JFYI
1 Like
That there 10x as many active installations on an unsupported version of NS than on the currently-supported version doesn’t seem like a good thing.
It would be interesting if the statistics were complete so one could tell the geo location of the various installations and versions.
IMVHO this is consistent with several technical changes applied to the project.
Edit: It’s not shared if these numbers are storical (every installation reported) or some sort of live (installation expires after x months of not phoning home)
It’s steadily decreasing, which is a good thing, it’s not possible to mass-switch all the machines at once and unfortunately this data is not entirely apples to apples comparison:
- If you have a NS7 machine that does both firewall and service provider, you have to split them or virtualize them. This manages to generate attrition on the commercial side of things, since the installer need to make the end user pay the maintenance for the installation. (maintenance is a carefully chosen word, since some are reluctant even to pay for regular servicing of the machine installed)
In Italy, since most of people is a tech illiterate, for tech we adopt the following: "If it ain’t broke, don’t fix it”. You can indeed interpreter that as an “won’t upgrade, don’t care”. - Pike is indeed right, the technical changes didn’t contribute, on top of that many have existing integration to update to the new format.
- Integrations is another great stopper: some are struggling to migrate not because we broke their integration, but because their integration runs on some ancient software that used to run fine on the 6.10, and pretend to keep it. PHP 5.4, Python 2… you name it.
- We might not end up with the same number of machines, because counting got switched. Now, a NethSecurity translates to a NethSecurity and that’s fine, but a NS8 installation can be one node, two, or even 6 nodes, on the stats, however we only count one. In addition to that multi-tenancy of NS8 is being used at its fullest, we’re seeing shutdown of multiple NS7, for a single NS8.
It is live, the unit gets excluded if it doesn’t ping every 48 hours. Ping is every 24~ish hours, depending on what you installed
2 Likes
Not only in Italy it’s - “never touch a running system”. NS7 compared to NS8 and Nethsecurity was easy to install and maintain for SME.
You should also consider, some simply do not want the new technique. Stephdl mentioned his personal experience he made on a meeting in discussions, somewhere I don’t rememeber.
In fact, it doesn’t matter anyhow.
So you’ve never driven a Fiat in your life? 
That’s the point, some don’t want to learn a new thing or it’s too much hassle to do with the new toolset and they keep the old version, I think it’s fine both ways 
510 Neth6 still alive??? Is close of 1/6th of your current NS8 install base!
For all the other folks: 93% of NS8 installs are 1 node clusters. Biggest advantage of NS8 is simply ignored.
Versions older than that don’t have the ping to the phonehome, trust me, they are still plenty out there. This sadly applies to every software.
It is, we work with SMEs and technicians usually don’t even bother to learn and use the tool provided well.
Again, sadly, this applies to everything, not only us.
The small percentage that uses the multi node cluster have reached very low cost of maintenance. For example, they have reduced and consolidated installations (from the NS7 in-place at the customer, to a cloud NS8 node), this allowed them to reduce hardware costs (nowadays cloud hosting is very cheap) and reduced maintenance time
An addition to that, if you haven’t noticed, the grafana dashboard is still sluggish, I have plans on improving that with even additional infos on the features used by the installations.
I (and the team too) think that hiding such stats is pointless, this helps us (and you all, as @pike pointed out) where we should point our efforts next
1 Like
… no? I mean…
Project decided to change the product. NS7 is one man band distro, NS8 is cluster management tool. Project did not found customers, because former ones did not considered the change.
As stated before, one host cluster is an overheaded cluster, this reflect in CPU and disk space occupation compared between NS8 and NS7.
I hope that the acknowledgment and the awareness of possibilities that NS8 can achieve the deserved acclaim and diffusion.
However the project left adopter in the dust and it’s trying to find new ones, “helped” somehow by the big bad evil overlord known as Broadcom.
