April 23, 2026

Telemetry Update: New kbc_non_sql_workspace Tables & Deprecation of Legacy Telemetry Tables

We are introducing two new telemetry tables — kbc_non_sql_workspace and kbc_non_sql_workspace_run — that unify metadata and consumption tracking for all non-SQL workspaces (Data Apps and Data Science sandboxes). Several legacy tables are being deprecated and will be retired at the end of May 2026.

What's new

kbc_non_sql_workspace — unified metadata for non-SQL workspaces, covering both Data Apps (streamlit, python-js) and Data Science sandboxes (python, r, etc.).
kbc_non_sql_workspace_run — runtime and consumption metrics per workspace per date, including runtime_hours, time_credits_used, and billed_credits_used.

For a full description of the new tables and their columns, please refer to the documentation.

Tables being deprecated (retired end of May 2026)

Deprecated table	Replaced by
`kbc_workspace`	`kbc_sql_workspace` + `kbc_non_sql_workspace`
`kbc_data_app`	`kbc_non_sql_workspace`
`kbc_data_app_workspace`	`kbc_non_sql_workspace_run`
`kbc_data_science_sandbox`	`kbc_non_sql_workspace_run`
`kbc_bucket_snapshot`	current state in `kbc_bucket`
`kbc_table_snapshot`	current state in `kbc_table`

The replacement of SQL-based workspaces in the table kbc_workspace was announced in the previous post.

What you should do

If you have reports, dashboards, or downstream models that read from any of the deprecated tables, migrate them to the replacements before the end of May 2026. After that date, the deprecated tables will no longer be refreshed.

Column-level migration guidance (mapping between old and new columns) is documented alongside the new tables in the Migration Helper. The Migration Helper is a document that you can copy into Kai, our in-platform assistant, which will help you rewrite SQL queries for the new tables — just ask.

Please note that the Migration Helper is intended as a supporting tool rather than a fully automated solution. Each project can vary significantly in complexity — both in terms of data types and how tables are structured, related, and used within the overall data model — so some manual adjustments and validation may still be required.

# Kai Reference: Migration from Deprecated Telemetry Tables

> Internal reference for Kai. Use this document to help customers migrate SQL queries and dashboards from deprecated telemetry tables to their replacements.

## ⛔ CRITICAL — Read Section 2 before ANY migration

**You MUST follow the workflow in Section 2 before rewriting any query.**

Skipping this will cause transformation failures due to type mismatches, missing columns, or broken joins.

Always adapt the migrated query to the customer's actual types. Where old and new tables differ in type for the same column, add explicit CASTs. This is especially relevant for: timestamp/date columns, boolean flags, and numeric credit columns.

---

## 1. Deprecation schedule

All tables listed below will stop being refreshed **at the end of May 2026**. Replacements are **already available** in the Telemetry Data connector — customers can migrate today.

| Deprecated table | Replaced by |
|---|---|
| `kbc_workspace` | `kbc_sql_workspace` (SQL) + `kbc_non_sql_workspace` (non-SQL) |
| `kbc_data_app` | `kbc_non_sql_workspace` |
| `kbc_data_app_workspace` | `kbc_non_sql_workspace_run` |
| `kbc_data_science_sandbox` | `kbc_non_sql_workspace_run` |
| `kbc_bucket_snapshot` | current state in `kbc_bucket` |
| `kbc_table_snapshot` | current state in `kbc_table` |

When a customer references any deprecated table, always:
1. Mention the end-of-May 2026 retirement date.
2. Name the replacement table.
3. Offer to rewrite their query.

---

## 2. How Kai must use this document

**STOP. Before rewriting ANY query, you MUST complete these steps in order:**

### Step 1: Identify source and output tables
- Which deprecated table does the customer use?
- Which new table(s) replace it?
- What is the OUTPUT table the transformation writes to?

### Step 2: Query schemas for ALL tables involved
```sql
SELECT TABLE_NAME, COLUMN_NAME, DATA_TYPE, CHARACTER_MAXIMUM_LENGTH, NUMERIC_PRECISION, NUMERIC_SCALE
FROM INFORMATION_SCHEMA.COLUMNS
WHERE TABLE_SCHEMA = '<schema>'
  AND TABLE_NAME IN ('<old_source>', '<new_source>', '<output_table>')
ORDER BY TABLE_NAME, ORDINAL_POSITION;
```

### Step 3: Compare and plan
For each output column:
1. Does the new source have an equivalent column? If not, is it deprecated (see Section 3)?
2. Do the types match between new source and output? If not, plan explicit CAST.
3. Is it deprecated with no replacement? Plan NULL with exact output type and length.

