Where does a million sales go?
A shop chain has 200 shops. Every till (the machine where you pay) sends a sale every few seconds. Cameras record video. Customers send emails.
By the end of the day there are millions of records. The files have many shapes. Where do they all go? And how does the marketing team find what it needs?
Key words
| Word | What it means | Example |
|---|---|---|
| Big data | Data so large, fast or mixed that a normal spreadsheet cannot hold it. | Millions of sales a day |
| Data lake | A store that keeps all data in its raw form, in any format. | Till logs, videos, emails |
| Data warehouse | A store of cleaned data, organised in one agreed shape, for reports. | A table of all sales |
| Data mart | A small slice of a data warehouse for one team. | Sales for marketing |
| Drawback | A bad point of something. A benefit is a good point. | A lake can become a mess |
| Raw data | Data kept as it arrived. Nobody has cleaned it. | A till log |
| Cleaned data | Data with mistakes and repeats removed and put in one shape. | A tidy table |
| Structured data | Data in rows and columns with set types. | A sales table |
| Unstructured data | Data with no set rows and columns, such as video or email text. | A camera video |
The cells students fill in are left blank in this preview.
Quick check: raw or cleaned?
Have a go
Quick check: raw or cleaned?
4 questionsYour answers are marked here and are not kept. A college account keeps every student’s marks.
The data lake: keep everything
A data lake stores data as it arrives. Video, emails, logs and tables all go in. Nobody cleans it first.
It is cheap and flexible: it takes any kind of data. You can decide later what to do with the data. The drawback: a lake with no care becomes a mess. People cannot find or trust what is inside.
The data warehouse: clean and ready
A data warehouse holds data that is cleaned and put into one agreed shape. All sales sit in the same table with the same columns.
Reports run fast and give the same answer to everybody. The drawback: it takes time and skill to clean and load the data. It is also less flexible than a lake.
The data mart: a slice for one team
A data mart is a small part of a warehouse. It holds only what one team needs, such as marketing.
It is small, quick and simple to use. The drawback: the copies can stop matching. The finance team and the marketing team may see different numbers.
The three stores side by side
| Store | Holds | Who uses it | Benefit | Drawback |
|---|---|---|---|---|
| Data lake | Raw data of any kind | Data scientists | Cheap and flexible | Messy. Hard to trust. |
| Data warehouse | Cleaned, structured data | Analysts and managers | Fast, agreed answers | Costs a lot to clean and load |
| Data mart | A slice of the warehouse | One team | Small and quick | Copies may differ |
The cells students fill in are left blank in this preview.