Structured Working and Data Backup

This is part 1 of our data backup strategy.
Part 2 on backup strategies findest du hier →

Introduction

Loosing data is a nightmare

It's worse to restore jobs that you've already been paid for and now have to do all the work again.

That's why the effort we put into backing up our data can't be high enough.

It doesn't even have to be the accidental deletion or overwriting of data. Files can also break. Data storage, on the other hand, always breaks! You just don't know when.

To avoid data loss or corrupted data, we use a combination of unique file naming, a file handling system and backup.

With these three methods, you can reduce the risk of data loss to practically zero.

And the good thing is, it doesn't even cost much.

---

Handling data securely

In order to manage data securely, to keep a chronological version history as far as possible, to search and find data, we apply the following guidelines:

1. Create a data folder

• Customer data is ALWAYS stored centrally.
• There is a superordinate folder in which all jobs currently being processed are stored. (We call it _Jobs) are stored.
  
• Each customer has a master folder. Their projects and jobs are stored in this folder.
• Jobs for the customer are assigned a job number at the latest when the job is entered.
• The new job is placed in the jobs folder and given a unique name.
  

2. Order folder

An order is labelled according to the order number (or quotation number), e.g:

0000_Customer Letterhead Invoice A4 DE

Everything relating to this order is collected in this folder.

Subfolders are usually

• Links
• Fonts
• Offers
  (own and external companies, as well as customer orders)
• delivered (optional)
• Print PDF (optional)
• data given to the customer (optional)
• old data (more on this later)

3. File names

We give (almost) all data the following names, for example

0000_Customer Letterhead Invoice A4 DE 2021-0118a.indd

Why the date?

We have often had cases where the actual date stamp of the file was no longer meaningful or correct due to editing or archiving software, copying of data and the like. The date we added ourselves is unique and is not changed by software or the operating system.

When will the date be set?

As soon as we edit a file, the file is copied BEFOREhand in a closed state.

The COPY is then given a new name.

Files are labelled with an 'a', 'b' or 'c' on the same day, the next day then starts with an 'a' again.

0000_Customer letterhead invoice A4 EN 2021-0118a.indd
then becomes
0000_Customer letterhead invoice A4 DE 2021-0118b.indd
0000_Customer letterhead invoice A4 DE 2021-0118c.indd

The next day we continue with
0000_Customer letterhead invoice A4 EN 2021-0122a.indd

Old data goes everytime into the 'old data' folder

This procedure ensures that changes can be tracked. It also ensures that there is always a file that was still OK after the last save.

Once the order has been completed, it is moved completely to the customer folder. If you work a lot for a customer, you can of course create additional subfolders such as

• Adverts
• Brochures
• Advertising material

Updates to existing jobs

If an update of this job becomes necessary after some time, we MOVE the entire folder back into the jobs folder. This ensures that there is only ONE current version and there is no need to guess whether this is a copy or the current version or something else.

If a new project based on an old job is continued at a later date, we copy practically everything from the old project into a new folder and then into the 'old data' folder. The new job is given a new number and everything starts all over again.

This means you have all the relevant and necessary data for this job in one place. You can also search for an old job and always have the complete data to hand.

Which data is important?

Before we back up data, we should consider which data is important to us.

We differentiate between

• Customer data (sacred and precious)
• Our own data (important, but if it's gone, it's just stupid and usually only costs grey hair and pimples)
• Own computer and operating system (not quite as important, annoying if it crashes)
• Programmes (rather unimportant because they have to be obtained again and again)
• Other data

The more important the data, the greater the effort we should make in the event of a failure. And here we are thinking, for example, of a hard drive crash, raid error, burglary, fire, etc.).

Organisation in the data

Structured data organisation means: structured backups

The easiest way to make a backup is as I said: simply back up everything.

This also works as long as the amount of data is manageable. As each backup takes time to determine what needs to be backed up, short backup intervals can mean that the backup is still running while the next one is already due. Not sensible...

It therefore makes sense to store all data in a structured manner.

We have two requirements for this:

• Data is stored in a clearly defined location
  (and not 'cached' somewhere on the desktop)
• the data storage location is always available
  (i.e. no removable disc or anything else)

In our company, all the data is stored centrally on a raid system. Several users have access to this system.

Our backup strategy

As we already use a structure similar to an incremental backup for our file handling and our dated data, we do without this for our backup and only create a copy of our data.

We now know what we secure, how we secure and when we secure.

We need to take a closer look at where we are securing.

---

Data storage locations

We store backups of current and long-term data on an external medium.

But that's not really enough. Data should also be stored outside the company/flat so that we are not left without data in the event of a burglary/water pipe burst, for example.

To ensure this, we create additional backup jobs that also back up the data to the cloud. We opted for Backblaze because of its simplicity and price. The advantage here is that we can access the data via FTP or web browser in an emergency.

In contrast to Amazon Glacier, which always argues with the very low price, there is no waiting time for a restore with Backblaze, the backend and access is absolutely simple. The costs at Backblaze are also manageable. On average, you pay around USD 3.50 per TB per month.

Emergency testing

The most important thing for the entire backup process is: how do I get my data back when I need it? You should definitely try this out. It must be clearly understandable and easy to handle. In an emergency, you have stress and don't have the time and nerves to try things out and then realise that you can't access the data.

Storage media

We now store our data locally on a removable medium such as an external hard drive and store data externally in the cloud.
But is that enough?

Please read the following article →

Summary

To handle and store data securely, we need:

• Order in our databases
• Sensible naming
• Use of our own time stamps
• Software that copies data automatically
• External storage media in different locations

Sources:

rsync : https://de.wikipedia.org/wiki/Rsync

rdiff : https://rdiff-backup.net/

Carbon CopyCloner : https://bombich.com/de

Backblaze : https://www.backblaze.com/b2/cloud-storage.html

NeoFinder : https://www.cdfinder.de/

AbeMeda : http://www.abemeda.de/de/