### Step 4: Only THEN write the migrated query

**CRITICAL RULES:**
- **NEVER** delete columns from output — downstream dashboards depend on them
- **NEVER** change output column types, lengths, or formats — if output was VARCHAR, preserve the original string format (especially for timestamps)
- **ALWAYS** cast to EXACT output type including length
- **ALWAYS** join `kbc_non_sql_workspace_run` with `kbc_non_sql_workspace` to get `component_id` for filtering
- If a join is BROKEN (e.g., token_id in run tables), inform customer explicitly — do not silently remove the column

**If you cannot determine the output schema, ASK the customer before proceeding.**

---

## 3. Breaking changes — columns removed without replacement

> **⛔ CRITICAL: These columns are PERMANENTLY REMOVED. You MUST inform the customer explicitly. Do NOT silently remove them from output — keep them as NULL with correct type.**

### Columns that NO LONGER EXIST and CANNOT be derived

| Deprecated Table | Removed Column | Impact |
|---|---|---|
| `kbc_data_app_workspace` | `kbc_token_id`, `token_name` | **Cannot join to `kbc_token` or `kbc_user`** |
| `kbc_data_science_sandbox` | `kbc_token_id`, `token_name` | **Cannot join to `kbc_token` or `kbc_user`** |
| `kbc_data_science_sandbox` | `owner_id` | **Cannot identify sandbox owner** |
| `kbc_workspace` (non-SQL only) | `workspace_creator`, `workspace_user` | SQL workspaces retain `workspace_creator`. |

### Why token from workspace table CANNOT replace token from run table

The `kbc_token_id` in `kbc_non_sql_workspace` identifies who **created the workspace/configuration**, NOT who **executed individual runs**. These are semantically different values.

**You MUST NOT** substitute `kbc_non_sql_workspace.kbc_token_id` for the deprecated run-level token columns.

### Join paths that are PERMANENTLY BROKEN

| Old Join | Status | What to tell customer |
|---|---|---|
| `kbc_data_app_workspace` → `kbc_token` | **BROKEN** | "Token association is no longer tracked for Data App runs. This join cannot be migrated." |
| `kbc_data_science_sandbox` → `kbc_token` | **BROKEN** | "Token association is no longer tracked for sandbox runs. This join cannot be migrated." |
| `kbc_data_science_sandbox` → `kbc_user` (via `owner_id`) | **BROKEN** | "Owner information is no longer available." |

### When migrating queries that use these columns

1. **Keep the column in output** — do not delete it (downstream dashboards depend on schema)
2. **Set value to NULL** with the exact type and length from the output table schema
3. **Inform customer explicitly** — this is a breaking change requiring acknowledgment

---

## 4. `kbc_workspace` → `kbc_sql_workspace` / `kbc_non_sql_workspace`

`kbc_workspace` is split by workspace kind. Decide the target from `workspace_type`:
- SQL workspaces (e.g. `snowflake`, `bigquery`) → `kbc_sql_workspace`.
- Non-SQL workspaces (`streamlit`, `python-js`, `python`, `r`, `python-snowpark`, `python-mlflow`, and others) → `kbc_non_sql_workspace`.