The race nowadays are with bigger fishes: XCP-NG, ProxMox, Docker, VMware and currently NS8 delivers only NS8 modules, no other containers. So any small company tinkerer comparing ProxMox and NS8…
NS8: works in a click, but i can use only the “approved” modules, if i don’t want to develop.
PM: take a longer road, but it’s hypervisor, cluster management and container manager. And I can install LXC, Docker and at least another container option.
Without NS7 to NS8 upgrade module… why any Ns7 adopter should do something different than backup old metal and restore old install into ProxMox Guest? 
In current state, NS8 is a nice solution for a problem (or customer) yet to be found.
Only if, in your mind, clustering is the “biggest advantage of NS8.” And I don’t think that’s even close to being the case. An advantage to be sure, but far from the biggest.
What’s the biggest advantage? Well, a base OS that’s still receiving security updates is a strong contender. So is being, at least slightly, independent of the base OS. More sensible management of TLS certificates. Multitenant email. I’m sure I could come up with other advantages to NS8 (vs. NS7) that are more impactful than clustering; these are just the first few that came to mind.
I suppose if you’re trying to run it on an old i386, where every cycle of CPU and every byte of RAM are precious, the new architecture might be a problem. But when even an older Raspberry Pi can handle it, IMO that objection is grossly overblown.
Yep, that’s the tradeoff. Just as it was in NS7. And NS6. And SME. And e-smith. What you gain in ease of use, you give up (to a degree) in flexibility.
Yes, it’s a completely different product. A few of its capabilities overlap with NS8; most don’t. And its container option is LXC and only LXC–such support as it has for Docker containers is to convert them to LXC and run them that way.
You know this exists, right? Not a straight “upgrade,” but migration?
LOL troll. 
My Prox cluster with 2 nodes beats your NS8 clusters hands down for every aspect of ops.
@dan said it all.
Well spoken.
Speaking (many times) only for me and my SME companies: a migration to something new or different does only make sense to me, if there is an improvement to what we did before. As long as there are no changes in the way we make money, we keep the IT profile as low as possible. We are producers. We print, we drill, we punch and we mill. Yes, we need a LAN. No cloud at all. Of course, we do need email. IT (as well as our machines) has to be rock-stable and safe. There is no time for breaks at all.
Everybody has different needs depending on how you earn your money. For those who need an orchestrator NS8 is surely a perfect choice. IMVHO the target group is shifted away from SME. IDK if this decision will pay off for Nethesis in the future. BUT - this is not my business at all.
Both ways are fine. That’s o.k. for me.
Yes, that is absolutely right. One of my biggest teachers of management (CEO of big global company) said: “It is necessary to continue the delusion that has been started.” And yes this the way SME produces money. Afterwards I was working in sports betting and gaming industry with big shooters and people willing to take risk. But the same applied even in this businesses, especially for IT 
. People don’t need cars, but faster horses…
I have still like tens of virtualized OS Debian 2-4 with working solutions/applications, which I took over, and there is no way to modernize them. If it is working - don’t touch it. And there is already a problem to virtualize them in a modern proxmox environment. The same situation is in banking and automotive industry. They are hiring cobol and fortran developers to keep their old code…
I just really think that you should use the right tool for the right job.
On my personal internal infra (since it’s running on a RasPI 4) I don’t have nethserver on it, not because I couldn’t, but because I wanted to learn how to handle deployment and management of containers through ansible (with backups, proxying and all)
Some choose NS8 cause the to learn how to do that is too much, and they prefer something already baked with all the bells and whistles. I saw many customers use the multi-node approach in another way: start off as a local node, then join a remote one, to end up moving all applications with 3-4 clicks to the remote and dismiss the local node.
Nowadays if you deploy an application that have any degree of dependencies you use containers, that’s unless you are constrained by the OS you’re based on.
Are you trading off performance for ease of use? Yes, absolutely.
Can you run everything on a container? No.
Do you separate the container runtime from the OS allowing to use whatever you want on the container despite the OS? Yes.
I really think the third point is a way greater advantage than the other two? Yep, but your opinion might vary
2 Likes