P2V - single server with ALL the roles...?

[Koldov]

It's an old DELL PowerEdge T420 single server in a very small rural Junior School, there's no money to throw at it...

 

Obviously this is going to be a nightmare, but I really want to get this server up to at least 2019 host and split it into 2 VMs.

 

Current server is on 2012R2 - SQL/SIMS/DNS/DC/FMSO, holds all the data/file shares... well just everything really!

 

It uses WSB to a USB (not tested - I know... file restores OK though).

 

Currently it runs:

 

1 x Intel® Xeon® CPU E5-2440 v2 @ 1.90GHz

 

32GB RAM

 

PERC H310

 

2 x 300GB (15K SAS 6G AL13SXL300N) set in RAID 1 for C:\ OS

4 x 600GB (15K SAS 6G AL13SXL600N) set in RAID 10 split into 2, roughly 500GB each for D:\ SIMS and E:\ DATA

 

Is it easier to P2V (but isn't that just going to make a VHDX of the whole server - which might be easier, but not what I really want?) or just create the 2 x HYPER-V servers and migrate SIMS to one (which I don't actually know how to do) and the DC to the other, then upgrade the host/demote etc...?

 

I would upgrade the host first, but worried an in-place might do some damage to SIMS or the DC and it might be better for them to be on their own VMs first...

 

In case you haven't guessed, I've never done anything like this and the worry of messing it up is just going to mean otherwise I'm going to leave it as is... forever...

[mavhc]

Literally 0 money? Or a small amount?

 

What I did was buy some Intel SSDs, SATA, wait until half term, make a new software raid, install a new bare metal os, and then from within that maybe it's possible to read in the old HDDs and convert it to a VM. Does your server have unused SATA ports?

 

Once you have that you can then do things a) remotely and b) without downtime.

 

Test install new VMs, move things across slowly. Or take a snapshot and test in place upgrade the old server VM.

[Koldov]

Literally £0 budget... I'm not in charge of it and have no say in what it is allocated to, anyway I've asked and most of it (if not all of £10k) has already been set aside for a new class set of iPads and various other shiny things important to T&L...

 

Although I will put in a request to the Business Manager saying 2012R2 is EOL in October 2023 and so the OS needs to be upgraded (although might have to wait until next year now to have some allocated), plus all the benefits of VMs etc...

[mavhc]

There's always spare money, point out that the whole school relies on a system that's not had any investment for 10 years, they'll need to pay for new versions of Server anyway, bit more for some SSDs would a) make it faster, and b) make it more reliable

[jmak]

I did a P2V of a "one server does everything" using Microsoft Disk2VHD. I did it over a half term weekend and saved the VHD to a USB HDD. I then booted the VHD from hyper-v on my PC which was sloooow, but enough to prove that everything functioned.

 

I then took a brave pill and wiped the server and rebuilt as a hypervisor and added the VHD. Next stage was to create more VMs and move roles across. I never got rid of the original VHD altogether, just left it with things I was worried would stop working and moved other roles off. After the first virtualization it was straightforward to take snapshots before I made any changes, so it was always easy to revert if something didn't work out.

Edited June 23, 2022 by jmak

[chaplic]

Best practise is out-of-the-window here so I don’t quite get the benefits of splitting the roles in two.

 

I would be quite tempted to P2V the server and have the bare metal runnning nothing but the hypervisor, then dump the VHDs to another machine regularly so it’s easy to stand up the old server on a spare PC when the poweredge fails (spare pc probably faster too).

[jmak]

My main aim at the time was to give me a better chance of a quick recovery in the event of hardware failure. As you say, hook up a recent VHD (I had daily snapshots) to a decent PC and you can get the school running in minutes rather than days. At the time I knew I didn't have the knowledge to quickly build a server and get an active directory domain going from scratch, but I could install a hypervisor and attach a VM.

There were some benefits to splitting out roles; I could install software and updates with restarts without disrupting what was happening on other servers. It also felt less risky than an in place upgrade to a new version of server for domain controllers: create new VM, add to domain, DC promo then demote the original. There are probably better ways to do it, but it worked for me.

