It's a worthwhile poll... The issue is integration into other platforms... Simply put my customers use VMware but are scared... I work for a partner... Until veeam and other release integrations into promox, scale, and others; it will be a non starter... KVM has been around forever but lack of enterprise tool kits are slowing transition.
-
Important Announcement for the TrueNAS Community.
The TrueNAS Community has now been moved. This forum will now become READ-ONLY for historical purposes. Please feel free to join us on the new TrueNAS Community Forums
VMware Alternatives
- Thread starter wsoteros
- Start date
I am not an expert on this matter.
Having said that I don't think Core is comparable to VMWare.
It just lacks a lot of features. At least I don't know how to share a GPU with many Windows VMs or similar.
In the past I had annoying issues with VMs not properly starting after installation, which could be fixed, but why should I?
Also many drivers and tools for FreeBSD are simply worse than under Linux or simply not available.
Sure you could use Core to run a Linux VM that runs Kubernetes, but it's just worse than most other solutions.
Or did I misunderstand something?
Scale on the other hand could be a potential replacement in the future.
I guess that's why they are checking the competition to see if they should make Scale more proxmoxy (e.g.).
- Joined
- Jan 11, 2021
- Messages
- 415
Nor am I an expert. And I agree: It's not comparable. But it's not true (as a general assumption) that bhyve "lacks features" per se. It depends on your perspective (i.e. use case). Some enterprise appliances can benefit from VMware, others from SmartOS (KVM and bhyve) or TrueNAS Core (bhyve and iocage).
ESXi, for example, *cough* "lacks" *cough* drivers even more than FreeBSD (compared to Linux). The stability of bleeding edge drivers is another matter. The (long term) availybility of drivers or the compatibility of the userland is another (Linux e.g. has no standardised userland, it's just the kernel). Are management tools bundled or do you have a choice? Do you want to invest in staff training? Do you need to develop things yourself for your particular use case? What about licensing? And so on...
If the features you need tick the boxes, then anything might be good enough.
ESXi, for example, *cough* "lacks" *cough* drivers even more than FreeBSD (compared to Linux). The stability of bleeding edge drivers is another matter. The (long term) availybility of drivers or the compatibility of the userland is another (Linux e.g. has no standardised userland, it's just the kernel). Are management tools bundled or do you have a choice? Do you want to invest in staff training? Do you need to develop things yourself for your particular use case? What about licensing? And so on...
If the features you need tick the boxes, then anything might be good enough.
firesyde424
Contributor
- Joined
- Mar 5, 2019
- Messages
- 155
Is this sarcasm? Always hard to tell over the internet.
- Joined
- May 17, 2014
- Messages
- 3,611
Does VMWare support live migration from 1 host server to another?
From my understanding it can, and do so automatically for load balancing. (VMWare over subscribes it's CPU resources...) In fact, a few years back I vaguely recall this biting me. A destination host for a Linux VM was setup for a rare, specific VLAN. But that destination host's network side was failing it's LACP. Thus, the Linux VM client would not run on that second VMWare host.
I know that Solaris LDOMs and AIX LPARs both support live migration.
Obviously both host servers need access to shared storage, the same VLANs, adequate memory & CPUs. Plus, dedicated resources like a USB license device, or access to specific GPU are not going to be easy. But most VMs are just plain computers, performing some task which don't need dedicated GPUs or USB devices.
From my understanding it can, and do so automatically for load balancing. (VMWare over subscribes it's CPU resources...) In fact, a few years back I vaguely recall this biting me. A destination host for a Linux VM was setup for a rare, specific VLAN. But that destination host's network side was failing it's LACP. Thus, the Linux VM client would not run on that second VMWare host.
I know that Solaris LDOMs and AIX LPARs both support live migration.
Obviously both host servers need access to shared storage, the same VLANs, adequate memory & CPUs. Plus, dedicated resources like a USB license device, or access to specific GPU are not going to be easy. But most VMs are just plain computers, performing some task which don't need dedicated GPUs or USB devices.
- Joined
- Nov 25, 2013
- Messages
- 7,776
No. Single node hyperconverged - most flexible and reliable platform I ever had.
firesyde424
Contributor
- Joined
- Mar 5, 2019
- Messages
- 155
What do you do for HA? Migration and maintenance? Performance monitoring? Analytics? Capacity projection? Contention?
firesyde424
Contributor
- Joined
- Mar 5, 2019
- Messages
- 155
For clarification, genuine questions. I've not considered Scale for this before.
