• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Question about resource usage

Draugauth

Limp Gawd
Joined
Mar 29, 2007
Messages
302
I am looking at building a Linux box for some VM usage.

What I want to do is use it to host 3-4 VM's.

2 NAS/SAN iSCSI VM's (fail over setup)
1-2 DHCP ADPDC (would like to do this in Linux)

I've already picked out the raid controller for the iSCSI setup and will be running 2 R6 sets. 1 for each NAS/SAN.

the DHCP/ADPDC servers will be stored on a raid 10 using onboard Sata 2 ports.

Primary OS will be stored on a R0 using onboard Sata 3 ports.

The iSCSI will be hosting the VM's for 2 other servers that will be running Win2012 with Hyper-V Roles which will host 2 VM's each Hyper-V and ESXi which will be duplicated as well as WHS2011 (or a newer version) on each.

The WHS2011 will be the file/print server for the home network and setup as a fail over. It will have passthrough access to the Raid controllers which will have 24 drives each in a R60.

The Hyper-V's and ESXi's will be used for lab purposes to learn CLI interfaces and how to control/setup fail over clusters as if in an IT environment.

I have the hardware chosen for the 2 Win2012 servers.

Right now I am trying to figure out what motherboard, cpu, and how much ram I will need for the Linux box. I'm leaning towards the same MB,CPU, Ram setup as the 2 Win2012 servers.

Win2012 server setups minus Raid controllers and quad port nics.

Norco 4224
Asus P8BWS
Intel E3-1230V2
32GB ram

I also currently have 2 Dell Powerconnect 5224 switches which I plan on adding in a 48 port gigabit switch to the bunch.

I figure the linux box for best performance will use 2 Quad port NIC's. each SAN/NAS will have 1 dedicated to it in bond/team mode. 1 onboard nic will be used for the DHCP/ADPDC server and the last onboard nic will be used for remote server management.

I know this is pretty expensive for a home lab setup but I want to have the stuff at home to practice on for when I get a job in IT again. That way I can screw up my home servers and not the work ones /lol.

And to anyone looking for an entry level IT worker in the Las Vegas area. Let me know please ;)

Haw
 
I take it that this is not the right forum or did I not provide enough info? I know it was late last night when I posted and I was tired so if something needs clearing up please let me know.
 
I need a good linux setup to run 2 VM's that will be handling failover iSCSI servers for the main systems.

Possibly 1-2 more VM's that will handle backup duties for DHCP/ADPDC or primary duties as such on a seperate vlan for home lab usage.
 
Iopoetve: Yes. 3 different servers though. 2 running Win 2012 standard with the Hyper-V Role and 1 running linux which will be hosting the shares holding the VM's for Hyper-V and ESXi as well as possibly the OS image for some thin clients I am going to build as HTPC's.

The question though is only in regards to the Linux server. I want to make sure I have enough CPU/RAM to handle multiple VM's running on it as well as enough bandwidth that I'm not waiting on transfers. ;)
 
Ah

Bandwidth is the last thing you need to worry about. IOPS are all that matter. RAM doesn't matter that much either unless you're doing dedupe or something.
 
You are overcomplicating what could be a very simple lab. Build a single box with your storage and a single box for your VM's. There is no need to 5-9's uptime for your lab NAS.
 
Redyouch: I'm creating a home based network and with that network I will be creating a sandbox home lab. So it will have it's own vlan and be separate from the home network incase I screw anything up.

Basically the idea is to have a business class network setup that can host a vlan for home lab stuff to test out before I implement it on the network. It will also allow me to have a learning tool without much risk to the items that I need to have stable.

Iopoetve: Won't the VM's need ram for them? I'm only thinking 8-16GB of ram for the linux box. Haven't decided yet on the CPU though. either 4 core Xeon or a 4 core W/HT Xeon (8 cores then). As for the iops it will be using SSD's for storing the VM's on so I won't have the spindle problems.
 
Still, you are overcomplicating this and will end up spending way too much money with no real goal in sight.
 
Redyouch: I'm creating a home based network and with that network I will be creating a sandbox home lab. So it will have it's own vlan and be separate from the home network incase I screw anything up.

Basically the idea is to have a business class network setup that can host a vlan for home lab stuff to test out before I implement it on the network. It will also allow me to have a learning tool without much risk to the items that I need to have stable.