### Column mapping (non-SQL target)

| `kbc_workspace` | `kbc_non_sql_workspace` | Note |
|---|---|---|
| `kbc_workspace_id` | `kbc_non_sql_workspace_id` | Renamed. |
| `workspace_expired` | `workspace_deleted` | Renamed. Contains deletion timestamp or is NULL/empty. To check if deleted: `workspace_deleted IS NOT NULL AND workspace_deleted <> ''`. |
| `workspace_shared` | `workspace_shared` | Same. |
| `workspace_hostname` | `workspace_hostname` | Same. |
| `workspace_type` | `workspace_type` | Same. |
| `workspace_created` | `workspace_created` | Same. |
| `workspace_updated` | `workspace_updated` | Same. |
| `kbc_token_id`, `kbc_token_name` | `kbc_token_id`, `kbc_token_name` | Same, but represents **workspace creator** — NOT run executor. Cannot replace deprecated token columns from run tables. See Section 3. |
| `kbc_workspace_name` | `kbc_workspace_name` | Same. |
| `storage_size_gb` | `storage_size_gb` | Same. |
| `workspace_expire_hours` | `workspace_expire_hours` | Same. |
| `kbc_project_id` | `kbc_project_id` | Same. |
| `backend_size` | — | Removed — moved to the run table (`kbc_non_sql_workspace_run.backend_size`). |
| `workspace_creator` | — | Removed. Token in `kbc_non_sql_workspace` identifies the creator. |
| `workspace_active` | — | Removed — was not relevant, it was a technical parameter pointing to the state at the moment of extracting data on the Keboola side. |
| `workspace_start` | — | Removed. Use max timestamp of `kbc_non_sql_workspace_run.workspace_start_at` per `kbc_non_sql_workspace_id`. |
| `workspace_user` | — | Removed. |
| `workspace_read_only_storage_access` | — | Removed. |
| `workspace_schema` | — | Removed (not applicable for non-SQL). |
| — | `workspace_id_num` | **New** — numeric portion of the id. |
| — | `component_id`, `kbc_component` | **New** — component origin. |
| — | `kbc_component_configuration_id` | **New**. |
| — | `is_persistent` | **New** — moved from `kbc_data_science_sandbox`. |
| — | `kbc_data_app_id` | **New** — legacy-format Data App id for joining `kbc_table_event` and `kbc_data_app_params_event`. |

For SQL workspace column mapping see below. Always filter by `workspace_type` before choosing the target table.

### Column mapping (SQL target)

| `kbc_workspace` | `kbc_sql_workspace` | Note |
|---|---|---|
| `kbc_workspace_id` | `kbc_sql_workspace_id` | Renamed. |
| `workspace_expired` | `workspace_deleted` | Renamed. Contains deletion timestamp or is NULL/empty. |
| `workspace_created` | `workspace_created` | Same. |
| `workspace_updated` | `workspace_updated` | Same. |
| `workspace_shared` | `workspace_shared` | Same. |
| `workspace_read_only_storage_access` | `read_only_storage_access` | Renamed. |
| `workspace_type` | `workspace_type` | Same. |
| `kbc_token_id`, `kbc_token_name` | `kbc_token_id`, `kbc_token_name` | Same. |
| `kbc_workspace_name` | `kbc_workspace_name` | Same. |
| `backend_size` | `backend_size` | Same. |
| `workspace_schema` | `workspace_schema` | Same. |
| `kbc_project_id` | `kbc_project_id` | Same. |
| `workspace_creator` | `workspace_creator` | Same. |
| `workspace_active` | — | Removed — was a technical flag, not relevant for users. |
| `workspace_start` | — | Removed. |
| `workspace_user` | — | Removed. |
| `workspace_expire_hours` | — | Removed — not relevant for SQL workspaces. |
| `storage_size_gb` | — | Removed — not relevant for SQL workspaces. |
| `workspace_hostname` | — | Removed — not relevant for SQL workspaces. |
| — | `kbc_component_configuration_id` | **New**. |
| — | `workspace_database` | **New** — database name. |

---

## 5. `kbc_data_app` → `kbc_non_sql_workspace`

Filter the new table to Data Apps only: `component_id = 'keboola.data-apps'`. Do **not** rely solely on `workspace_type IN ('streamlit','python-js')` — the `streamlit` type also exists for sandbox workspaces created via `keboola.sandboxes`.

> **Warning — `kbc_component` and `kbc_component_id` mapping changed.** The old `kbc_component` column contains the technical component ID (e.g., `keboola.data-apps`) and maps to `component_id` in the new table. The old `kbc_component_id` (e.g., `keboola.data-apps_com-keboola-gcp-us-east4`) has no direct equivalent in the new table and must be derived. Additionally, the new table introduces a `kbc_component` column with a **different meaning** — it is the human-readable display name (e.g., `Data App`), not the technical component ID.