[sigma]

It's an old DELL PowerEdge T420 single server in a very small rural Junior School, there's no money to throw at it...

 

Obviously this is going to be a nightmare, but I really want to get this server up to at least 2019 host and split it into 2 VMs.

 

Current server is on 2012R2 - SQL/SIMS/DNS/DC/FMSO, holds all the data/file shares... well just everything really!

 

It uses WSB to a USB (not tested - I know... file restores OK though).

 

Currently it runs:

 

1 x Intel® Xeon® CPU E5-2440 v2 @ 1.90GHz

 

32GB RAM

 

PERC H310

 

2 x 300GB (15K SAS 6G AL13SXL300N) set in RAID 1 for C:\ OS

4 x 600GB (15K SAS 6G AL13SXL600N) set in RAID 10 split into 2, roughly 500GB each for D:\ SIMS and E:\ DATA

 

Is it easier to P2V (but isn't that just going to make a VHDX of the whole server - which might be easier, but not what I really want?) or just create the 2 x HYPER-V servers and migrate SIMS to one (which I don't actually know how to do) and the DC to the other, then upgrade the host/demote etc...?

 

I would upgrade the host first, but worried an in-place might do some damage to SIMS or the DC and it might be better for them to be on their own VMs first...

 

In case you haven't guessed, I've never done anything like this and the worry of messing it up is just going to mean otherwise I'm going to leave it as is... forever...

 

I think I look after a school that was probably set up by the same company (Wildings/Wildwater?) as I have the same config also for a very small rual Primary.... I'm looking to replace the server/do something.

 

They are a maintained school and so I'm feeling it hinges on when County go to Cloud based SIMS as it would seem to be a good time to consider going serverless. I don't want to spend any money on a server migration if I don't have to, so I've been sitting it out as is. (Procrastinating).

[mavhc]

Waiting for someone to do something with sims is a bad idea, civilisations will rise and fall in that time, or at least Windows Server versions

[Koldov]

My main aim at the time was to give me a better chance of a quick recovery in the event of hardware failure. As you say, hook up a recent VHD (I had daily snapshots) to a decent PC and you can get the school running in minutes rather than days. At the time I knew I didn't have the knowledge to quickly build a server and get an active directory domain going from scratch, but I could install a hypervisor and attach a VM.

There were some benefits to splitting out roles; I could install software and updates with restarts without disrupting what was happening on other servers. It also felt less risky than an in place upgrade to a new version of server for domain controllers: create new VM, add to domain, DC promo then demote the original. There are probably better ways to do it, but it worked for me.

 

This...

 

This set-up has been rock solid for years, I know her inside and out, all the quirks, there is hardly a day I have to do more than a quick look over the Server Manager to see if anything is being weird (which is very, very rare - unless it's SIMS or an MS update)... it works OK and does everything it needs to do (which isn't much in a small school).

 

Yes, I've had to change a couple of drives, but in all honesty I do appreciate not having to come in and fight fires or struggle with issues related to the server at least... I have enough of that with pupils/teachers...

 

I think one my main issues is that I'm not really a 'poke the bear' type and try not to make work for myself... I like (need) low stress, no drama...

 

But having said that I'm really not looking forward to the day when something goes 'pop' or we get some sort of Malware/Virus attack (plus the server OS only has a year to go) and so as others say, I know I need to have a better Back-Up, DR plan and splitting roles and Hyper-V should help me with that in the end.

[Koldov]

Trying to start this journey and falling at the first hurdle...

 

I teamed both the NICs in the server a few years ago, so which do I choose....?

 

I'm presuming the team, but I don't want to mess anything up!

 

[mavhc]

I'd simplify things and unteam them first

[Koldov]

Thanks, it did cross my mind (can't remember why I did it in the first place now - possibly just because I could - maybe resilience/load balancing?) I don't think it gains anything from it speed wise as our old Cisco switches are only 1GBps at the core anyway...

 

So NIC1 for the HOST and NIC2 for the VMs...?

Edited June 28, 2022 by Koldov

[paulkerton]

Keep it backed up. When it does start to go, they will invest.

