Had a look at nodeweaver's website - https://www.nodeweaver.eu

It looks interesting. They have a clear edge computing, IoT, Industry 4.0 focus on their product and their customers. So it's not a generic IT HCI product but rather geared towards those types of applications.

So what's the difference really? Well, the difference is that it's probably going to be managed by technicians and not IT. So it's design is likely geared towards easy of use and stability rather than towards bleeding edge features and lots of settings.

And also possibly tuned to run on very modest hardware as that is common feature of edge devices. It's also very likely that it is tuned for deterministic real time applications, which is common on edge devices and IIoT. KVM has PREEMPT-RT which are patches to give better performance for real time applications. Basically you want to know that an application will respond within a certain time, every time, regardless of what it is doing.

So this could be a good thing or a bad thing. Most things like real time patches improves things where it is needed but doesn't cause much performance loss where it is not needed. Anyway, as a product I think if you want a turn-key system it will probably be a good fit. If you want to be able to tinker with everything it will probably not be the right choice.

Update to 19.10. Has been stable and good for us. Also 19.04 was trouble free.

I don't understand how the use of RDP could do anything to cause multiple logins?

If you RDP in to your desktop using the same login as usual then everything is exactly the same as if you're physically there.

@Dashrender said in Looking for solutions to allow remote users access to their internal psychical computers:

@Pete-S said in Looking for solutions to allow remote users access to their internal psychical computers:

@Dashrender said in Looking for solutions to allow remote users access to their internal psychical computers:

@Pete-S said in Looking for solutions to allow remote users access to their internal psychical computers:

Also are you using split tunneling on the VPN connection or is all traffic passing over VPN when connected?

I'm curious how this plays into the current conversation?

OP said he wanted to "make the best use of the remote session in terms of data transmission". It also plays into the security issue, together with credentials and logins.

aww - definitely understand the bandwidth portion, but not the creds/logins though.

Split tunneling is in general considered less secure because the user's computer is basically bridging the internet and your corporate network. And you have zero or little control over the traffic outside the VPN, unless you have some security in place for this.

When not using multi-factor authentication for the VPN tunnel (which the OP isn't) you are more susceptible to phishing attacks.

With split-tunneling and no 2FA on the VPN it's much easier to trick the users to enter their credentials into something that looks just like real thing.

That's how it ties into the security - overall risk.

Here are some covid-19 recommendations for VPNs.
https://www.us-cert.gov/ncas/alerts/aa20-073a

Microsoft have some new recommendations on how to do split tunneling VPN, particular with O365 and on-prem, to make it secure and to take load and bandwidth off the VPN connection.

Not related to security but it's also possible to have a bandwidth limitation in the VPN appliance without having a bandwidth limitation on the WAN link. There is usually a maximum VPN bandwidth in the firewall / VPN appliance.

@JaredBusch said in Virtual Desktops / Workspaces / Multiple Desktops:

@Pete-S I must have only skimmed that thread, because I use two workspaces consitently.

When on my laptop (single screen) for a long time, I will often be using 3.

Yeah, I count myself as being addicted to it. I use it also when running two or three monitors.

I always have 12 set up because keyboard shortcuts fits the the 12 function key. I very seldom need them all.

Looks like this on an old laptop I'm using right now (win7):

@scottalanmiller said in Virtual Desktops / Workspaces / Multiple Desktops:

@Dashrender said in Virtual Desktops / Workspaces / Multiple Desktops:

@scottalanmiller said in Virtual Desktops / Workspaces / Multiple Desktops:

@Pete-S said in Virtual Desktops / Workspaces / Multiple Desktops:

@travisdh1 said in Virtual Desktops / Workspaces / Multiple Desktops:

@Pete-S I remember that discussion. Blew me away that people didn't know what they were, I've been using them since the mid 90s!

Perhaps it's because of the "easy to use" trap - hide everything powerful so new users don't get confused.

On some distros you have to go looking for it to find it.

Yeah, it used to be very in your face in the 1990s. I think people coming over from Windows found it overwhelming as "too much power and choice" in Linux. It made the screen busier and most people aren't sure how to use it. So making it a hidden option made things seem more like Windows. For the average user, it's not useful, so I think that this tends to make sense.

I guess I don't see it being overly useful - all those things have to be running, right? I mean if you can lock an app to a specific screen to a specific set of dimensions, then maybe, so when you launch it in the future it's always in the same place, I suppose that could be helpful, but mostly it's a meh for me.

That's the idea.

Let me give an example...

You are busily working on a design project. Graphics and whatnot. you have all the different things that you need set up so that you can work efficiently.

