Latest / The Nerdy Photographer Podcast / 180 - Backup or Bust – Building a Digital Safety Net

Transcript

180 - Backup or Bust – Building a Digital Safety Net · The Nerdy Photographer Podcast

• 0:02
  Casey Fatchett: Welcome back to the Nerdy Photographer podcast.
• 0:05
  I'm your host Casey Fatchett. Today, we're going to talk about
• 0:07
  the least glamorous, most consistently ignored, and
• 0:10
  arguably the most important topic in all of digital
• 0:13
  photography: data redundancy, backup systems, the safety net
• 0:18
  under everything you do. I know it's not exciting. Nobody
• 0:23
  watches a photography channel hoping to learn about hard drive
• 0:26
  arrays, but here's the framing that I want you to hold on to
• 0:30
  for the next few minutes. Every photograph you've ever taken on
• 0:33
  a digital camera that you care about is currently sitting on a
• 0:36
  physical object that will eventually fail. Not might fail,
• 0:40
  will fail eventually. It's going to happen. Mechanical hard
• 0:43
  drives have moving parts that wear out. SSDs have limited
• 0:48
  write write cycles. Memory cards get corrupted. Laptops get
• 0:52
  stolen, dropped, caught in house fires. Things happen.
• 0:56
  Eventually, that drive is going to fail. The question is not
• 1:00
  whether you'll experience data loss in your career. The
• 1:02
  question is whether you'll have a system in place when it
• 1:06
  happens. And right now, most photographers don't have
• 1:10
  anything. So, let's fix that. Okay, the three-two-one rule.
• 1:16
  The framework that data management industry has
• 1:18
  converged upon, and that professional photographers,
• 1:22
  archivists, IT departments use is called the three-two-one
• 1:25
  rule. It's simple enough to remember and comprehensive
• 1:28
  enough to protect against almost every realistic failure
• 1:31
  scenario. Here's how it works: you need three copies of your
• 1:35
  data, not one, not two, three. The reason is probability. Any
• 1:40
  single storage device can fail if you have two copies and
• 1:42
  they're both in the same location. A single event, a
• 1:44
  flood, fire, power surge, can potentially take out both
• 1:48
  simultaneously. Three copies, especially when distributed
• 1:51
  thoughtfully, means that two separate simultaneous failures
• 1:55
  would need to occur to lose your data permanently. That rate is
• 1:58
  rare, and it's enough to let you sleep at night. Those copies
• 2:02
  should live on two different types of media. This protects
• 2:05
  against a class of failures that can affect multiple devices of
• 2:09
  the same type at once. If all three copies are on the same
• 2:12
  brand of hard drive and that model has a firmware bug, which
• 2:16
  historically has actually happened, you could lose all
• 2:19
  three simultaneously, mixing media types such as an internal
• 2:23
  SSD, an external hard drive, and a cloud service means no single
• 2:27
  technical failure can cascade across all of your copies, and
• 2:32
  one copy needs to be off-site. This is the one most
• 2:35
  photographers skip, and it's the one that protects against the
• 2:38
  scenarios that feel unlikely that they will until it has
• 2:41
  happened: a house fire, a flood, burglaries, burst pipe. If all
• 2:45
  your drives are in the same building, they share the same
• 2:48
  physical risks. An offsite copy breaks that dependency. In
• 2:52
  practice, for most photographers, a three-to-one
• 2:55
  system looks something like this: your working drive where
• 2:57
  you're doing your editing, a local backup drive or NAS system
• 3:01
  at home, and a cloud backup running continuously in the
• 3:04
  background. That's three copies, two media types, one offsite.