Iopoetve: Won't the VM's need ram for them? I'm only thinking 8-16GB of ram for the linux box. Haven't decided yet on the CPU though. either 4 core Xeon or a 4 core W/HT Xeon (8 cores then). As for the iops it will be using SSD's for storing the VM's on so I won't have the spindle problems.

You don't run VMs on the linux box, you run it on a hypervisor - that's where we were all getting confused. Why are you using a linux machine? Your architecture makes ~no~ sense to me. :(
 
I can't actually find a question.

But I'll offer some:

You're trying to configure redundant iSCSI hosts and using R0 for something (not quite sure what)?

You want to run Hyper-V and ESXi nested inside an unspecified hypervisor running on Linux?

And two servers running 2012 with nested Hyper-V and ESXi hypervisors inside them?

I think you need to draw us some kind of diagram and tell us what your questions are.

Unless you're tripping over money or already have all this stuff laying around, it sounds like way overkill for someone looking for an entry level position. JMHO.
 
Iopoetve: VM's can be run under Linux using VM Player. Also if memory serves some of the Xen stuff is in the latest Linux Kernels as well.

Markwo: Question is about resource usage on a Linux whitebox that will be running anywhere from 4-8 VM's. (4 always and up to 4 more in a segregated vlan for home lab usage/testing)
 
Iopoetve: VM's can be run under Linux using VM Player. Also if memory serves some of the Xen stuff is in the latest Linux Kernels as well.

Markwo: Question is about resource usage on a Linux whitebox that will be running anywhere from 4-8 VM's. (4 always and up to 4 more in a segregated vlan for home lab usage/testing)

Yes, they can be, but a type-2 hypervisor is a horrendously inefficient method for running nested type-1s. I'd far prefer to run a type-1 and then nest more type-1s for testing, as it'll be a better use of resources. Player/workstation/etc are for online testing more than always-on servers.
 
Well I would love to build a dedicated ESXi box but that is a lot of money to get HCL components for a whitebox.

Otherwise I would look at building an ESXi whitebox and use it.
 
Well I would love to build a dedicated ESXi box but that is a lot of money to get HCL components for a whitebox.

Otherwise I would look at building an ESXi whitebox and use it.

Who said you had to buy HCL? You planning on calling support? It runs on a great many things that are not HCLed.
 
Who said you had to buy HCL? You planning on calling support? It runs on a great many things that are not HCLed.

Well it didn't work on my current server hardware. Well most of the hardware worked but not my raid controller.

After a long time thinking about it I've decided to just use WHS2011 in place of Linux. If I need to I can create the failover file servers in a sandbox until I can set things up things up the way I want them. It's just taking to long trying to get the information I've asked for.
 
Well it didn't work on my current server hardware. Well most of the hardware worked but not my raid controller.

After a long time thinking about it I've decided to just use WHS2011 in place of Linux. If I need to I can create the failover file servers in a sandbox until I can set things up things up the way I want them. It's just taking to long trying to get the information I've asked for.

You're asking to build a moon rocket, and starting with the Nina, Pinta, and Santa Maria.

You're going about things in a completely illogical and obtuse way instead of building a standard whitebox that will get the entire job done in one machine. You're buying hardware, but worried that an existing RAID controller won't work with a type-1 hypervisor? A PERC card costs somewhere in the $100 range on ebay and will do the job great. Or a P400 series. Or any other HCLed raid controller. Or, hell, a local sata controller - I run ESXi on a bunch of Dell D630 laptops right now for a dev cluster; it honestly works great on them.

What you're trying to do is completely nonsensical - really. That's why no one has answered; it's not a real question, so we're not sure where to start. It made sense 5 years ago when ESX had almost no compatibility with standard hardware and really required server-grade bits. Now the hardest part should be figuring out what $30 Intel NIC you want to use. There's no reason to embed T-1 hypervisors in a T-2 unless you're trying to turn a workstation into a brief "here's how it'd look" demo - performance is so lousy when passing through a hypervisor to another hypervisor on top of a hosting OS that other than showing it's possible, it's 100% totally useless. I should know, I used to do it. It's painful.

