constrained to start as late as the logic allows.
StartNoEarlierThan	timestamp	possible	Constraint date: the activity must not start before this.
StartNoLaterThan	timestamp	possible	Constraint date: the activity must not start after this.
ActualFinish	timestamp	possible	When the activity actually finished. Empty until it has.
ActualStart	timestamp	possible	When the activity actually started. Empty until it has.
ActualWorkHours	double	never	Work hours expended on the activity so far.
CurrentProgress	double	never	Progress reported on the activity, as a percentage — 20 means 20%.
EarlyStart	timestamp	possible	Earliest the activity can start under the schedule logic.
EarlyFinish	timestamp	possible	Earliest the activity can finish under the schedule logic.
LateFinish	timestamp	possible	Latest the activity can finish without delaying the schedule.
LateStart	timestamp	possible	Latest the activity can start without delaying the schedule.
FreeFloatHours	double	never	Hours the activity can slip before it delays a successor.
TotalFloatHours	double	never	Hours the activity can slip before it delays the end of the schedule.
PlannedWorkHours	double	never	Work hours planned for the activity.
RemainingWorkHours	double	never	Work hours still to be expended on the activity.
IsCancelled	boolean	possible	Whether the activity has been cancelled.
CancelledDate	timestamp	possible	When the activity was cancelled. Empty unless it was.
PlannedProgress	double	never	Progress the activity was planned to have reached, as a percentage — 20 means 20%.
FinishOnOrAfter	timestamp	possible	Constraint date: the activity must not finish before this.
AlwaysOnSchedule	boolean	never	The ILAP IsAlwaysOnSchedule flag as imported.
DurationHours	double	possible	Planned duration of the activity in hours.
RemainingDurationHours	double	possible	Duration still remaining, in hours.
CalendarRef	string	possible	The IlapId of the calendar in CalendarId. Use this to join across snapshots.
Label	string	possible	Human-readable name of the activity.
LocalSystemId	string	possible	Identifier of the activity in the planning system it was imported from.
GlobalSystemId	string	possible	Identifier of the activity in the system of record it originates from.
Code	string	possible	Short code identifying the activity in the source planning system.
IlapId	string	possible	Stable ILAP identifier for the activity, carried across imports and deployments.
BaseUri	string	possible	Base URI recorded alongside the ILAP identifier when the activity was imported.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

ActivityType is one of: NotSet, RegularActivity, MilestoneStart, MilestoneFinish, Hammock, TaskDependent, ResourceDependent, LevelofEffort, WBSSummary.

This file carries metadata columns. See Metadata columns.

resource.parquet

The resources available to the schedule. One row per resource.

Column	Type	Empty?	Meaning
DataId	int32	never	The imported data set this resource belongs to. Resources are scoped to the data set rather than to one schedule.
Description	string	possible	Free-text description of the resource.
Label	string	possible	Human-readable name of the resource.
LocalSystemId	string	possible	Identifier of the resource in the planning system it was imported from.
GlobalSystemId	string	possible	Identifier of the resource in the system of record it originates from.
Code	string	possible	Short code identifying the resource in the source planning system.
IlapId	string	possible	Stable ILAP identifier for the resource, carried across imports and deployments.
BaseUri	string	possible	Base URI recorded alongside the ILAP identifier when the resource was imported.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

This file carries metadata columns. See Metadata columns.

calendar.parquet

The working calendars activities are scheduled against. One row per calendar; the rules that make it up are in calendarOperation.parquet.

Column	Type	Empty?	Meaning
DataId	int32	never	The imported data set this calendar belongs to.
HoursPerDay	double	never	Working hours in one calendar day, used to convert between hours and days.
Start	timestamp	never	Start of the period the calendar covers.
Finish	timestamp	never	End of the period the calendar covers.
Description	string	possible	Free-text description of the calendar.
Label	string	possible	Human-readable name of the calendar.
LocalSystemId	string	possible	Identifier of the calendar in the planning system it was imported from.
GlobalSystemId	string	possible	Identifier of the calendar in the system of record it originates from.
Code	string	possible	Short code identifying the calendar in the source planning system.
IlapId	string	possible	Stable ILAP identifier for the calendar, carried across imports and deployments.
BaseUri	string	possible	Base URI recorded alongside the ILAP identifier when the calendar was imported.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

This file carries no metadata columns.

successor.parquet

The links between activities and the lag on each. One row per link, held against its predecessor activity.