• 3:08
  It's achievable. It's not particularly expensive, and it's
• 3:11
  the baseline you should be building towards. Let's talk
• 3:16
  about RAID. You'll often hear RAID recommended as a part of a
• 3:20
  backup strategy, and it's worth understanding what it actually
• 3:22
  does because it's frequently misunderstood in ways that
• 3:25
  create a false sense of security. RAID stands for
• 3:29
  redundant array of independent disks. It's a way of combining
• 3:33
  multiple physical hard drives into one single logical unit,
• 3:38
  and depending on the configuration, they can provide
• 3:41
  varying levels of performance and redundancy. The important
• 3:44
  distinction is between RAID levels. RAID zero stripes your
• 3:48
  data across two or more drives to increase speed, but it
• 3:52
  provides zero redundancy. If either disk fails in a RAID zero
• 3:56
  array, you lose all your data across both drives. It's
• 4:00
  actually riskier than using a single hard drive because you've
• 4:04
  doubled the number of components that can fail. Do not use RAID
• 4:08
  zero as part of any sort of backup strategy. RAID one is
• 4:12
  what's most relevant to photographers interested in
• 4:14
  redundancy. It mirrors your data. Everything written on one
• 4:16
  drive is simultaneously written to an identical second drive. If
• 4:20
  one drive fails, the other has a complete current copy of
• 4:23
  everything. You pull the failed drive, replace it, the array
• 4:27
  rebuilds itself for a working drive that you are actively
• 4:31
  editing from.
• 4:31
  A RAID one network attached storage device (NAS) is an
• 4:36
  excellent solution, but here's the critical point: RAID is not
• 4:40
  a backup. It protects against drive failure, but it does not
• 4:43
  protect against accidental deletion, file corruption,
• 4:47
  ransomware, or physical disasters. If you delete a
• 4:50
  folder on a RAID one array, it's deleted from both drives
• 4:53
  simultaneously. RAID buys you uptime and protection against
• 4:57
  hardware failures. It does not replace the other components. Of
• 5:00
  a three-to-one system, cloud backup. This is, and here's a
• 5:06
  really important thing to say about cloud backup. Sync isn't
• 5:08
  the same thing as backup. The offsite component of your
• 5:12
  three-to-one system is most practically handled through a
• 5:16
  cloud backup, and it's worth distinguishing between two types
• 5:18
  of cloud surfaces that look similar but behave very
• 5:22
  differently. Sync services and backup devices. Sync services
• 5:27
  such as Dropbox, Google Drive, iCloud are designed to keep your
• 5:32
  files consistent across multiple drives. They mirror whatever is
• 5:36
  on your computer to the cloud in real time. This sounds like a
• 5:40
  backup, but there is a catch. If you delete a file, or a virus
• 5:44
  encrypts your files, or you accidentally overwrite a folder,
• 5:48
  the sync service will faithfully replicate that change to the
• 5:51
  cloud. You, your corrupted or deleted files will be corrupted
• 5:55
  or deleted everywhere. Most sync services do maintain a version
• 6:00
  history and a recovery window. Dropbox, for instance, can
• 6:03
  recover deleted files or previous versions with a rolling
• 6:06
  window of 30 to 180 days, depending on what your plan is.
• 6:11
  So they're not completely without recourse, but that
• 6:13
  window again is limited. The recovery process is manual and
• 6:17
  cumbersome, and it's not designed for the scale of a
• 6:20
  photography catalog, true backup services work differently. They
• 6:24
  maintain version snapshots of your data over time, often going
• 6:28
  back months or years, and they're specifically designed
• 6:31
  for recovery, not just access. For photographers, Backblaze is
• 6:35
  one of the most practical options that backs up your
• 6:38
  entire computer continuously for a flat monthly fee, keeps a year
• 6:43
  version history and has a straightforward recovery
• 6:45
  process. It's not the only option, but it's worth knowing
• 6:48
  about. And if you want to check that out, I will have a link in
• 6:51
  the episode notes where you can use our special sign-up link.
