2015-09: Backups

### Discussion Questions

– What do you care about backing up? What do you not care about?
– What backup products do you use?
– What backup workflows do you use?
– What products do you use? rsync? Backup Exec? Crashplan? Other?
– How do you test your backups?
– How do you test disaster recovery?

### Linux backups for publishing

Linux server backs up files from the last 45 days to a DVD (as much as would fit)

+ Back up once a month
+ Plus nightly backups with rdiff-backup (which includes history) on a 3-disk RAID array
+ Would restore files from DVD
+ Low strain of hardware, low attack surface from web
+ The DVD backup was in use for 15 years
+ Each book project is responsible for its own backups
+ Restructuring a full backup would be tedious
+ But the most recent files are accessible (and these are the most important)
+ System was designed when most files were smallish, but Adobe Indesign makes big files

### At the Working Centre

+ Symantec Backupexec for most file servers, mail server, Active Directory
+ Remote sites we use Windows Server Backup + backupexec once a week
+ Active Directory is important but remote domain controllers is a backup
+ My Documents are mapped and get backed up
+ Desktops, bookmarks, PST files do not get backed up on clients
+ Weekly full backups + daily differentials for Windows
+ Linux: tarballs of important files + rsync to a central server
+ Backup server in our server room + (maybe) remote sync
+ Copy most important files on encrypted USB keys

### Big Backups don’t sync!

– DFS doesn’t work!
– Syncthing doesn’t work!

### Sidebar: What is Active Directory?

– SMB is Samba storage and network (Windows File shares)
– Active Directory is an LDAP database:
+ User accounts and passwords
+ Security groups
+ Computer accounts
– On Linux: there are RedHat, SuSE LDAP solutions
– On Samba 4, you can mimic an Active Directory domain
+ Linux clients can (kind of) join a domain, and log in using Active Directory credentials
+ Active Directory implements a bunch of open standards, so you can connect to it from Linux clients

### What happens at REEP?

– Two servers at one location
– Everything is backed up there: file shares, Active Directory, …
– Image for Windows is what he uses
+ Does an image backup for the servers
+ Saved onto a separate internal drive
+ Full backups are once a month + differentials
– Puts the monthly full on an external drive
– Also uses shadow copy
+ This gives end users previous versions
– JungleDisk backs up file shares to their Amazon servers
+ $10-15/month for 200GB of storage
+ It has some historical information
+ Do you trust them? Yes
+ Information is not highly confidential
+ Backs up twice a day over a tunnel
– Incremental backups happen every day
– There is a RAID mirror in the server as the first level
– He uses Exchange online and so does not worry about backups

### Why backupexec?

– Wanted more granular restores of Exchange
– It thinks in terms of tapes

### Lessons

The cloud is convenient

Redundancy is important!

### Hardware confidence?

– We have spare parts for hardware servers
– We have virtual servers
– RAID hard disks in the servers
– Brendan wants servers that are under warranty
+ 5 year warranties built in
– TWC uses gamer-quality hardware with good reviews
– Custom-built servers: depends
+ Gamer-quality or low-end business quality
– We use Hyper-V for virtual servers
– Touch Machines bought in pairs
+ both machines run 24-7
+ AMD processors

### Backup Levels/options

– Encrypted USB keys
– DVD backups
– RAID arrays
– Previous versions/shadow copies
– Offsite backups
– Backups to internal drives
– Cloud backup

### How much should we trust encrypted backups?

– What if the company does the encryption?
– What if we did the encryption?
– Maybe it doesn’t matter: if we are on the Internet we are in the cloud
– There is consent vs non-consent
– Is Linux more secure?
– Should we throw up our hands and give up?
– FLOSS means that although people CAN examine source code it doesn’t mean that it WILL be

### Disaster recovery

– If you participated in disaster recovery other people would have to do this the next time
– They paid for another company to provide services for backups
– How do you set up disaster recovery environments?

Some people do not like backups because they can profit from the data recovery.