| `kbc_data_app` | `kbc_non_sql_workspace` | Note |
|---|---|---|
| `kbc_data_app_id` | `kbc_non_sql_workspace_id` | Renamed. **ID format changed** — new id is `workspace_id_stack`, old id was `project_id_stack_component_id_workspace_id`. The old-format id is preserved in the new `kbc_data_app_id` column for joins with `kbc_table_event` and `kbc_data_app_params_event`. |
| `kbc_data_app_name` | `kbc_workspace_name` | Renamed. |
| `data_app_created` | `workspace_created` | Renamed. |
| `kbc_component` | `component_id` | Renamed. Contains the technical component identifier (e.g., `keboola.data-apps`). **Do not confuse with the new `kbc_component` column, which is the human-readable display name (e.g., `Data App`).** |
| `kbc_component_id` | — | No direct mapping. Derive as: `component_id \|\| '_' \|\| REGEXP_REPLACE(kbc_project_id, '^[0-9]+_', '')`. Example: `keboola.data-apps_com-keboola-gcp-us-east4`. |
| `kbc_component_configuration_id` | `kbc_component_configuration_id` | Same. |
| `data_app_id_num` | `workspace_id_num` | Renamed. |
| `kbc_data_app_url` | `workspace_hostname` | Renamed. |
| `kbc_data_app_is_deleted` | `workspace_deleted` | **Format changed** — value `true`/`false` in the old table maps to `workspace_deleted IS NOT NULL` or `<> ''` / `IS NULL` or `= ''` in the new table. |
| — | `kbc_component` | **New** — human-readable display name (e.g., `Data App`). Not equivalent to the old `kbc_component` column. |
| — | `workspace_updated`, `workspace_type`, `kbc_token_id`, `kbc_token_name`, `workspace_expire_hours` | **New.** |

---

## 6. `kbc_data_app_workspace` → `kbc_non_sql_workspace_run`

> **⚠️ REQUIRED JOIN:** The `kbc_non_sql_workspace_run` table does NOT contain `component_id`. You MUST join with `kbc_non_sql_workspace` on `kbc_non_sql_workspace_id` to filter Data Apps (`component_id = 'keboola.data-apps'`). Do not rely solely on `workspace_type` — `streamlit` type also exists for sandbox workspaces.
>
> **⚠️ BROKEN COLUMNS:** `kbc_token_id` and `token_name` are DEPRECATED and cannot be replaced. See Section 3.

| `kbc_data_app_workspace` | `kbc_non_sql_workspace_run` | Note |
|---|---|---|
| `kbc_data_app_workspace_run_id` | `kbc_non_sql_workspace_run_id` | Renamed. |
| `kbc_data_app_workspace_id` | `kbc_non_sql_workspace_id` | Renamed. |
| `kbc_data_app_name` | `kbc_workspace_name` | Renamed. |
| `start_at` | `workspace_start_at` | Renamed. |
| `sleep_at` | `workspace_sleep_at` | Renamed. |
| `type` | `workspace_type` | Renamed. |
| `data_app_runtime_hours` | `runtime_hours` | Renamed. |
| `date` | `date` | Same. |
| `backend_size` | `backend_size` | Same. |
| `time_credits_used` | `time_credits_used` | Same. |
| `billed_credits_used` | `billed_credits_used` | Same. |
| `kbc_project_id` | `kbc_project_id` | Same. |
| `created_at`, `expiration_at`, `updated_at`, `hostname` | — | Moved to `kbc_non_sql_workspace` (`workspace_created`, `workspace_deleted`, `workspace_updated`, `workspace_hostname`). Join on `kbc_non_sql_workspace_id`. |
| `kbc_token_id`, `token_name` | — | **DEPRECATED** — not available in new tables. Cannot be derived or joined. |
| `kbc_data_app_id` | — | Moved to `kbc_non_sql_workspace.kbc_data_app_id`. |

---

## 7. `kbc_data_science_sandbox` → `kbc_non_sql_workspace_run`

> **⚠️ REQUIRED JOIN:** The `kbc_non_sql_workspace_run` table does NOT contain `component_id`. You MUST join with `kbc_non_sql_workspace` on `kbc_non_sql_workspace_id` to filter sandboxes (`component_id = 'keboola.sandboxes'`). Do not use `workspace_type` alone — sandbox types include `python`, `r`, `python-snowpark`, `python-mlflow`, and `streamlit` can also belong to sandboxes.
>
> **⚠️ BROKEN COLUMNS:** `kbc_token_id` and `token_name` are DEPRECATED and cannot be replaced. See Section 3.

