How does SCALE scale?

[vitaprimo]

I redeployed a tiny TrueNAS appliance I had removed a while back, at least a year back. TrueNAS CORE and SCALE were still fresh, SCALE's installer would crash half the time. SCALE is now official and I figure we'll end up there anyway so I got that this time.

Aided with my notes from last time I flew through the config with time to spare, even after the unexpected missing gpart util from FreeBSD—I managed with gparted Live and fdisk. With all of the extra time I got exploring what makes SCALE, SCALE (i.e: ˝SCALE" and "scale") but I found no obvious clustering/scaling options.

How is it that TrueNAS SCALE scales again?

I read something about Gluster, something HCI and something about Kubertenes. That could mean to something like vSAN/DFS/S2D/SOFS, something like vCenter (VM orchestration), something like a Kubernetes node (container orchestration), or all of the above.

Maybe settings are exposed only when related services are enabled (could it?) — I'm only using iSCSI, SNMP, AD-joined. NFS, SMB and most others are disabled.

Thanks. :)

 

[danb35]

IIRC, most of the clustering functionality is controlled through TrueCommand, so you'd need to have that installed somewhere to handle that.

 

[sretalla]

The current stable build (Angelfish) supports (via TrueCommand) the clustering of storage via glusterfs and the clustering of SMB shares, which have been aggregated into the same workflow recently with the release of TrueCharts 2.2.

The next build currently in nightlies (Bluefin) will support clustering of Apps via kubernetes.

 

[morganL]

There is also the option for scaling out using Minio as an App on multiple nodes.

 

[vitaprimo]

Oh I see… I secretly kinda was hopinhh for a built-in multi-master or distributed orchestrating <insert_word>*, but that's something silly I've been daydreaming about for a while for nearly everything that comes in groups.

I guess I'll stick with vSAN for now. TrueNAS handles storage for the hypervisors themselves; ESXi is I think 400MB so I'm in no hurry to scale anything. I'm [loosely] watching out for Kubernetes implementations though, I'm curious which one will be the first that will allow for any unskilled user to grow from a single self-container Kubernetes cluster one host at a time without planning ahead.

VMware's Tanzu is sort of there but there's still the decision of self-contained vs cluster even if the cluster only has one host. Pay no attention to my nonsense. :)

Thanks for answering guys.

*: for something that's neither an appliance(VM/container) nor lives entirely on a single host, but rather it splits itself and/or multiplies parts of itself when redundacy or extra oomph is needed(distributed)

 

[Patrick M. Hausen]

I secretly kinda was hopinhh for a built-in multi-master or distributed orchestrating <insert_word>*

TrueCommand

Public documentation for TrueCommand, the TrueNAS fleet monitoring and managing application.

www.truenas.com

 

[sretalla]

VMware's Tanzu is sort of there but there's still the decision of self-contained vs cluster even if the cluster only has one host. Pay no attention to my nonsense. :)

I think maybe you're seeking the kind of experience that comes from running Rancher running on top of Harvester... nice run-through here: https://www.youtube.com/watch?v=tVsMen_e6OI

I don't know if it's even on the roadmap for SCALE to go as far as that.