It's exactly what I had to do at an old role. No money, until the HDDs started to fail and magically they took a move to the cloud for files for students, and found the cash to replace kit for the staff side of things.

[Koldov]

Yeah it's one of the reasons I'm trying to get it split up into Hyper-V machines if anything ever needs to be moved onto a new machine quickly. Also for the many other benefits including checkpoints and restores... at the moment it's WSB to a USB, I can only imagine how painful that would be in a DR situation!

 

Ah, I remember now, the NIC Team was due to WSUS hogging the bandwidth once a month as the updates were released... thinking of going with the 'Do not store update files locally; computers install from Microsoft Update' anyway, as there are only maybe 20 laptops, 1 ICT Suite and 10 admin PC's with an OK internet connection.

[mavhc]

Just trash WSUS and move to WUfB, unless you have an isdn, or 4000 clients, all laptops

[Koldov]

Does that require any InTune/Azure funny business... we don't use either and I'm not really looking to take on any more learning curves just yet... it would appear I can't even do the basics right!

 

Made some space on the server last night to accommodate the VMs, shrunk one volume and without thinking decided I would tidy up and give the old volume a new Drive letter. Ignored the 'programs won't work if they rely on the drive letter' warning, as nothing uses it by drive letter anyway (or so I thought)....

 

Came in this morning to mass panic and zero shares available on the network!

 

So, that's been fun...

[houlta]

Its group policy set-

https://docs.microsoft.com/en-us/windows/deployment/update/waas-manage-updates-wufb

 

Just looking at your original post though, if budget is that tight then I would use Urbackup to make an image backup of the server.

Its free, makes a vhdx and can also do file level, incremental etc. https://www.urbackup.org/

The intent is of course to have more than 1 machine to run it with

It will also backup databases separately if you like and vm's with the client installed.

 

Until they pay then it's not your job to put too much effort in really, I would put everything in writing though and copy in several people with clear text stating the risk.

This is basically how these small single school setups fail, sod all investment in IT and an impression that it runs forever with no need to spend anything.

Next thing they know, their desktops are locked up with a request for bitcoin and their entire fileserver is just gone, along with SIMS and anything else they bundled on it then they panic and blame everyone else. They left the backup connected with a single USB drive and ofc the hackers encrypted that too.

 

As a minimum really you need a vmhost and a backup server on site- a synology NAS is usually most cost effective, then a cloud backup (veeam licensed through the cloud backup company not veeam direct).

Ideally you'd have some old servers you would re-purpose as you purchase new kit and use them for backup redundancy.

[Koldov]

Yeah, I agree in principal, but in practise I have to do what I can.

 

I've seen the GPOs for WUfB, but how do you manage it all (see who has installed what update, approve/remove etc.)?

 

I've seen 'Update Compliance' but:

 

Update Compliance is a Windows service hosted in Azure

[mavhc]

It's free, and easy enough to set up http://www.edugeek.net/forums/windows-10/207846-update-compliance.html

[houlta]

Its not that detailed. The idea is you apply the policy to the users/computers you want and just disable it when you need to and amend it if need be.

Theres a tool for setting it all here-

https://www.microsoft.com/en-us/download/details.aspx?id=101056

 

You won't see who has managed to install what, you'd need something separate for that. Its more like automated updates with a few caveats is all.

Do you have O365 there?

Edited June 29, 2022 by houlta

[PotNoodleTech]

Literally £0 budget... I'm not in charge of it and have no say in what it is allocated to, anyway I've asked and most of it (if not all of £10k) has already been set aside for a new class set of iPads and various other shiny things important to T&L...

 

Then leave it to die.

 

If you cobble something together with weeks of your time and ingenuity and a pinch of good luck - they'll literally never ever see the need to spend any money on servers.

 

Sad as it is, schools like this living 20 years in the past when IT systems didn't matter are only gonna learn their lesson by downtime.

 

Sounds harsh but I've seen it happen - (some don't even learn after a week or two's worth of nothingness)!

[mavhc]

There's always money, they must keep some for emergencies after all.