Ten lame excuses for not backing up your server
There are lots of excuses given for not backing up a server, below, in not
particular order are ten of my favorites with some comments:
- OS only
- Which you will never need to patch, upgrade or alter in any way.
Obviously there are many variations on this theme, but the argument
given is along the lines that the server has an Operating System image
installed with, perhaps, a defined set of applications, all of which
could easily be re-installed / re-imaged on to the server. In real life
servers tend to be "living" things that change and grow over time;
administrators spend their lives implementing changes and tweaks and
checking logs and configurations all of which become very precious once
lost.
- Data never changes
- So it can be stored on read-only media and a few copies can be kept
in safe places - there is your backup solution.
- Cache / transient data only
- In which case do the job properly and keep the data in RAM. An
example of this I once encountered was an FTP server that was only to
be used for transient data and would not need to be backed up.
Following the crash of the server, a long time was spent re-creating
the accounts information and then it was discovered that people had
been using the service as a collaboration facility and had lost
important information.
- Data is not important
- Then remove it from your system. It is amazing how childish some
people can be, often without realizing it. The data is important so it
needs to be backed up, but someone does not want to pay for the backup
devices and / or media so they claim the data is not worth backing up.
And when you suggest removing the data the same people protest. Their
message is "the data is important, but you can not make me admit that".
Any data worth keeping is worth backing up.
- High cost of backup media
- The high cost of down time and the cost of re-entering or re-creating
data, assuming this is even possible, can be extremely high.
- Systems are highly reliable and have high redundancy
- And when disaster strikes? A version of this excuse is that the data
is distributed accross multiple systems located in multiple geographic
locations and the chances of loosing all of the copies of the data are
negligably small; which if fine until your synchronisation process
copies a corrupted version of the data to all the repositories.
- Staff won't / can't change media
- Buy a robot or get management to do their job.
- The server is about to be decommissioned.
- So if the data is no longer needed, switch the server off now,
otherwise keep backing it up.
- The server is new / experimental / in development.
- Which means that it is probably in a state of rapid change with
people doing a lot of configuration work which they will not want to
loose. This is similar to the above situation.
- It is too difficult
- No it is not.
I have worked as a system administrator at various times for a number of
years and have encountered all the above excuses in one form or another at
various times. The person or people advocating that a server is not backed up
is almost never the person or people who will have to recover data if it is
lost. If a system is not backed up you can be sure that a failure or problem
of some kind will occur and despite any previous promises to the contrary
someone will want the data back.