Hardware Thread, New server system for primary school in Technical; Hi there, new here, and hoping to get some input from people more knowledgeable than me!
A primary school I ...
29th July 2013, 05:10 AM #1
- Rep Power
New server system for primary school
Hi there, new here, and hoping to get some input from people more knowledgeable than me!
A primary school I work at is looking to update their aging server system, and tasked me in some preliminary investigation into what we could do with both hardware and software, and get some idea on cost. Ive done a little investigating so far so I will detail what Ive got.
Currently we have separate curriculum and admin networks. 3 server machines, 1 for admin domain, 2 for curriculum for redundancy. these are all Acer/ Altos G330 Mk2's running Windows server 2003.
There are maybe 100 systems connecting to this, a mix of laptops and desktops with anything from XP through to windows 8. There are also 30 iPads.
The school is also looking at getting maybe 50 new laptops, or a mix of laptops/desktops and some more iPads.
Now the school is not very big, and has a limited ICT budget of course, but will be flexible if I come up with a plan that they are sold on.
I don't see us buying 3 brand new pieces of hardware to replace the 3 aging server systems, and think it would be best to combine these into a single network. There are only around 4 machines that actually use the admin network.
It has been suggested that I could buy a single beefy machine with plenty of upgrade potential. Install server 2012. Copy over and run both current 2003 domains in VM's temporarily. Mash the ram and HDD space etc of the 3 old systems and use as redundancy backup. This would allow me to setup and test the new system on the side with minimal pain, and cut over without changing to much of the existing system (Its only me there at the moment, and for a few hours a week only). This would have the system running faster and more stable with updated hardware to begin with. then as the school can afford it, buy the CALs necessary to eventually cut over to a completely server 2012 system.
Does this sound reasonable? The only issue I can see, is the hardware needed to support this being quite expensive.
I have been quoted so far a HP ProLiant DL385 G8P, 3x1TB HDDs. Server 2012 with 5 CALS for around 9k AUD.
I guess I am looking to find out, does this upgrade path sound sensible? Is the system quoted overkill? Is it likely I could do any reasonable upgrade for around 5k AUD (I know most of you are from UK so I think thats closer to 3k GBP)? This is simply so I have a basis for selling this idea to the school. Lastly, is there any hardware you guys would recommend that I could checkout and seek quotes for?
Thanks a lot for anyone who reads this! Any help would be great
IDG Tech News
29th July 2013, 07:27 AM #2
29th July 2013, 11:14 AM #3
You're in Victoria (as am I) so I'll tell you what I would do in your situation. You really don't have enough money with only 5K to do virtualisation properly. Here's why. Lets say you buy the best box you can for 5K and then you install Win SVR 2012 and migrate the three existing servers into HyperV (your most cost effective VM solution for your circumstances), and for 5k you can buy a decent server.
Originally Posted by iMangles
Now, everything is running great, you're backing it all up using one of several solutions (several can be had for under $100 per host). Then your host server goes down, which could happen for any number of reasons. Now you have all three servers down, and none of your existing servers are probably capable of acting as a HyperV host (if they are correct me here). Now, you're really down and either have to wait until its fixed (could be days) or have to quickly find (buy) another server somewhere while the whole network is down in the meanwhile. I'm not going to endorse that route for obvious reasons.
You can buy a decent server for 5k that could act as a HyperV host. The problem is that you really need two of them so you have redundancy and load balancing. Or, you could buy three decent new servers and forgo the VM route for around 7k - they don't need to be as highly spec'd because there are three of them. You are stating that you will have close to 200 clients on the near future, so if you go cheaper than that you'll be up the creek without a paddle I reckon. We also haven't talked about the server that will act as a backup server. Now, this could be a decently spec'd desktop PC with sufficient disk space and Gigabit NIC if you already have one or you might be able to repurpose one of the existing servers. You'll need a backup server and software no matter what solution you go (virtualised or physical).
Virtualisation is not always the best solution especially in smaller environments, I think that your environment is big enough to warrant virtualisation, but you really need at least 10k to implement it successfully and safely and it sounds like the school is scrimping in its ICT spend. If you really are that cash strapped, I would contact the guys at Scorptec to see what they can build you in your budget, you'll likely get more bang for your buck than going through HP, Dell, etc. You might also look at some decent refurb or ex-demo gear (Grays online, eBay). If I were you, I would go back with a plan that covered:
- Two servers big enough for HyperV hosts that can run all of your servers (overspec them by at least 30% to account for increased usage with additional clients), that means a minimum of 4-core Xeon or AMD, 16-24GB RAM, RAID5/6 storage (7200RPM)
- Software licenses for the servers
- HyperV backup software
- Backup server and storage (might be able to repurpose existing equipment, if not budget $1,000 for this at least.
I reckon you need at least $11,000. 5K isn't going to cut it unless you buy refurb or you'll end up with a subpar system that has no chance of coping with any growth or node failures.
I'm here in the Southeast suburbs of Melbourne so if you have questions just let me know.
29th July 2013, 11:25 AM #4
Take a look at some of these barebones server units and load them up with the drives and CPUs that you need. I think it will be best value for money. The guys at Scorptec know their stuff and are very responsive. They can help you build a system to meet your requirements to a budget.
29th July 2013, 12:14 PM #5
I've been thinking about this.... I'm not sure you do need two, if your business continutity planning is robust, and SLT accecpt that an element of risk is required to avoid paying for redundacy of provision.
