I thought I had a lot of gear at home but WTF?!?
-
18 switches? damn?
140 devices? what the what? -
@NashBrydges said in I thought I had a lot of gear at home but WTF?!?:
Ran into this guy's home setup and wasn't sure if it should go in the "I can't even" or "water closet" section but anyway, have a read.
Not really any different than my house used to be. My house is pretty small, so 90 runs are not needed. But I do have one room alone with 12 individual runs of its own. Every bedroom has several runs.
The only difference is is that I had more servers and stuff, and he has more entertainment and IoT devices. If I was doing all that "smart home" stuff, I'd have a dozen additional runs and a big list of trivially small gear.
But unlike him, I have way more desktops, laptops, near-computers (dedicated compute devices), VoIP phones (one per room), APs and so forth. I used to have a rack (okay, it is still here), in a dedicated server room with dedicated HVAC, with a separate networking rack, with enterprise class UPS, almost ten servers, NAS, SAN, and more.
-
@Dashrender said in I thought I had a lot of gear at home but WTF?!?:
18 switches? damn?
140 devices? what the what?Yeah, but seems mostly consumer stuff. Just loads of gadgets.
-
If you think about a larger house for someone with disposable income....
You can end up with a big number of IP connected lightbulbs, cameras, light switches, power outlets, door bells, sound system components, televisions, etc.
My house has five televisions and I'd like two more. That's seven devices right there. Plus each room with a TV has a video game unit of some kind, so that's five more there, too. Silly consumer stuff adds up super fast. Plus everyone in the house has a wifi phone (that's five more) and a tablet (that's another five more.) So media devices alone gets me 20 devices in my small house.
The consider we all have laptops and a few desktops. The numbers just keep going up and up.
Oh and don't forget your things like Alexa or Google Home devices themselves.
-
The thing with this network is that it's focused on the gadgets, not the tech. From a tech side, it's pretty weak and the author doesn't really even know the stuff that he has...
I chose ProxMox as my virutalization HyperVisor, at least initially, because it's free and open source, and has most of the features as VM Ware, which I had originally intended to use (with a low cost VM User Group license). It has high availability cluster, live migration from node-to-node of VMs, and although it's not really a type 1 hypervisor, it sort of runs on the bare medal, with a basic install of Debina at the core, and the ability to run KVM containers as well as full VMs (for OSes like MacOS and Windows) on the same host.
So things that are wrong here...
- ProxMox isn't a hypervisor, it's a hypervisor manager: it manages KVM and OpenVZ
- You don't choose ProxMox over normal KVM or Xen because it is free and open source, that makes no sense and implies he doesn't even know what the normal options are to compare against so really just chose this because "marketing."
- He thinks this isn't a Type 1 hypervisor, when it is absolutely a type 1.
- He calls Debian "Debina"
- There is no such thing as KVM containers, that's just made up. It uses OpenVZ in the old versions and LXC in the newer.
-
And more...
FreeNAS - I had planned to use a Drobo B810i on iSCSI as my main Storage Area Network device for my ProxMox cluster, but it was not able to run an SSH server to allow the cluster to login, so I had to go another route. I chose to use FreeNAS on another Dell server. This is kind of the recommendation of the ProxMox folks, and it's running on Unix (in this case FreeBSD because it has the ZFS file system natively) and allows me to use the ZFS file system, which allows me to use all storage types on a network host that appears to the VMs as local storage. Having "remote" storage allows me to live migrate a VM - move it from one physical server to another while running, without the VM knowing that it is being moved.
He was consider Drobo and FreeNAS. What? So he has this whole clustering thing, but not only built his learning lab on a pyramid of doom, but considered two of the worst possible approaches to do it. And he even points out how poor the ProxMox community is, that they actually recommended this "by the book example of exactly the opposite of what you should do" approach.
And of course, stuff that is wrong...
- It's FreeBSD because it uses ZFS natively. UFS is the native filesystem in FreeBSD, ZFS is native only to Solaris. ZFS is no more native on FreeBSD than it is on Ubuntu. So both his "fact" is wrong, and his "since I have it, I must have this" is wrong. Yes, FreeNAS runs on FreeBSD, but his logic tells us he doesn't understand anything going on under the hood.
