Backing Up VM Image Files to Internet Backup Services

Here in rural Ireland, I can only get 38/10 so the Virgin offer looks incredible to me. I had uploaded about 7Tb to Backblaze during my subscription, it took about six months but would tick along quite well after that. It depends obviously on what your overall amount currently is and how much you add on an ongoing basis.

One of the things I really liked about BackBlaze was that you could set it to use the whole of your upload bandwidth when required, eg first upload, compared to CrashPlan where I never managed to get it to upload at more than about 30% of my max.
This was a couple of years ago when I ran them both (different computers) for a year as a trial.

New letter from CrashPlan. They say they made two mistakes in the recent letter about stopping some files types. The first mistake was sending the letter to the wrong circulation. The para about the second mistake says:

"The second mistake involves the actual file changes that we made. As part of this update, we stopped archiving 32 file types and directories. The email notification included a link to an updated list of files that are excluded from CrashPlan backups. One of the file types we began excluding from backups is the .sparseimage file format. We believed that this file format was obsolete because in 2007 Apple introduced a new format called .sparsebundle, which we thought replaced .sparseimage for the use case we track. After we implemented the changes in May, some of our customers made it clear they still have valid use cases for .sparseimage. We now believe we made an error in excluding .sparseimage, and we have since added it back to the list of files we support via backup. "

Looks like my sparsebundle will be OK.

2 Likes

This all sounds like a solid justification for not using cloud storage solutions for backing up your stuff. You may make individual selections for Drop Box, but the ā€œbigā€ services like Backblaze, CrashPlan and Arq are limited by bandwidth constraints and well as capacity concerns. The point has been made that just about any broadband service limits upload speeds dramatically. In the past I seem to remember that Crash Plan could retrieve a hard drive from a client to make their first backup. But since they no longer backup up system files and applications, that would seem to be problematic.

The problem seems to me to be twofold: First, the process of checking files for changes being slow; the second is upload bandwidth. Question: if you upload a sparsebundle of your entire system, is that replaced every time you upload a new one, or are they saved as versions of your system? The sparsebundle does seem to be a way around their exclusions. And they would download more quickly than they uploaded; it might be a viable way to restore your system. That is, if, after a disaster, you still have a computer to download and restore it to, and broadband to download it with.

My system has gone wonky on me a few times, and I needed both TimeMachine to restore from an earlier date and a clone to get my more recent data (because the clone was also corrupted and wouldnā€™t boot). But for the most part, Time Machine would seem to be the most reliable way of saving and restoring dataā€”though I have my VMs excluded from my Time Machine backups, as I donā€™t have that much hard drive space to spare. They are included, however, in my clones. Even so I somehow lost a copy of my Windows 7 VM. And I havenā€™t been able to get it to reinstall. Fortunately I upgraded to Windows 10 before I lost 7, which had the license that enabled me to get 10 for free.

All that said, I keep clones of my system and most important data, onsite and in a safe deposit box. But I donā€™t see a viable reason to use cloud storage. I even use iCloud with restraint, as Apple has never done cloud services very well.

I apologize for this long post, but after reading the article all the comments ahead of me, I had a lot to think about.

1 Like

Hi Jeff, see my comments earlier in the thread about this. The answer is NO !

As sparsebundle (as opposed to a sparseimage) breaks the image up into thousands of ā€œbandsā€. CrashPlan only uploads the changed bands. In practice CrashPlan only uploads a few GB of changed bands in the sparsebundle of my entire system every day. But I would use Time Machine or a local clone to recover before I used the CrashPlan sparsebundle, which is the earthquake and fire offsite backup.

A sparseimage would upload the entire system everyday.

A general comment is that speed is not an issue for these backup services. It all happens in the background with no detectable impact. I would be very upset if my sync services like Dropbox were as slow.

You can of course be just as selective in what CrashPlan backs up as you can with what Dropbox syncs.

Donā€™t apologize; itā€™s a great thread, and even though Iā€™ve never used a VM, this is a very helpful and informative thread. At a few different jobs, my husband and I both found offsite backup services to be more trouble and time consuming than they were worth. If it wasnā€™t for IT people being able to futz with recovery after hours, we probably would have jumped out windows. So we also just do time machine and time capsule backups as well as to removable hard drives. Itā€™s easy, brainless, relatively quick and saves money.

Thatā€™s really not the intended takeaway. The point is to be aware of the limitations inherent in any portion of an overall backup strategy and to ensure that the strategy as a whole is providing the level of protection you need. Itā€™s far better to use a cloud backup service (as a tertiary backup) that might not back up everything than it is to have no offsite backup at all (or one thatā€™s updated too infrequently because itā€™s too much trouble).

1 Like

Personally, I donā€™t see storing a clone in a safe deposit box as viable. I live only a 10 minute cycle from the bank, but I would never manage to regularly update a clone and take it to the bank, so it would end up hugely out of date. I struggle to keep my on-site clone updated more than once a month. My network Time Machine and Crash Plan backups, on the other hand, are constantly working in the background, and Iā€™ve used them in the past to quickly retrieve files I somehow deleted or needed a previous version of.

So I see cloud backups as ideal for an off-site backup that is kept up to date, as @ace says. Definitely not as the only backup though! But once the initial backup is done (over weeks in my case), the basic upload bandwidth available seems sufficient to keep cloud backups reasonably up to date (and far more so than a safe deposit box copy would be).

Saying all that, after this article (and the fact CrashPlan never informed me directly about the changes even though Iā€™ve been a customer for years!), Iā€™m planning to switch to Arq backup.

1 Like

Iā€™m sort of glad you said that since it answers the question of why we should cover things that the companies are theoretically informing all their customers about. But I am disappointed in Code42 these days.

1 Like

Very interesting article and comments. Iā€™d never really considered a personal Cloud backup solution, having multiple SuperDuper and Time Machine backups on site. Because of this thread I started looking at costs and reviews of various options starting with thewirecutter.com. Their take is that Backblaze is overall the best option, but iDrive comes in a strong second.

I donā€™t see iDrive mentioned here at all, so Iā€™m curious if anyone on this list has experience with that option.

And, I see that many find Arq to be an excellent solution but thewirecutter.com excludes it because it requires more fiddling and technical nous than the ones they recommend. What do those who use it think?

We wrote about Arq a while back.

Note that Arq now offer an integrated cloud backup with a simplified app where you pay $6/month per terabyte. Iā€™m not sure this existed when Glenn wrote his article, and Iā€™ve not used it. But from the sounds of it, itā€™s a like the other backup services in terms of being simple setup and integrated.

https://www.arqbackup.com/arqcloudbackup/

I was about to respond to @jeff5 with a similar answer, @jzw, so thanks to yours, I can keep mine much shorter. :wink:

Another key advantage of cloud backups is that they ā€œtravel with you.ā€

Yes, Time Machine does manage local backups on your Mac when the machine is not connected to the TM drive. Thatā€™s worthless, though, if the machine and/or drive fail before youā€™re back home to sync those files with your TM backup storage.

In contrast, cloud backups work just as well regardless of location as long as you have reasonably fast internet connectivity. Thatā€™s an added layer of safety I wouldnā€™t want to miss.

1 Like