Consider that with server based hardware you can get 4hr response times on hardware warranties.
Anecdote: I've never had a server actually fail. Bits of one, yes, but never a whole box. You do need proper Windows Server certified hardware, with redundant disks (RAID 5/6) and a UPS to minimise the risk. And something like this: Veeam Backup & Replication Cloud Edition for VMware and Hyper-V would negate the need for backup tapes and manual rotation.
I'd also be tempted to suggest that you don't P2V migrate the existing servers. If you can wait for a few months until 2012 R2 is out, build a new clean environment and migrate the services into that - this will give the primary the best value from the work. It also is supposedly easier to integrate with the Microsoft Cloud, which is certain a direction that a primary should be looking.
I am assuming that you've got a "reasonable to good" internet link.
Last edited by psydii; 29th July 2013 at 12:16 PM.
29th July 2013, 12:22 PM #6
If money is really tight, see what you can get on refurb. I know from a performance point of view a HP Proliant G5 380 has enough grunt to serve your needs for less than £1000 (GBP) and at that sort of price you could afford two. You would need to be quite fussy about the CPU though as Hyper-V has certain requirements that older machines may not meet.
29th July 2013, 12:49 PM #7
- Rep Power
I act as the Network Manager at a medium size Primary School (400+ children). We haven't bothered with redundant servers. We are running the latest version of 2008 having upgraded from 2003. The old 2003 server was powered up and connected for a while but only offered load-sharing - which proved unnecessary from a performance perspective - and not true standby capability so I turned it off and have successfully been crossing my fingers ever since.
Since we are not running air traffic control at Heathrow or the signalling at Clapham Junction and we are seriously strapped for cash, redundancy in the server side of the network is a luxury we are doing without.
29th July 2013, 12:57 PM #8
I have had servers fail, and top quality ones at that. It does happen. Not everything on a server is redundant and nothing is bullet proof. The 4hr response time does little good if the part you need isn't in stock in the country. I had a SAN logic board pack it in earlier this year and they had to fly the only logic board available for it in country from Perth (3,400 km). Te best RAID or UPS won't help you there, and even RAID can fail you (I've seen it).
Originally Posted by psydii
I do agree with the P2V, I would be reluctant to do that and would suggest a fresh build. I wouldn't do it on 2012 though, and I wouldn't do it on 2012 R2 until its been out for a couple of months. Personally I would go with the tried and true 2008 R2, but there are some advantages with 2012 for you in a small environment with HyperV.
29th July 2013, 01:00 PM #9
Crossing fingers is not a good strategy. It's all good until (not if) it goes down, then you'll have to pay the piper. Speaking from 20 years experience here.
Originally Posted by catch21
29th July 2013, 01:05 PM #10
Sadly most senior execs in primary and secondary education, when presented with the up front cash cost of hedging against a hardware failure are are happy to take the risk that services will be down for a day or three during school hours in a given three year window. That's what contingency/reserves is for.
Originally Posted by seawolf
You can only do what you can do.
Last edited by psydii; 29th July 2013 at 01:07 PM.
29th July 2013, 01:29 PM #11
Until it happens right during report writing time or they are unable to communicate with parents during an emergency and aren't prepared with a paper based backup to deal with such contingencies. No, they'll just blame you.
Originally Posted by psydii
Our job is to convince the execs why taking that position is a bad one. Here's an idea. Imagine you pulled the plug for a day or three? Just what wouldn't the school be able to do, what would they still be able to do? Write a risk assessment showing the impacts that would actually occur, include worst-case scenarios. Present that to the execs and watch them squirm as reality hits them of what the impacts are. If you determine the impacts are low, fine. For us, they would be serious and crippling. The more you use technology at a school the more dependent you are on it.
29th July 2013, 01:38 PM #12
I have recently bought a raft of servers, refurbed, the beefiest coming in at £1000 (DL385G2 - two quad core Opterons, 32GB RAM, 8x72GB 10k SAS). Obviously the supply in Oz will differ and prices on older kit may well be higher. Adding your storage requirements to this wouldn't add much cost for SATA drives. This older kit should be more than up to a primary virtualisation setting.
For redundancy, I have a few of these hosts around - I can just transfer the VM backups across to another host quickly and easily. The peace of mind this gives is worth it - you could still do this in a tight budget with refurb kit.
29th July 2013, 01:40 PM #13
Agreed. Good quality refurb gear is a viable option. Gear is a bit more expensive here in OZ though.
Originally Posted by 3s-gtech
29th July 2013, 03:04 PM #14
Get refurbished gear and find out if they can supply share SAS as that will give you the shared storage for HA.
29th July 2013, 03:12 PM #15
They might try, but since they will have signed off on the proposal, which highlighted the risks of not having redundancy, and the costs of implementing it, their arguement would fail.
Originally Posted by seawolf
However as the 'architect' of the technical solution, it *is* our responsibility to ensure that processes are resilient in the event of failure... so to not work with the executive to have paper based fall back systems ready to go would be to some extent our fault.
By TonyMiles in forum Hardware
Last Post: 24th June 2013, 02:35 PM
By CHR1S in forum Hardware
Last Post: 17th June 2013, 04:27 PM
By _Bat_ in forum Downloads
Last Post: 1st October 2008, 03:01 PM
Last Post: 20th September 2006, 08:18 PM
Users Browsing this Thread
There are currently 1 users browsing this thread. (0 members and 1 guests)