- Joined
- Nov 25, 2013
- Messages
- 7,776
Monitoring: TrueCommand and Observium
HA: nothing - I wrote "single node"
Desaster recovery: we run two machines, snapshot and replicate all VMs from the active node for that VM to the inactive one. Should one machine fail, I can manually boot all VMs in half an hour or so
Service like Domain Controllers simply run one VM on each machine.
HA: nothing - I wrote "single node"
Desaster recovery: we run two machines, snapshot and replicate all VMs from the active node for that VM to the inactive one. Should one machine fail, I can manually boot all VMs in half an hour or so
Service like Domain Controllers simply run one VM on each machine.
- Joined
- Nov 25, 2013
- Messages
- 7,776
P.S. I run all this on CORE, not SCALE.
depasseg
FreeNAS Replicant
- Joined
- Sep 16, 2014
- Messages
- 2,874
Hyperconverged? A stand alone server with a lot of storage isn't what I would consider hyperconverged. Show me a 3+ node TrueNAS cluster that presents a single storage pool across all three nodes and allows me to move a running Virtual Machines from node to node, and then I'll stand corrected.
firesyde424
Contributor
- Joined
- Mar 5, 2019
- Messages
- 155
Niche case, I'll grant you, but we do have a use case that requires console access. We're an archive vendor and the various stages of that process require access to the customer's environment. Each customer (hopefully) has a secure method of required access to their systems which usually boils down to VPN software. Typically, that means connecting to the customer's VPN gateway with their preferred software, which kills all local LAN connections, preventing RDP.
I engineered a system several years ago using VMware ESXi and linked clones. There is a single master template containing all versions of the software required for access to customer systems. Rather than maintain and update 90+ VMs, they are considered disposable. When Windows updates must be run or there's a new version of (insert name of VPN software here), we update the template, power off and delete the 90+ VMs, then recreate them all from the template. Total time from start to stop is under 10 minutes.
These VMs must be accessed via their VM console because RDP will not work once they connect to a customer's systems.
- Joined
- Nov 25, 2013
- Messages
- 7,776
Hyperconverged means storage and virtualisation on one system, not necessarily clustered. And yes, it cannot do that. Nonetheless it has proven to be a perfectly stable solution for all my company's virtual workloads including our domain controllers and similar mission critical things.
ZFS and hourly snapshots and replication makes up for the lack of clustering. For us, not necessarily for everyone.
I was just a wee bit irritated that the initial poll did not even mention TrueNAS.
ZFS and hourly snapshots and replication makes up for the lack of clustering. For us, not necessarily for everyone.
I was just a wee bit irritated that the initial poll did not even mention TrueNAS.
firesyde424
Contributor
- Joined
- Mar 5, 2019
- Messages
- 155
We have no choice. It's not so much the changes to ESX as the changes to vSAN licensing. I'm a big fan of VMware vSAN and, until recently, a big evangelist for it. However, we are extremely storage heavy for our size and the changes to vSAN licensing means that our licensing costs are set to increase 4-5X depending on licensing model, which isn't sustainable for us. We're moving away from VMware, we simply have no choice.
- Joined
- Nov 25, 2013
- Messages
- 7,776
That sounds like a support and deployment nightmare. With a clever solution, granted.
We have gateway to gateway VPNs and SSH access from the developers' and operators' machines to the infrastructure we manage. I guess I can consider us lucky.
- Joined
- Nov 25, 2013
- Messages
- 7,776
Drums. Drums in the deep. They are coming.
- Joined
- Nov 25, 2013
- Messages
- 7,776
BTW - the VPN client systems will have to keep the connection to their default gateway up and even have a host route for the customer's VPN endpoint, so they cannot kill *all* LAN communication.
Have you tried NATing the RDP connections to the default gateway on that very gateway?
firesyde424
Contributor
- Joined
- Mar 5, 2019
- Messages
- 155
When I say "kills all LAN connections", I mean that it redirects all LAN traffic through the software VPN tunnel. We're also talking hundreds of customers, all with different requirements. Just off the top of my head these systems have OpenVPN, Cisco Anyconnect, VMware Horizon client, Citrix Workspace, FortiClient, Ivanti, Pulse, and a half dozen others, if not more. There's simply too many configurations to manage this way and we just end up having our employees use the console for these specific systems.