Column	Type	Empty?	Meaning
ActivityId	int32	never	The predecessor activity the link runs from, into activity.parquet.
SuccessorObjectId	int32	possible	The successor activity the link runs to, into activity.parquet.
CalendarId	int32	possible	The calendar the lag is measured against, into calendar.parquet.
Type	string	never	Which ends of the two activities the link joins — see the values below the table.
LagHours	double	never	Lag on the link, in hours. Negative values are leads.
SuccessorRef	string	possible	The IlapId of the successor activity in SuccessorObjectId. Use this to join across snapshots.
CalendarRef	string	possible	The IlapId of the calendar in CalendarId. Use this to join across snapshots.
IlapId	string	possible	Stable ILAP identifier for the link, carried across imports and deployments.
BaseUri	string	possible	Base URI recorded alongside the ILAP identifier when the link was imported.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

Type is one of: StartToStart, StartToFinish, FinishToStart, FinishToFinish.

This file carries no metadata columns.

structure.parquet

The breakdown structures defined over the schedule — the WBS and any other activity hierarchy. One row per structure; its nodes are in structureElement.parquet.

Column	Type	Empty?	Meaning
DataId	int32	never	The imported data set this structure belongs to.
Description	string	possible	Free-text description of the structure.
Type	string	never	Whether the structure is a WBS, a flat list or a hierarchical list — see the values below the table.
Label	string	possible	Human-readable name of the structure.
LocalSystemId	string	possible	Identifier of the structure in the planning system it was imported from.
GlobalSystemId	string	possible	Identifier of the structure in the system of record it originates from.
Code	string	possible	Short code identifying the structure in the source planning system.
IlapId	string	possible	Stable ILAP identifier for the structure, carried across imports and deployments.
BaseUri	string	possible	Base URI recorded alongside the ILAP identifier when the structure was imported.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

Type is one of: Wbs, FlatList, HierarchicalList.

This file carries no metadata columns.

structureElement.parquet

The nodes within a breakdown structure, each pointing at its parent. One row per node; walk ParentElementId to rebuild the hierarchy.

Column	Type	Empty?	Meaning
StructureId	int32	never	The structure this node belongs to, into structure.parquet.
Description	string	possible	Free-text description of the node.
ParentElementIlapId	string	possible	The IlapId of the parent node. Use this to rebuild the hierarchy across snapshots.
ParentElementId	int32	possible	The parent node within this snapshot. Empty at the root of the structure.
Label	string	possible	Human-readable name of the node.
LocalSystemId	string	possible	Identifier of the node in the planning system it was imported from.
GlobalSystemId	string	possible	Identifier of the node in the system of record it originates from.
Code	string	possible	Short code identifying the node in the source planning system.
IlapId	string	possible	Stable ILAP identifier for the node, carried across imports and deployments.
BaseUri	string	possible	Base URI recorded alongside the ILAP identifier when the node was imported.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

This file carries no metadata columns.

activityStructureElement.parquet

Which activities hang off which structure nodes. One row per membership — the join table between activity.parquet and structureElement.parquet.

Column	Type	Empty?	Meaning
ActivityId	int32	never	The activity, into activity.parquet.
StructureElementId	int32	never	The structure node it hangs off, into structureElement.parquet.

This file carries no rows of its own beyond the pair above — no Id and no Created, unlike every other planning file.

This file carries no metadata columns.

resourceAssignment.parquet

The resources assigned to activities, with the quantities involved. One row per assignment.

Column	Type	Empty?	Meaning
ActivityId	int32	never	The activity the resource is assigned to, into activity.parquet.
ProfileId	int32	possible	The distribution profile that spreads the quantity over time, into profile.parquet.
CalendarId	int32	possible	The calendar the assignment is scheduled against, into calendar.parquet.
ResourceId	int32	possible	The resource assigned, into resource.parquet.
PlannedHours	double	never	Hours of this resource planned for the activity.
ActualHours	double	never	Hours of this resource expended so far.
EarlyStart	timestamp	possible	Earliest the assignment can start under the schedule logic.
EarlyFinish	timestamp	possible	Earliest the assignment can finish under the schedule logic.
LagHours	double	possible	Lag between the activity starting and the assignment starting, in hours.
CurrentProgress	double	never	Progress reported on the assignment, as a percentage — 20 means 20%.
CalendarRef	string	possible	The IlapId of the calendar in CalendarId. Use this to join across snapshots.
ResourceRef	string	possible	The IlapId of the resource in ResourceId. Use this to join across snapshots.
ProfileRef	string	possible	The IlapId of the profile in ProfileId. Use this to join across snapshots.
Label	string	possible	Human-readable name of the assignment.
LocalSystemId	string	possible	Identifier of the assignment in the planning system it was imported from.
GlobalSystemId	string	possible	Identifier of the assignment in the system of record it originates from.
Code	string	possible	Short code identifying the assignment in the source planning system.
IlapId	string	possible	Stable ILAP identifier for the assignment, carried across imports and deployments.
BaseUri	string	possible	Base URI recorded alongside the ILAP identifier when the assignment was imported.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

This file carries metadata columns. See Metadata columns.

resourceAvailability.parquet

The windows in which a resource is available, and at what quantity or rate. One row per window.