- He does the standard "ZFS does something horrible, and that's a unique feature" marketing trick. Yes, ZFS will allow you to screw up your storage. But so will everything else. He is implying that he needs ZFS because it's crap, but assumes nothing else is crap in the same way, when it's basic storage knowledge that everything works that way and you just don't do it and ZFS just has a bad community that is clueless and recommends stupid things.
-
It's a home lab and all just for fun, so whatever. But what it isn't, is impressive. It's a lot of "stuff", yes. But my take on it is that this is very much a bench tech who likes running cable, having racks, having gear and gadgets, and isn't really into the IT side beyond just the cool factor of "having lots of stuff." So his network design is based around "excuses to rack more gear" rather than "serious lab for learning how to run a production environment." Nothing wrong with that, it's just not very impressive from an IT perspective.
-
@scottalanmiller said in I thought I had a lot of gear at home but WTF?!?:
The thing with this network is that it's focused on the gadgets, not the tech. From a tech side, it's pretty weak and the author doesn't really even know the stuff that he has...
I chose ProxMox as my virutalization HyperVisor, at least initially, because it's free and open source, and has most of the features as VM Ware, which I had originally intended to use (with a low cost VM User Group license). It has high availability cluster, live migration from node-to-node of VMs, and although it's not really a type 1 hypervisor, it sort of runs on the bare medal, with a basic install of Debina at the core, and the ability to run KVM containers as well as full VMs (for OSes like MacOS and Windows) on the same host.
So things that are wrong here...
- ProxMox isn't a hypervisor, it's a hypervisor manager: it manages KVM and OpenVZ
- You don't choose ProxMox over normal KVM or Xen because it is free and open source, that makes no sense and implies he doesn't even know what the normal options are to compare against so really just chose this because "marketing."
- He thinks this isn't a Type 1 hypervisor, when it is absolutely a type 1.
- He calls Debian "Debina"
- There is no such thing as KVM containers, that's just made up. It uses OpenVZ in the old versions and LXC in the newer.
I was about to weigh in on this myself...
I used to be a grease monkey, and while mechanics can appreciate a $500k vehicle that has a million gadgets. That person is generally not a mechanic and they don't have the same troubleshooting logic as a mechanic.
Many of those people buy their vehicles. They either buy parts that are extravagant and well above what they need, or they don't do the labor themselves. Neither of those scenarios translate into real world experience in the field of auto mechanics because they are anything but real world.
At the end of the day, it's an extreme waste of money that translates into zero real world value. With vehicles, something at that level of customization is of so little resale value its ridiculous. Obviously IT equipment retail value is even worse. It's probably the worse investment of time and money you can make.
-
@IRJ said in I thought I had a lot of gear at home but WTF?!?:
@scottalanmiller said in I thought I had a lot of gear at home but WTF?!?:
The thing with this network is that it's focused on the gadgets, not the tech. From a tech side, it's pretty weak and the author doesn't really even know the stuff that he has...
I chose ProxMox as my virutalization HyperVisor, at least initially, because it's free and open source, and has most of the features as VM Ware, which I had originally intended to use (with a low cost VM User Group license). It has high availability cluster, live migration from node-to-node of VMs, and although it's not really a type 1 hypervisor, it sort of runs on the bare medal, with a basic install of Debina at the core, and the ability to run KVM containers as well as full VMs (for OSes like MacOS and Windows) on the same host.
So things that are wrong here...
- ProxMox isn't a hypervisor, it's a hypervisor manager: it manages KVM and OpenVZ
- You don't choose ProxMox over normal KVM or Xen because it is free and open source, that makes no sense and implies he doesn't even know what the normal options are to compare against so really just chose this because "marketing."
- He thinks this isn't a Type 1 hypervisor, when it is absolutely a type 1.
- He calls Debian "Debina"
- There is no such thing as KVM containers, that's just made up. It uses OpenVZ in the old versions and LXC in the newer.
I was about to weigh in on this myself...
