a little direction, for 2 FreeNAS, with replication

[ecourt]

I've hit the search button quite a few times, but I've not really seen anyone doing what I'd like to do. Since this is the 'new to freenas' section, I am building my first freenas server.

Id like to build 2 FreeNAS servers, that replicate between each-other

I have 2 ESX servers that are already setup to boot from local disk
Media/data is stored on freeNAS. If one goes down, the second FreeNAS can be picked up by ESX and everything keeps running.

Optionally (and something Ive not searched as much) I'd like to have tiered storage, where FreeNAS can tell what files are used often, and what files are not. Store frequently accessed files on fast storage, and infrequently used files on slower storage.

My level of technical experience/expertise is fairly high, as I've already listed ESX, I'm pretty good with Linux (centOS/Redhat being my favorite, bsd not bad, debain/ubuntu least favorite, windows servers have their place as well, I'm not an OS bigot (unless you talk about crapple, then just STOP)

so the questions are --
Are the replication goals achievable in FreeNAS ?
Are the fail-over goals achievable in FreeNas ? (this may be more of a ESX question that a FreeNas one)
Are the storage tiers achievable in FreeNas ?

Appreciate any direction/help you guys can give me.

 

[SweetAndLow]

This isn't the purpose of replication. Replication is not real time. The tiered storage is kind of how the arc or l2arc work. They keep most frequently used data available for reading.

Sent from my Nexus 5X using Tapatalk

 

[depasseg]

Are the replication goals achievable in FreeNAS ?

Yes Freenas systems can replicate data from one system to another. I believe the smallest snapshot and replication time window is 5 minutes.

Are the fail-over goals achievable in FreeNas ? (this may be more of a ESX question that a FreeNas one)

It depends on how ESX is setup. If this is just a network mount, and you can live with 5 minutes of lost data, then you could implement some sort of DNS failover to point from freenas1 to freenas2. There isn't any clustering built into FreeNAS 9.

Are the storage tiers achievable in FreeNas ?

No, not in a traditional sense. You can however improve performance with a low latency device (SLOG) for sync write workloads (which sorta works like a tier 0), combined with more RAM and a large (or multiple) SSD's as an L2ARC read cache.

 

[ecourt]

thanks for your response, great info. 5 minute snapshots should be fine, this is a setup for my home (a bit elaborate, but I have old toys laying around, so why not)
ESX, I was thinking ISCSI, but network mount would work as well. probably play with both just to see which works better.

you mention no clustering in freeNAS 9, is it in 10. I'm not against trying the beta out.
Im waiting on a few drives to arrive to start my configuration

On the tiered storage, yeah, as I said home setup, SSD would be overkill (and overly expensive) I've seen a few 'data management systems' that can show you a disk volume, and some items on that volume may be on fast storage, while other things in the volume (not accessed often) are stored on slower disk.
That's the kind of thing I was wondering if was available.
so if I setup a 8TB raid array, and have a single 8TB hard drive setup as well. infrequently accessed stuff could be on the single drive, while frequently accessed stuff is on the faster array, and I don't have to mess with manually deciding which goes where, the system determines it from how often the files are accessed.

 

[SweetAndLow]

thanks for your response, great info. 5 minute snapshots should be fine, this is a setup for my home (a bit elaborate, but I have old toys laying around, so why not)
ESX, I was thinking ISCSI, but network mount would work as well. probably play with both just to see which works better.

you mention no clustering in freeNAS 9, is it in 10. I'm not against trying the beta out.
Im waiting on a few drives to arrive to start my configuration

On the tiered storage, yeah, as I said home setup, SSD would be overkill (and overly expensive) I've seen a few 'data management systems' that can show you a disk volume, and some items on that volume may be on fast storage, while other things in the volume (not accessed often) are stored on slower disk.
That's the kind of thing I was wondering if was available.
so if I setup a 8TB raid array, and have a single 8TB hard drive setup as well. infrequently accessed stuff could be on the single drive, while frequently accessed stuff is on the faster array, and I don't have to mess with manually deciding which goes where, the system determines it from how often the files are accessed.

Why not just put that 8tb drive in the vdev and get the added performance and added storage?

Sent from my Nexus 5X using Tapatalk

 

[ecourt]

since the 8tb raid array is 5 2TB drives, I'd be wasting 6 tb of that drive, unless something has changed on how devices are configured in arrays.
again, the idea is to make better use of faster storage, and cheaper, slower storage.
Not to just throw more storage at the NAS, and needlessly store a file that's accessed once a year, onto a 5 spindle array.