Skip to main content

Module 06: Dataset Monitoring Dashboard

Every day, data arrives from exchanges, brokers, data vendors and your own systems. The dataset monitoring dashboard tells you, at a glance, whether each delivery arrived, whether it arrived on time and in full, and whether it passed its quality checks. In this module you will learn to read it.

Learning Objectives​

By the end of this module you will be able to:

  • Explain what a dataset, a dataset feed and a dataset delivery are
  • Choose which datasets you monitor
  • Read the status icons on each dataset: delivery, score, quality, smart loader and alerts
  • Use the dataset tabs to see the completeness and quality check results for a date
  • Filter the list to find the datasets that need attention

Prerequisites​

Example data

Any dataset your company monitors will do, for example the ECB reference rates dataset ECB.FX.SPOT shown in the screenshots. If the Technical track's training scenario is set up, look for TRAINING.MYGASHUB.SETTLEMENTS and TRAINING.MYGASHUB.NOMINATIONS. Setting up datasets and their checks is a developer or administrator task and is covered in Technical Module 06.


Step 1: Understand the Terms​

TermWhat it isExample id
Dataset feedThe expected time window for a feed from one providerICE.IFEU
DatasetOne product within a feed, and how much data is expected each dayICE.IFEU.B
Dataset deliveryThe record of what actually arrived for a dataset on one day, with its timing, completeness and qualityICE.IFEU.B:2024-06-12
Smart loaderAn optional loader that re-runs the load until the data is complete

A dataset id is always PROVIDER.FEED.PRODUCT.

A day in the life of a delivery​

The delivery lifecycle

  1. At the start of the day each monitored dataset gets a new delivery record with status waiting and a score of 4
  2. When data arrives, the platform checks for corrections, counts what was loaded against what is expected, and sets the status to partial or loaded
  3. Quality checks run on the full dataset after each load
  4. A background check lowers the score if the data is late, and raises an alert if it is late or missing

Step 2: Open the Dashboard and Read the Overview​

Open the dataset monitoring dashboard from the portal menu. It has three parts:

  • Overview: a summary of the current status of all monitored datasets and alerts
  • Dataset list: detailed information about each monitored dataset
  • Reports: public and private reports about datasets

On the overview, use the view by dropdown to summarise the status by Provider, Category or Priority. Category and priority are labels your company assigns to datasets, so you can, for example, look at all your high-priority gas datasets together.


Step 3: Choose What You Monitor​

The dataset list shows only the datasets you have chosen to monitor.

  1. Select the Monitored button at the top left The Monitored button. The list now shows every dataset you are entitled to see
  2. Select the add button The add button on each dataset you want to monitor
  3. Select Monitored again to go back to your monitored list

Step 4: Read the Status Icons​

Each dataset in the list shows a row of status icons:

The status icons of a dataset

IconValuesWhat it tells you
Delivery Statuswaiting, active, partial, holiday, loaded, late, missingWhere today's delivery is in its lifecycle
Delivery Score0 to 4How promptly the data arrived (see below)
Quality Statusunchecked, valid, warning, failed, severeThe overall result of the quality checks
Smart Statuswaiting, running, loaded, failed, holidayThe smart loader's status, if one loads this dataset
AlertsA numberHow many alerts were raised for this dataset

Delivery scores​

ScoreMeaning
4Data arrived on time
3Up to 1 hour late
2Between 1 and 4 hours late
1More than 4 hours late
0No data expected, for example a holiday
What needs attention?

Look for a delivery status of late or missing, a quality status of failed or severe, a smart status of failed, or an alert count above zero. Partial means some, but not yet all, of the expected data has arrived.


Step 5: Look at One Delivery​

Select a dataset to open its tabs on the right of the screen, for the selected date.

The General tab with completeness and quality checks

TabWhat it shows
GeneralThe status information for the date, and the result of each completeness check (expected against actual) and quality check
FeedThe configuration of the feed this dataset belongs to, and smart loader logging
DataThe data for this dataset for the selected date
TimelineEverything that happened to this dataset on the selected date
AlertsThe alerts raised for this dataset on the selected date
InsightsInsight reports about this dataset

Other tabs may appear with documentation about the dataset, written by OpenDataDSL or by your own teams.

Actions on a dataset

Hover over a dataset in the list for its actions: Edit opens a side panel to edit the dataset, Remove takes it off your monitored list, and Copy copies its id. Depending on your access rights, the dataset tabs may also offer actions such as running the quality checks again. Ask your data team before changing a dataset's configuration.


Step 6: Filter the List​

To find the datasets that need attention quickly, filter the list by:

  • Delivery Status or Quality Status, for example only late and missing
  • Provider, Feed, Product or DSID
  • Category or Priority
  • Smart Feed Status
  • Source: private, common or public

Exercise​

  1. Open the dataset monitoring dashboard and look at the overview by Provider, then by Priority
  2. Add two datasets to your monitored list, for example ECB.FX.SPOT and one your team relies on
  3. For each, note the delivery status, score and quality status for today
  4. Open the General tab of one dataset. How many tenors were expected, and how many arrived?
  5. Filter the list to show only datasets that are late or missing. If any are listed, open the Timeline tab to see what happened

Summary​

StatusHealthyNeeds attention
Deliveryloaded, holidaylate, missing (partial while still loading)
Score43, 2 or 1
Qualityvalidwarning, failed, severe
Smart loaderloadedfailed
Alerts0any

Further reading: Dataset Monitoring dashboard guide, Data Monitoring topic, Dataset Monitoring overview

Next Step

In Module 07 you will run reports and insight reports, including the reports about your datasets.