Then you have to work on a server that is down. For that you need an RDP session to the server and some documentation open.

Then you have to do some meeting and have paperwork ready for that meeting.

With "workspaces" you can set up all three as if they were totally different environments and hop between them "instantly" with everything in place and ready for the task at hand. It's all but useless for people on a focused task. but super useful for people who have to switch between essentially unrelated tasks.

That's a great example.

Sometimes it useful also when you are working on the same task if it requires lots of windows.

The only time I feel multiple desktops are not as useful is when working directly on a small laptop where you are basically running every app in full screen. Then I have little need for multiple desktops because I only see one application at a time and I'm not really using the "desktop" at all.

@IRJ said in Does intra-VM traffic leave the host?:

2.) Why would you want separate virtualization platforms on the same subnet, anyway? Managing traffic by subnet is MUCH easier than trying to manage it by host. Flat networks are frowned upon these days because they are much more difficult to manage traffic.

I'm increasing the segmentation granularity so the traffic is already managed by subnet, however there are a few subnets I would like to lock down even more and it's just too much to make a subnet for every single host.

@JaredBusch said in iperf results wired vs wireless ...:

@Pete-S umm wut? I know exactly what I am reporting here. Did you mean to reply to the OP?

iPerf is the right tool for this job. It tests the line speed (wireless or wired) of connection.

It has nothing to to with day to day speeds, and should not.

Your test was perfect. No, it was meant for the OP or as a general comment.

@Grey said in Private DNS architecture?:

@Pete-S said in Private DNS architecture?:

@Grey said in Private DNS architecture?:

This all sounds very complicated. Why not use the DNS and DHCP at your datacenter and turn off all the others, and then give the routers an ip helper address config? Does your network hardware not support that?

@Grey It may very well be too complicated. At the same time it has to be fast, robust and the parts have to be able to work independently if a VPN link goes down.

Ok, cut the line to the internet. Can they still function? What doesn't work? What gets cached at your app server? How much data is transferred when the line returns?
How much actual resilience does the business need vs what they can sustain, and what's the risk? Has anyone answered these questions before?

The diagram is a simplified. It's only internal company traffic that goes over the VPN in the drawing. The data centers also serves other clients that are not connected over VPN. That actually their primary job - they are serving customers, not just internal workloads.

When it comes to resilience and risk, it's the data centers that have to be up and running. So they have redundant everything. The rest is just ordinary SMB stuff.

PS. Also in the data center we are doing HA in the application layer and not the hypervisor layer. So having two DNS servers made sense to me since that will be natural HA in the application layer.

Just plug them into a server that doesn't have a stupid perc controller and wipe them there.

@notverypunny said in SAS to USB:

@Pete-S said in SAS to USB:

Just plug them into a server that doesn't have a stupid perc controller and wipe them there.

We're almost exclusively Dell, so finding someplace to connect a SAS that isn't a PERC is going to be difficult, if not impossible.

I'd like to avoid having to use a server at all to blank them, might have to dig through some of the boxes of "junk" to see if there might be a SAS-compatible card that I could throw into a desktop

That's probably your best bet.

Dell also makes pure HBAs, like HBA330. Maybe it's even called PERC as well, I don't know. Anyway, if you put the disk in a SAS controller that isn't raid it doesn't think it's part of an array or something.

Check the registry according to this:
https://www.isumsoft.com/windows-10/how-to-disable-use-of-usb-storage-devices-in-windows-10.html

@biggen said in Add 2.5" U.2 (NVMe) SSDs to custom build?:

So I'm looking at the motherboard I'm pillaging from another system and it has two M.2 slots built into it! Doh! When I bought it last year I must have had a premonition I wanted a dual NVMe setup for a future lab!

Scott is right. Putting U2 drives in your build is like Ferrari suspension on a Prius.
And that MB can't take enterprise M2 drives so you're better of with consumer drives.

The IcyDoc you're looking at isn't really suitable for U.2 because you need server level cooling over the drives.

That said U.2 doesn't need a hot swap backplane because it's possible to cable them up similar to a SATA drive. Problem with that motherboard is that you need two available slots with x4 PCIe and that one doesn't have it. Or you can hook up two PCIe drives to one x8 or larger PCI slot - if you have bifurbication support on your PCIe slots and consumer motherboards usually don't have that.

@CCWTech said in Dymo vs. other print servers:

@Pete-S said in Dymo vs. other print servers:

Print servers is very old school. Normally you just use label printers that have ethernet.

Software vendor requirement unfortunately.

If they just support Dymo printers it has to be some kind of SOHO product.