I used to be a grease monkey, and while mechanics can appreciate a $500k vehicle that has a million gadgets. That person is generally not a mechanic and they don't have the same troubleshooting logic as a mechanic.
Many of those people buy their vehicles. They either buy parts that are extravagant and well above what they need, or they don't do the labor themselves. Neither of those scenarios translate into real world experience in the field of auto mechanics because they are anything but real world.
At the end of the day, it's an extreme waste of money that translates into zero real world value. With vehicles, something at that level of customization is of so little resale value its ridiculous. Obviously IT equipment retail value is even worse. It's probably the worse investment of time and money you can make.
Good comparison. This is a network designed to look cool to people who like "tech gadgets" with lots of lights, wires, fans, etc. It's the same thing that makes you get really excited about the special effects and lights in a video gaming rig. Nothing wrong with enjoying that stuff, but that's not serious system design, it's the "effects". It's like the difference between being excited and showing off how you tuned an engine vs. showing off that you added flame decals to a car.
This network is all flame decals, no substance. A serious lab is way more likely to be cheaper, less to look at, and way more solid. This one looks good in pictures, but doesn't hold up in an interview to explain it - even when taking the time to write a wiki article about it to show off, he wasn't able to articulate clearly what he has or why, because he built all this stuff, spent all this money, and never even took a few moments to research what he was doing with it or learning what it was once he had it.
-
@scottalanmiller said in I thought I had a lot of gear at home but WTF?!?:
@IRJ said in I thought I had a lot of gear at home but WTF?!?:
@scottalanmiller said in I thought I had a lot of gear at home but WTF?!?:
The thing with this network is that it's focused on the gadgets, not the tech. From a tech side, it's pretty weak and the author doesn't really even know the stuff that he has...
I chose ProxMox as my virutalization HyperVisor, at least initially, because it's free and open source, and has most of the features as VM Ware, which I had originally intended to use (with a low cost VM User Group license). It has high availability cluster, live migration from node-to-node of VMs, and although it's not really a type 1 hypervisor, it sort of runs on the bare medal, with a basic install of Debina at the core, and the ability to run KVM containers as well as full VMs (for OSes like MacOS and Windows) on the same host.
So things that are wrong here...
- ProxMox isn't a hypervisor, it's a hypervisor manager: it manages KVM and OpenVZ
- You don't choose ProxMox over normal KVM or Xen because it is free and open source, that makes no sense and implies he doesn't even know what the normal options are to compare against so really just chose this because "marketing."
- He thinks this isn't a Type 1 hypervisor, when it is absolutely a type 1.
- He calls Debian "Debina"
- There is no such thing as KVM containers, that's just made up. It uses OpenVZ in the old versions and LXC in the newer.
I was about to weigh in on this myself...
I used to be a grease monkey, and while mechanics can appreciate a $500k vehicle that has a million gadgets. That person is generally not a mechanic and they don't have the same troubleshooting logic as a mechanic.
Many of those people buy their vehicles. They either buy parts that are extravagant and well above what they need, or they don't do the labor themselves. Neither of those scenarios translate into real world experience in the field of auto mechanics because they are anything but real world.
At the end of the day, it's an extreme waste of money that translates into zero real world value. With vehicles, something at that level of customization is of so little resale value its ridiculous. Obviously IT equipment retail value is even worse. It's probably the worse investment of time and money you can make.
Good comparison. This is a network designed to look cool to people who like "tech gadgets" with lots of lights, wires, fans, etc. It's the same thing that makes you get really excited about the special effects and lights in a video gaming rig. Nothing wrong with enjoying that stuff, but that's not serious system design, it's the "effects". It's like the difference between being excited and showing off how you tuned an engine vs. showing off that you added flame decals to a car.
This network is all flame decals, no substance. A serious lab is way more likely to be cheaper, less to look at, and way more solid. This one looks good in pictures, but doesn't hold up in an interview to explain it - even when taking the time to write a wiki article about it to show off, he wasn't able to articulate clearly what he has or why, because he built all this stuff, spent all this money, and never even took a few moments to research what he was doing with it or learning what it was once he had it.
