Hi Everyone,
Drawing from my recent lab experiences I am curious as to how these open source technologies
play at a more enterprise level (TrueNas and XCP-NG).
My company is about to do a major IT reboot and our
planned approach was to use Azure HCI on clusters that are globally replicated.
I love the simplicity of TrueNas (and performance generally of ZFS) but we don’t really want to be locked into IX hardware just to get HA functionality. Is there another way of achieving this?
Similarly, I like what Tom has been saying about XCP being a challenge to VMWare, and the mentions of HCI on XDP sound fantastic, but from everything I see only I can’t tell if XOSAN v2 is a viable project and if it’s actually being actively developed. Do you know what the status is? Or what is the current best approach to doing this?
Many thanks as always – and @LTS_Tom I hope your shoulder is feeling better! Glad it wasn’t more serious!
1 Like
I have deployed Truenas enterprise hardware with 2 XCPng hosts for an enterprise business and it has been doing extremely well. We have pay for the premium license that has everything we need for backups, cloudinit and so on.
Unfortunately as of right now you can only get truenas HA with the enterprise hardware from IXSystems.
XOSANv2 is being developed, but I don’t know their priority on the project. Maybe a search on their forum will shed some more light.
unfortunately it’s a bit early to challenge VMWare in big appliances in my view.
If it would be 3-5 hosts, you can use almost everything as a solution. But if you scale it 3 or 5 times … no.
Especially (from your nic) in medicine with whole bunch of regulations, where GE and Siemens and other big guys rules.
I don’t look at using the hardware from IX systems as a lock in when it comes to an instant failover system, it’s all about custom Hardware having the features.
XOSAN v2 is being very actively developed I would expect to release maybe later this year.
Do you know the scalability of XCPng? I think the last time I looked you can have 64 hosts in a single pool.
I’m not sure what you mean when you are talking about the NICs. Could you expand on that?
First, it’s only my opinion.
Second. VirtualSwitch. Where it is in XCP-NG?
Third. Distributed Switch (across DC-s)
Fourth. vSAN
I know TrueNas is all we have, (I’m a fan too) but vSAN solution has its perks.
Fithf. Backup server for XCP-NG is impressive ( when @LTS_Tom speaks about, may Allah prolong his days) but I’ve being using vSphere replication for years. It’s my last stand, it helps me many times, even when I thought it’s all over.
Believe me, I speak only for production usage, where failure is not an option.
So whatever hypervisor XCP-NG may support, sustainability of construction as a whole is a key. Cost of ownership is also a key. (I mean “cost” widely inc hours of support team etc.)
And as I always say “trust” is keypoint. @LTS_Tom grows our trust to XCP-NG enormously … but not enough for me.
So everybody is free to listen or not to listen, and have right for theyr own mistakes and victoryes.
SY Trifon1
Yes, 64 hosts max per resource pool, here are the other limits
Great to hear about XOSAN v2!
Lock-in isn’t probably the right word. For us we have a bias towards
commodity servers due to a partnership with SMC and so where we
can we like to be able to custom build our hardware.
Just curious if anyone has experience with Starwind at all?
It’s been years since I tried their product but I did a deep dive last night and it seems like they provide a nice tie in with both hardware and hypervisors. Their product segementation (VSAN vs NAS/SAN vs HCI) is unnecessarily confusing but it does seem like they offer a lot of options.