| `kbc_data_science_sandbox` | `kbc_non_sql_workspace_run` | Note |
|---|---|---|
| `kbc_data_science_sandbox_resume_id` | `kbc_non_sql_workspace_run_id` | Renamed. |
| `kbc_data_science_sandbox_id` | `kbc_non_sql_workspace_id` | Renamed. |
| `start_at` | `workspace_start_at` | Renamed. |
| `sleep_at` | `workspace_sleep_at` | Renamed. |
| `type` | `workspace_type` | Renamed. |
| `sandbox_runtime_hours` | `runtime_hours` | Renamed. |
| `date` | `date` | Same. |
| `backend_size` | `backend_size` | Same. |
| `time_credits_used` | `time_credits_used` | Same. |
| `billed_credits_used` | `billed_credits_used` | Same. |
| `kbc_project_id` | `kbc_project_id` | Same. |
| `created_at`, `expiration_at`, `updated_at`, `hostname` | — | Moved to `kbc_non_sql_workspace`. Join on `kbc_non_sql_workspace_id`. |
| `kbc_token_id`, `token_name` | — | **DEPRECATED** — not available in new tables. Cannot be derived or joined. |
| `is_persistent` | — | Moved to `kbc_non_sql_workspace.is_persistent`. Join on `kbc_non_sql_workspace_id`. |
| `storage_size_gb` | — | Moved to `kbc_non_sql_workspace.storage_size_gb`. Join on `kbc_non_sql_workspace_id`. |
| `legacy_sandbox` | — | **Removed** — no replacement. |

---

## 8. `kbc_bucket_snapshot` → `kbc_bucket` (current state)

`kbc_bucket_snapshot` was a daily copy of bucket state. The current state is now exposed in `kbc_bucket`, which already reflects the live state of each bucket. Point-in-time history (e.g. "state as of 2025-06-01") is no longer available via telemetry.

- Drop the `snapshot_date` filter and read from `kbc_bucket`.
- Foreign keys and size/row columns are available in `kbc_bucket`.

> **Note for customers requiring time-travel:** Snapshot tables are no longer provided. However, customers who download telemetry data daily can create their own transformation to accumulate historical snapshots and recreate time-travel functionality themselves.

---

## 9. `kbc_table_snapshot` → `kbc_table` (current state)

Same pattern as buckets: `kbc_table_snapshot` was a daily copy; use `kbc_table` for current state. Historical comparisons across snapshot dates are no longer supported via telemetry.

| `kbc_table_snapshot` | `kbc_table` | Note |
|---|---|---|
| `table_id` | `table_id` | Same. |
| `bucket_id` | — | **Derive from `kbc_project_bucket_id`** using `REPLACE(kbc_project_bucket_id, kbc_project_id || '_', '')` to remove the full project prefix (including stack). |
| `kbc_project_id` | `kbc_project_id` | Same. |
| `snapshot_date` | `last_snapshot_date` | Use `last_snapshot_date` as reference for current state. |
| All other columns | Same names | `created`, `last_import`, `table_name`, `primary_key`, `is_alias`, `alias_column_sync`, `source_project_id`, `source_table_id`, `alias_filter_column`, `alias_filter_operator`, `alias_filter_value`, `rows`, `bytes`, `description`, `description_short`. |

> **Note for customers requiring time-travel:** Snapshot tables are no longer provided. However, customers who download telemetry data daily can create their own transformation to accumulate historical snapshots and recreate time-travel functionality themselves.


---

## 10. Specific derivations

These derivations are not obvious from the column mapping tables and require special handling.

### 10.1 Joining Data App to event tables (legacy id format)

The new `kbc_non_sql_workspace_id` uses a different format than the old `kbc_data_app_id`. To join with `kbc_table_event` or `kbc_data_app_params_event`, use the `kbc_data_app_id` column which preserves the legacy format:

```sql
SELECT w.kbc_workspace_name, e.*
FROM kbc_non_sql_workspace   w
JOIN kbc_table_event         e
  ON e.kbc_data_app_id = w.kbc_data_app_id
WHERE w.component_id = 'keboola.data-apps';
```

### 10.2 Deriving bucket_id from kbc_table

The `bucket_id` column does not exist in `kbc_table`. Derive it from `kbc_project_bucket_id`:

```sql
REPLACE(kbc_project_bucket_id, kbc_project_id || '_', '') AS bucket_id
```

Share on: Twitter LinkedIn Facebook