There's definitely a CONNECTION between those two pieces of TECHNOLOGY...

I was so disappointed to learn the J series won't do half of what I bought synology for. They really need to make it easier to select the right models.

​@@jeremiahbullfrog9288 + model or non plus model (just don't buy J or non plus)

​@@jeremiahbullfrog9288They have a selection tool on their site that asks your use case. It could be more comprehensive but like dsm they try to keep it simple.

Nice! I guess if you think about it why wouldn't it work with a Celeron processor.

@@theroboticscodedepot7736 that was exactly what i thought.

@@theroboticscodedepot7736 according to official intel documentation J4125 only supports 8GB of RAM.. but I'm running 20GB (4GB onboard + 16GB module) just fine

You can increase any of them to 20GB if you have the right type of ram

Can you install a pci-e card tho?

Once copied to the backup device the backup program should not need to copy all 10TBs every time but only the files that have changed or are new files. I decided to use two external USB connected M.2 NVMe 2TB SSD drives for my main backups which allows me to keep one of them offsite. The backup program runs daily and I switch drives out every two weeks. For 10 TB's you could use an all SSD NAS like the Asustor flashstor or the UGREEN DXP480T to handle the 10TB of backups.

The replication is done at the fs / block device level. When any failure occurs, the passive NAS starts up all of the same services on the same dataset and takes the cluster IP address, but anything in memory (including user sessions) is lost. Since active sessions are lost, the not-google-drive will make you login again when failover happens.

@@apalrdsadventures ah ok, that's rather unfortunate. With both devices effectively having the same IP, I thought it was similar in implementation to how OPNsense handles ha and fail over. Is also a bit sad, since it seems most of the work is done, and just handling at least session hand over seems like a comparably small problem? But all that being said, it's still a very low price point to get any kind ok ha that kinda "just works". Which is also neat.

It's not nearly as simple as it sounds. In OPNsense, they are syncing the config changes with xmlrpc (not frequent), firewall state table with pfsync, and sharing the firewall IPs (CARP). When a CARP failover happens, the user session table is already entirely in sync so nothing needs to change. They aren't syncing state for any other processes, so things like DNS caches aren't being transferred. Here, they could transfer the virtual IP quickly, but all of the services running on the device each have their own state memory (i.e. Samba has active TCP sessions which won't transfer), and trying to sync all of those separately is a bit of a nightmare. Keeping memory in sync takes a lot more bandwidth than keeping disk in sync, but there are transactional database systems which try to do it. Every service running would need to be modified to deal with this, or they'd need to use services which are natively clustered (i.e. store all user state in Redis, and duplicate that). It's certainly not a small task.

Yes. Using rsync, one NAS of almost any brand can push or pull files to another NAS of almost any brand.

@@jeffnew1213 Thanks! I will look into rsync

@@jeffnew1213 Thanks! I looked at rsync and it appears to be the equivalent or the Windows program RoboCopy which I use to migrate backup files.

That shit is way too expensive for just a piece of software then

Lol that's crazy they expect you to buy their hard drives.

@@joeykeilholz925 for quite a few models, i think 8 bays and up for HDDs and for NVMe pool storage the requirement is a lot stricter, you basically have no choice but to, if you want access to all the available features of the unit! Bloody ridiculous for anything other than, and even then it's a bit of a push, enterprise level customers.

Which this? TrueNAS has great replication features, the rest not so much

The really big problem with btrfs (and why it has a reputation for 'eating' your data) is mostly because it doesn't solve the RAID5 parity 'write hole' problem, which happens with all parity-based RAID unless they are specifically designed to address this. tl;dr when writing data and recomputing parity, if all of the disks don't complete their writes, the parity will be inconsistent and the block lost. HW RAID traditionally used battery-backed RAM to prevent inconsistencies on disk (by fixing them on the next boot), zfs will always copy-on-write which can fragment the data in rare cases (which has negative performance penalties if the case actually happens, but is entirely safe and normally as fast), and btrfs will normally copy-on-write but potentially lose data under scenarios where a tiny write happens across a small block, in their design to avoid fragmentation. Linux mdadm (basic 'software raid') originally didn't solve this either, but has the option of journaling writes to avoid it (this comes with a performance penalty also). In a mirrored setup, btrfs should perform about as well as zfs and mirrors aren't affected by the parity RAID issues.

I have a QNAP NAS and for the most part I have been very pleased with it. I did run into an issue with their backup program HBS 3 (Hybrid Backup Sync 3) and it's still an ongoing issue with them but I did find a workaround for now.

It's odd that btrfs even lets you disable it, there's no reason to disable a feature like that

@@apalrdsadventures You should ask the same about ZFS. zfs set checksum=off Edit: Per Synology documentation - checksum is not recommended on folders used for databases, virtual machines, surveillance station recordings or anything that requires small random writes.

IPv6 only causes problems for network admins who disable it out of spite

@@apalrdsadventures consumers of these Synology use these mainly an only need local access don't need to be confused with ipv6 address

I've been using them since they came out and I agree 100% that they are moving more toward the business/enterprise end. I think there is a price gap in the market that they are trying to slide into for small/medium businesses and even less traditional larger businesses. With that said, I hope they don't abandon the home user market completely, it is nice having the same software on all of their units.

He tried to fight me over that chair! I unboxed later

Well it is his House ..

The WD Red Plus is CMR. The WD Red is not. WD definitely made this intentionally confusing.

You won't even get 10G replicating between two with something like Truenas and spinning drives unless you add a fast slog which should be mirrored and have plp, or have a lot of drives.