I still have automotive friends and we like to joke how a K and N sticker adds 30hp and how having a chrome differential cover on your 4x4 increases your offroad capabilities.
Actual mechanics see stuff like that as you don't know what the hell you are doing. Buying a $100 chrome cover for your axle is the biggest douche signal there is. Lol
-
Exactly, that stuff is so silly. I get the reasons that people do it, you have to do one thing to impress the serious people, and one to impress the clueless people who aren't interested in those things. Flame decals, CPU fan lLEDs, chrome axle covers, ProxMox are all "impress the masses" moves. If you do serious stuff underneath, then I can appreciate that stuff. But the lack of substance here is the real issue.
-
Now the networking gear is pretty nice. But it is nearly all just "plug it in and go", so while it is attractive and expensive and really good stuff, it doesn't really add much lab potential.
-
He's running his own email server but uses an Ooma for his phone... :man_facepalming:
-
@bnrstnr said in I thought I had a lot of gear at home but WTF?!?:
He's running his own email server but uses an Ooma for his phone... :man_facepalming:
My guess, from looking at the network, is that phones are beyond him, or at least quite challenging. Getting phones working is harder than email and less obvious.
These things aren't "levels", but his knowledge level of storage and virtualization suggests that VoIP would be rather a struggle. So not surprising.
-
I agree, mostly flash, little substance. At least imbalanced substance. Some good stuff that is overkill in one direction, then missing the basics in others. But hey, it is a lab, hopefully it is actually teaching him stuff and not just costing a lot of money.
-
@scottalanmiller said in I thought I had a lot of gear at home but WTF?!?:
Exactly, that stuff is so silly. I get the reasons that people do it, you have to do one thing to impress the serious people, and one to impress the clueless people who aren't interested in those things. Flame decals, CPU fan lLEDs, chrome axle covers, ProxMox are all "impress the masses" moves. If you do serious stuff underneath, then I can appreciate that stuff. But the lack of substance here is the real issue.
Yeah true, but I think the chrome axle cover bothers me more than anything else haha. So... You are putting a chrome cover on a 4x4 that is generally lifted with large tires, but people are supposed to believe you go offroad. If you go offroad, what are the chances your axle gets dirty? nearly 100%, right? lol
Other issues like K & N systems are generally stem from a lack of knowledge similar to what we commonly see in IT. Are K & N systems generally good and do they help the vehicle? sure, but it is it worth the $400-600 to get a 4-8hp? People get confused by the marketing because they see the system can add up to 60HP. The thing is that you will only get that much HP in a very extreme circumstance. Every motor is different and most are obviously fairly well tuned at this stage. You are not going to be a legitmae automotive company and designed a terribly restricted air system that decrease HP and fuel economy at a tremendous level. It would hurt you big time. Where we see most of these gains is in older EFIs from 80s and 90s when these things werent tuned. Even then you would be super lucky to get 20-30HP. I have never heard of a single instance of anyone getting more than that.
We could compare this to many things in IT where the marketing doesnt necessarily lie, but it is purposely misleading and they make their money from misleading customers.
-
@IRJ said in I thought I had a lot of gear at home but WTF?!?:
We could compare this to many things in IT where the marketing doesnt necessarily lie, but it is purposely misleading and they make their money from misleading customers.
Or from giving the customers the tools to mislead themselves. That's very much where ProxMox and FreeNAS fall. Rarely do they lie, but they sure are misleading and make it really easy for customers to convince themselves of things.
-
I think I would spend the money on having my servers in a colo and/or use Vultr.
-
@black3dynamite said in I thought I had a lot of gear at home but WTF?!?:
I think I would spend the money on having my servers in a colo and/or use Vultr.
Right, seems like way better ways to use that same money to learn more. Nothing wrong with servers at home and if he is a bench buy, which sure seems likely, then his "lab" is the physical pieces, not the logical ones. So it would make sense.
If we view this as an IT lab, it's a joke. If we view this as a bench lab, it's great. So perspective matters a lot.
-
I just wonder at the money he spent. I don't have the same interests as he does since I care little for all that automation stuff. I do want to know how much he spent though lol
