Remote backups with Backblaze B2

timbr

The roadmap states "S3 (and compatible) backup target". In some ways B2 is S3 compatible, but in many it is not. Backblaze B2 is a much more affordable place to store backups. They are big, reliable and have a solid API. I'd love to see B2 support be treated as important as S3 support.

WebInfra

We've used B2 extensively with backup applications which advertises as "S3 Compatible" like JetBackup and it worked fine.

One important thing to note is that B2's overall performance is nowhere near some other S3 players like Wasabi so you may encounter problems and errors if you backup a large amount of data within a short period of time.

alenguav

I currently use Wasabi with Jetbackup (and cPanel)
and was wondering how compatible is Enhance with it
as I only see Backblaze mentioned in this documentation
https://enhance.com/docs/backup-role/s3-backups.html

MediaServe

alenguav Use native Enhance backups. We use JetBackup on cPanel but Enhance backups trounce JetBackup. You just need a ubuntu server with a raw unformatted partition and space for backups. There's no better backup system, so forget S3 compatible backups and just rely on Enhance native backups. So easy to restore too!

My $0.02 anyway.

Dan

Completely agree with @MediaServe

The Enhance backup system works very very well and I really don't understand the want for S3 based backups.

MediaServe

Dan Perhaps S3 style backups are a second layer of backup redundancy for some, but for me I would like to have native incremental backup redundancy to 1 or more Enhance backup servers rather than S3, which is always full backups.

twest

MediaServe being able to run duplicates of enhance backups servers would be great for additional redundancy. We need more options for the config too, like lower retention, I don't need 7 days of backups for enhance 😢

MediaServe

twest Incremental backups on btrfs are so fast and efficient why not? We have retention at max currently (90 days) and backup daily only during overnight hours without any problems. Might have to deploy extra backup servers as the Enhance estate grows but we have nearly 200 websites migrated from cPanel now (slowly one at a time.)

Every time I help a client restore from backup the timestamp is always within minutes of midnight which is surprising. And our backup server is in a separate data center while our Enhance servers are on a single hardware node with 7 Enhance VMs and a NIC that is still hardware limited to 1 gbps. (We colo our own gear but use Vultr for offsite backups using their block storage provided on less expensive HDDs vs SSD/NVME.)

twest

MediaServe nice strat! I've also setup the backup server offsite at another DC for that little bit of added protection. It's on a 10Gbps port and running NVMe drives so the bottleneck is going to come down to the hosting node - not backup server (since in a disaster we want everything to run at top speed to recover asap).

I'm mostly interested in the Enhance backup as a server recovery mechanism - not really "website backup" specifically, we have other website level backups that are easier to manage for 90 days 'long term backups'... What I want from Enhance backups is to run it as often as possible (currently at 3-hour intervals) and only keep 1-2 days of backups, so in event of a server level disaster we can quickly redeploy the backups to a standby server that's already setup and ready to go. In this setup our clients might lose a maximum of 3 hours of data, and recovery should go super quick since all sites have dns managed via Cloudflare addon which updates in a snap. The bottleneck is how quickly the standby server can download the backups off the backup server, maybe redeploying a 700GB of data server will take less than 30 mins, that'd be great recovery time.

8Dweb

twest Not sure how others handle this. We have full block level backups of all our VM's and move those backups twice a day to one of our other data centers. We can then bring that copy back online by either restoring to the original site or to one of the other data centers.

We even take full copies of the Enhance backup servers, to make sure we have the ability to restart the entire Enhance cluster VM's anywhere, and keep going.

I realize this may be larger than some hosting companies can handle.

In our case most of our cloud structure is built out using XCP-NG (we moved from XenServer years ago)...

If you are able to have two decent hosts in two data centers - this is easily achieved. In our case, the hosts and SAN's are all connected at 25 and 40GB internally, then 10GB between data centers via dark fiber or MPLS. Since these are private lines, there is no public transit of our larger backup infrastructure. That being said, it is overkill for a smaller host.

I have now restored a specific customer's Enhance backup about 5 times, and one of my own portal sites once.. Each time the restoration of around 4.1 GB per site has been under 5 minutes. I am impressed with Enhance' backup/restore.

I think with 2 servers, geographically separated, you could run ProxMox, XCP-NG or other Hypervisor, setup a portion or full Enhnance Cluster and then keep copies of those VM's in your other data center. If you have a larger infrastructure two hosts plus a SAN (or NAS - i.e. TrueNAS) box would make a great strategy to keep things running.

Matt

twest

8Dweb I thought about running proxmox to have capability of block level backups - but rather not add another layer of complexity on servers. It's just a lot easier working with full bare metal, less overhead, less moving parts to manage, easier sysadmin duties.

We're doing 3-2-1 backups, with a proprietary incremental site-level backup that goes to S3, plus the enhance backup server. Our main/old hosting performs 3 daily backups with 2 going off-site and 1 local. I don't think the strategy we have with enhance is "lesser" since we're running the enhance backups tri-hourly to have very "fresh" backups all the time.

Alas, it's like they say "the hosts backups are not your backups", we don't have a robust backup system in place specifically for customers - it's mostly just for us, so we can ensure customer's data is as protected as possible, and within reason.

Nice to hear so many hosts around here take backups so seriously 🙂