Every POS printer, logistics printer, industrial label printer and such have models with an ethernet connection.

But if it is what it is then they you just have to buy the print servers. Not all print servers work the same.

A quick search shows that Dymo is very finicky with 3rd party print servers. And we know the AviMark software is finicky as well.

Is it really worth all this trouble to save what, 60 bucks? For something that might or might not work reliably?

Even if you need ten of them, I doubt you are going break even - if you can make it work at all.

And what happens when Dymo upgrades their driver and the DIY print server doesn't work anymore. You're back to square one.

Google's Android Emulator works on linux, windows and mac. It's part of the SDK for developing Android applications so it better work.

Also running Android x86 in Virtualbox works on both linux and windows. Virtualbox in general works with a much wider variety of guest OSs than KVM, Xen, Hyper-V, VMware etc.

Android X86 images ready to deploy on VirtualBox:
https://www.osboxes.org/android-x86/

@Dashrender said in Dymo vs. other print servers:

@Pete-S said in Dymo vs. other print servers:

@CCWTech said in Dymo vs. other print servers:

I have several clients that use the Dymo label printers. Many of them want to have multiple computers print to the label printer so I have used the Dymo LabelWriter Print Servers (est $125/each).

It seems really high compared to other brand of print servers out there.

Is there any particular advantage or reason to use their print server over just another vendors print server? I feel like money can be saved by using other vendors products instead of the Dymo brand, but wanted suggestions.

Print servers is very old school. Normally you just use label printers that have ethernet.

Do you have one that costs less than the combined price of Dymo label printer and the print server? (typically around $250)

We don't use Dymo but the Dymo LabelWriter Wireless printer seem to be in the $130-170 span.

@scottalanmiller said in Window server standard edition on Hyper V- means two Wins VMs ?:

@Pete-S said in Window server standard edition on Hyper V- means two Wins VMs ?:

Even back in the days I bet those "up to 20%" was a theoretical number when you had one 2.5" drive and no raid controller with cache. Which is basically never happens.

Well as essentially no RAID controllers accept SATA drives, it's decently common even there. But it would be interesting to see on MD or ZFS what kind of numbers are seen. But those stats were from VMware, so presumably tested in RAID.

That's a logical fallacy Scott. How could VMware do their test on RAID comparing SATA and SAS command queing performance when there are no RAID cards that accepts SATA?

Actually all raid cards I know support both SATA and SAS, for instance the Megaraid series from LSI/Broadcom that Dell, HP and everybody else rebrands.

It used to be that SATA drives were second rate citizens. It also used to be that SATA drives did a poor implementation of the command queing available with SATA aka NCQ because SATA was primarily a low cost drive for consumers.

But nowadays you can get the same high capacity enterprise drives with either SATA or SAS interface. And if you look at the stated performance on IOPS for random read/write operations you'll see that it's the same. On Exos 16 for instance it's 170/440.

On a file server it will be the file buffers in RAM on the guest OS/host and the RAID card cache that will have the largest impact. And SSD cache if it's available.

@Dashrender said in Video editing - suggestions:

oh man.. that guy is such a tool!

Not my cup of tea either but it doesn't matter. They shoot a lot of video and you can see them scrubbing and doing stuff on the 4K video stream and that is what you want. Looks like they are using Adobe Premiere Pro so should be similar performance with elements. Pro is $20 per month I think so not too bad if one would need more features down the line.

Performance on the hardware is decent enough but not awesome so it's not more than needed. Budget is $1000 which is what you had too. If you want to build or buy doesn't matter but this is the ballpark to aim for:

• CPU 6-core 3.6Ghz base, 4.2Ghz turbo.
• 16GB RAM
• 0.5TB NVMe SSD
• 3TB HDD
• Nvidia GTX 1650 4GB

@syko24 said in Dell PERC H740 with SSDs?:

So would SSD be overkill if using the H740? Or would I be better off saving some money and getting 10K RPM drives? This server is going to be hosting MS SQL Server with about 30 users.

Only go with HDDs when you don't have any other choice. SSDs will be much faster. A single SSD has IOPs in the several tens of thousands range while a HDD can only do a couple of hundred.

Read intensive SATA SSDs are getting close to the same price as 10K SAS. If you want to save money don't buy SAS SSDs, because you don't need them.

If you want to save even more money don't buy drives from Dell. Unless you're an enterprise customer, you pay 2-3 times as much for the same drive. For SSDs look for Samsung and Intel enterprise drives. They are the top two brands. Samsung PM883 or Intel S4510 in your case. There are other similar models. Expect to pay about $220 for 1 TB.