Column	Type	Empty?	Meaning
ResourceId	int32	never	The resource this window applies to, into resource.parquet.
CalendarId	int32	possible	The calendar the window is measured against, into calendar.parquet.
ProfileId	int32	possible	The distribution profile shaping availability across the window, into profile.parquet.
StartDateTime	timestamp	possible	Start of the window. Empty where the window is open-ended at the start.
FinishDateTime	timestamp	possible	End of the window. Empty where the window is open-ended at the end.
AvailableQuantity	double	possible	How much of the resource is available in the window.
AvailableRate	double	possible	The rate at which the resource is available in the window.
Label	string	possible	Human-readable name of the availability window.
LocalSystemId	string	possible	Identifier of the window in the planning system it was imported from.
GlobalSystemId	string	possible	Identifier of the window in the system of record it originates from.
Code	string	possible	Short code identifying the window in the source planning system.
IlapId	string	possible	Stable ILAP identifier for the window, carried across imports and deployments.
BaseUri	string	possible	Base URI recorded alongside the ILAP identifier when the window was imported.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

This file carries no metadata columns.

calendarOperation.parquet

One rule within a working calendar — either working time added or working time removed. One row per rule; when the rule applies is in weeklyRepeatingPeriod.parquet or connectedPeriod.parquet.

Column	Type	Empty?	Meaning
CalendarId	int32	never	The calendar this rule belongs to, into calendar.parquet.
Type	string	never	Whether the rule adds or removes working time — see the values below the table.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

Type is one of: RemoveWorkingTime, AddWorkingTime.

This file carries no metadata columns.

weeklyRepeatingPeriod.parquet

A calendar rule that repeats on a given weekday, with the times it starts and finishes. One row per weekday the rule covers.

Column	Type	Empty?	Meaning
CalendarOperationId	int32	never	The calendar rule this period belongs to, into calendarOperation.parquet.
DayOfWeek	string	never	The weekday the rule repeats on — see the values below the table.
StartTime	timestamp	never	Time of day the period starts. Only the time part carries meaning.
FinishTime	timestamp	never	Time of day the period finishes. Only the time part carries meaning.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

DayOfWeek is one of: Sunday, Monday, Tuesday, Wednesday, Thursday, Friday, Saturday.

This file carries no metadata columns.

connectedPeriod.parquet

A calendar rule covering one continuous range of dates rather than a repeating weekday. One row per range.

Column	Type	Empty?	Meaning
CalendarOperationId	int32	never	The calendar rule this period belongs to, into calendarOperation.parquet.
StartDateTime	timestamp	never	Start of the range the rule covers.
FinishDateTime	timestamp	never	End of the range the rule covers.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

This file carries no metadata columns.

profile.parquet

The distribution profiles used to spread a quantity over time. One row per profile; the curve itself is in profilePoint.parquet.

Column	Type	Empty?	Meaning
DataId	int32	never	The imported data set this profile belongs to.
IlapId	string	possible	Stable ILAP identifier for the profile, carried across imports and deployments.
BaseUri	string	possible	Base URI recorded alongside the ILAP identifier when the profile was imported.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

This file carries no metadata columns.

profilePoint.parquet

The (x, y) points that define a distribution profile. One row per point; order them by X to read the curve.

Column	Type	Empty?	Meaning
ProfileId	int32	never	The profile this point belongs to, into profile.parquet.
X	int32	never	Position of the point along the profile.
Y	double	never	Value of the profile at that position.
Id	int32	never	Identifier of the row, unique within this snapshot only.
Created	timestamp	never	When the row was created in ILAP Analytics.

A profile normally runs from (0, 0) to (100, 100), so both axes read as percentages — proportion of the way through against proportion of the quantity accrued.

This file carries no metadata columns.

See also

• Parquet export — snapshot layout, the completion marker, and the manifest
• Parquet reporting columns — every column of the ten reporting files
• Reading Parquet from Fabric — shortcutting the export into a Lakehouse
• Why the export is plain Parquet