If you really want to do this, the answer is incredibly simple: Buy the biggest, baddest, hardware you possibly can, a metric-crapton of ram, as many cores as you can afford, and ask the Linux section to make sure it's compatible and how to tune it, as lord knows we don't pay that much attention here - all I care about is if Linux runs on a 440BX board with intel NICs and an LSI SCSI card, as that's what the world emulates in a hypervisor (almost the same for Xen/Hyper-V). That's the beauty of virtualizing - I don't care about actual hardware anymore as long as it runs a hypervisor.
 
Yes but the raid controller I've got is one that I've had for a while now and I couldn't afford another raid controller right now to replace it. I do plan on upgrading it at a later date at which point ESXi won't be a problem. But for now ESXi runs as a VM under Server 2012.

Now the only question I had was on resource usage on a linux box that would be holding the VM's for my 2 Server 2012 boxes. A question that no one was able to answer and instead wanted to try to say oh just build this instead.

as I've stated I have the 2 servers already specced out and most of the hardware bought for the 2012 servers. I only needed help with a linux server. Since the linux server would be running it's own VM's I thought that the Virtualization section would be the ideal place to go to.

I apologize for the confusion.
 
Yes but the raid controller I've got is one that I've had for a while now and I couldn't afford another raid controller right now to replace it. I do plan on upgrading it at a later date at which point ESXi won't be a problem. But for now ESXi runs as a VM under Server 2012.

Now the only question I had was on resource usage on a linux box that would be holding the VM's for my 2 Server 2012 boxes. A question that no one was able to answer and instead wanted to try to say oh just build this instead.
Because there is no answer. It's not a real question. How many squirrels does it take to Bob Yahtzee? It all depends on what you're trying to do with them - we can't answer this any more than we can answer the meaning of life. Resource usage depends entirely on what you're going to run in the VMs - in and of themselves, it takes almost nothing to start a VMM (couple dozen MB of ram, few CPU cycles), it's what happens IN the VMM that matters. An Exchange VM is going to look very different than a webserver. I ran what you're trying to run on an old laptop with an external drive. I've also run it on a brand new I7 with SSD. They ran very differently, but both ran - what do you want to DO with the VMs? What do your really want them to BE? Just a reference machine? Production quality? Usable quality?

Rule 1 of virtualization: There are no "best practices" for general sizing. They don't exist. Everything is unique to the hypervisor, workload, servers, etc.
as I've stated I have the 2 servers already specced out and most of the hardware bought for the 2012 servers. I only needed help with a linux server. Since the linux server would be running it's own VM's I thought that the Virtualization section would be the ideal place to go to.

I apologize for the confusion.

We barely pay attention to Type-2 hypervisors anymore - it's a now-niche market that is quietly starting to stagnate. Not that much left to do, and there's no advances - you just feed as many resources to the system as you can and go from there. :)
 
Iopoetve: If you look at the OP you will see that I stated what would be running in the first 6 lines (that includes the blank lines fyi) of the post. If those lines were not clear or did not give enough information then people could have asked me for clarification and I would have been more than happy to give it.

I'm done with this thread and disappointed.
 
Iopoetve: If you look at the OP you will see that I stated what would be running in the first 6 lines (that includes the blank lines fyi) of the post. If those lines were not clear or did not give enough information then people could have asked me for clarification and I would have been more than happy to give it.

I'm done with this thread and disappointed.

You said those were vms, and then mentioned RAID cards. Does not compute for a reliable, or decent performing, setup. We were trying to help you build a better mousetrap.
 
If you re-read your OP, you'll see that there's not actually a single question in it.

The closest you come to a question is "Right now I am trying to figure out what motherboard, cpu, and how much ram I will need for the Linux box."

Throwing up your hands and saying "I can't do this because some guys on a forum didn't read my mind and give me answers to questions I didn't ask in a timely manner" is both hilarious and asinine.

Your OP is a mess of scattered thoughts and statements. Regroup it into something we can digest and ask a few questions and maybe you'll get some better responses.
 
I know this is pretty expensive for a home lab setup but I want to have the stuff at home to practice on for when I get a job in IT again. That way I can screw up my home servers and not the work ones /lol.

Haw

You must be a very young person...
If the company I work for did not provide a Lab environment I'd simply look for a better place. I personally don't use my home lab fot any work related activities. To me it is just against conventional logic. Don't get too passionate about your employer. Loyalty rarely gets you anywhere in your career. From my personal experience anyway :)
 
Back
Top