• 6:54
  The initial upload can take weeks if you're starting with a
• 6:58
  large catalog, and that's normal. And it's a reason to
• 7:00
  start today rather than starting later. Okay, here's another
• 7:05
  option: cold storage. Many photographers also keep what's
• 7:09
  called cold storage drives that aren't connected to anything,
• 7:12
  sitting on a shelf holding archives of old work. It's a
• 7:14
  legitimate part of a backup strategy, especially for
• 7:16
  long-term archiving. But there's a maintenance reality that most
• 7:20
  people don't know about mechanical hard drives. The
• 7:23
  traditional spinning kind can develop a problem called
• 7:25
  stiction if they're left idle for extended periods. The
• 7:29
  read-write heads against sit against drive platters, and over
• 7:33
  time they essentially stick in place. When you plug the drive
• 7:37
  back in a year or two after a year of or two of inactivity and
• 7:41
  try to spin it up, the motor may not be powerful enough to break
• 7:44
  that contact, and the drive fails, not from age but from
• 7:48
  prolonged stillness. The fix is simple: spin up your archive
• 7:52
  drives at least once a year, plug them in, let them run for a
• 7:54
  few minutes, verify that all the files are still accessible, and
• 7:57
  then safely eject them. That physical movement is enough to
• 8:00
  prevent stiction from developing. While you're at it,
• 8:03
  do a spot check. Open some files. Make sure they're not
• 8:06
  corrupted. A drive that shows up and mounts successfully isn't
• 8:09
  necessarily a drive whose files are all intact. Trust me, it's
• 8:13
  happened to me. SSDs don't have this mechanical problem, but
• 8:17
  they have their own long-term storage quirk. Unpowered SSDs
• 8:21
  can slowly lose their charge over time, which can lead to
• 8:24
  data degradation. For true long-term cold storage, we're
• 8:27
  talking years of activity. Mechanical drives in a cool, dry
• 8:31
  environment are actually still more reliable choice than a lot
• 8:34
  of other things, which surprises a lot of people. Now, verify
• 8:40
  your backups. One thing the photography community almost
• 8:42
  never talks about: a backup you haven't tested is not a backup
• 8:46
  you can trust.
• 8:47
  It's entirely possible to have a backup system running for
• 8:49
  months, believe your data is protected, and then discover
• 8:54
  during a recovery attempt that the backup files are corrupted,
• 8:57
  incomplete, or simply not what you thought they were. Drives
• 9:00
  can fail silently. Software can have configuration errors. Cloud
• 9:03
  services can have sync issues that you'd never notice until
• 9:07
  you needed to recover something. The habit to build is this:
• 9:11
  periodically, every few months is a reasonable amount of time.
• 9:14
  Pick a random folder from your backup and restore it to a test
• 9:17
  location. Open the files. Make sure they're intact. This isn't
• 9:20
  being paranoid. It's the same thing as a competent IT
• 9:24
  department will do as a matter of routine. The five minutes it
• 9:27
  takes to verify a backup is insurance against the moment
• 9:30
  when you desperately need to trust that backup. Okay, the
• 9:35
  photographers who lose irreplaceable work are almost
• 9:38
  never the ones who didn't care about it. They're the ones who
• 9:41
  meant to get a better system in place and hadn't gotten around
• 9:43
  to it yet. Data loss doesn't send an advance warning, so
• 9:48
  today's homework is not it. Right now, before you do
• 9:50
  anything else, locate every drive you own and map out where
• 9:53
  each copy of your important work lives. Identify your single
• 9:56
  points of failure, places where one event could take out the
• 9:59
  only. Copy of something. If your wedding archive lives on one
• 10:02
  external drive with no backup, this is a liability that needs
• 10:06
  to be resolved today, not eventually. Buy a second drive
• 10:09
  if you need one. Start a cloud backup trial. Spin up any cold
• 10:12
  storage drives that have been sitting idle for a long period
• 10:15
  of time. And once your system is in place, set a recurring
• 10:18
  calendar reminder to verify it every few months, your future
• 10:21
  self-the one staring at a drive that won't mount-will be very
• 10:25
  glad that you did this today. Do it today, and until next time,
• 10:31
  stay safe and stay nerdy.