small business server
Discussion
Hi
We use a windows based database that is around 20 years old - I know, I know . . . but I have yet to find anything that matches it for speed, simplicity and its usefulness within our business - in 20+ years, it has NEVER crashed.
It's a client server type and I would like to build a dedicated small server machine just for this . . . . that is always on
RAID 1
GB Ethernet
SSD HDD's - 250GB/500GB each is more than enough
Windows Server (can be as old as 98 but it works perfectly on Windows XP Pro too)
Must be BOMB PROOF
If you were going to build this little machine - how would you spec it, without breaking the bank?
Thanks
We use a windows based database that is around 20 years old - I know, I know . . . but I have yet to find anything that matches it for speed, simplicity and its usefulness within our business - in 20+ years, it has NEVER crashed.
It's a client server type and I would like to build a dedicated small server machine just for this . . . . that is always on
RAID 1
GB Ethernet
SSD HDD's - 250GB/500GB each is more than enough
Windows Server (can be as old as 98 but it works perfectly on Windows XP Pro too)
Must be BOMB PROOF
If you were going to build this little machine - how would you spec it, without breaking the bank?
Thanks
robbieduncan said:
But the OS will to a certain extent limit the hardware. Will Windows XP boot on a NVMe SSD on a modern Intel or AMD platform? I would have thought not.
It does, subject to driver availability.Amazingly, some 'newer' (not necessarily current!) drivers will actually work with the 64bit version of XP.
robbieduncan said:
OK. So we can have modern hardware. How much memory does it want? Will it run on a server OS (Windows 10, 8.1, XP are all client OS releases)? Do you have a rack or do you want a tower?
I would say minimal RAM but 16Gb would be more than enoughTower or small form factor ideally
khushy said:
<snip>
Must be BOMB PROOF
<snip>
Rules out SSDs with this kind of build, unless you can spend money on enterprise devices.Must be BOMB PROOF
<snip>
You would probably be better off going to Servers Direct and buying an off the shelf HP.
Said HP server will be quiet, performant, and reliable.
TonyRPH said:
Rules out SSDs with this kind of build, unless you can spend money on enterprise devices.
You would probably be better off going to Servers Direct and buying an off the shelf HP.
Said HP server will be quiet, performant, and reliable.
thanks for thatYou would probably be better off going to Servers Direct and buying an off the shelf HP.
Said HP server will be quiet, performant, and reliable.
What type of HDD's would you go for?
khushy said:
I would say minimal RAM but 16Gb would be more than enough
Tower or small form factor ideally
OK. I am struggling to find a reasonably priced case with redundant PSU capability. Perhaps a standard ATX tower case with something like this https://www.scan.co.uk/products/600plus600w-silver...Tower or small form factor ideally
Ideally I'd want something with ECC RAM for a server. That means Xeon or an AMD Ryzen/ThreadRipper/EPYC with a motherboard that supports ECC. In the interest of cost I'd go AMD as you don't need massive performance. Get a board that support 2xM.2 SSDs and use those as as your primary RAID. Stick a hot-swap drive bay or two for backups.
khushy said:
thanks for that
What type of HDD's would you go for?
I'll likely be shouted down for this, but probably Western Digital reds (the nas drives).What type of HDD's would you go for?
And depending on space requirements, I'd stay away from the very latest huge drives too (for reliability).
I have a home server running WD Blue x4 (not intended for this purpose) and I think they've been running 24x7 for about 5 or 6 years now, with 1 failure during that time.
TonyRPH said:
I'll likely be shouted down for this, but probably Western Digital reds (the nas drives).
And depending on space requirements, I'd stay away from the very latest huge drives too (for reliability).
I have a home server running WD Blue x4 (not intended for this purpose) and I think they've been running 24x7 for about 5 or 6 years now, with 1 failure during that time.
I've got 4x3Tb Reds in our home NAS. Been running 24x7 for about 6 years apart from a home move. No failures yet (touch wood)And depending on space requirements, I'd stay away from the very latest huge drives too (for reliability).
I have a home server running WD Blue x4 (not intended for this purpose) and I think they've been running 24x7 for about 5 or 6 years now, with 1 failure during that time.
anonymous said:
[redacted]
I would love a cloud solution and have tried a virtual server - but our broadband is s
tE - we have tried BT, Satellite and are now on an EE 4G card (which is not bad tbh) but for the database, which we use all the time for our business, its way too slow - hence the need for a physical solution.The DB is currently run on a PC we use in the office - WIN 8.1 pro etc - but I would like a dedicated machine for this
Hence the questions
anonymous said:
[redacted]
In my previous role in an MSP, moving things like local files and DB's to the cloud was just not possible due to most the UK's industrial estates not having a fast enough internet connection. A local server with these files was needed for ease of use and speed. Backups however were stored in the cloud. HantsRat said:
In my previous role in an MSP, moving things like local files and DB's to the cloud was just not possible due to most the UK's industrial estates not having a fast enough internet connection. A local server with these files was needed for ease of use and speed. Backups however were stored in the cloud.
we set automatic backups to the cloud every day - and to our local storage tooBB is rubbish here :-(
Otherwise a cloud solution might be for us
Though not a direct answer to the OP, here is A different approach (and hopefully will provide food for thought)
The way to make something bomb-proof is to minimise the risk / probability of failure, then reduce the impact of that failure. All hardware will have varying failure rates and as an end user, there are arguably only a few steps you can take to minimise the probability of a failure:
- Purchase hardware known to be more reliable (see comments of SSD's above)
- Purchase Hardware from vendors with a reputation for quality (hard to determine fact from marketing spin)
- Reduce the 'single points of failure' (RAID, Multiple PSU's, etc)
Each of the above usually come either at cost, or performance.
If the impact of the failure has been minimised, then you may not need so concerned about the probability of failure. As an example - I am currently working with a client on their DR environment. If any of their server HW fails, they can be back up and running within 10 minutes (in this case by using a cloud DR service, which I appreciate would not work for you). We have considered what the greatest impact on their data would be, and they are comfortable with it. As a result, the HW they use matters less, as their emphasis is on DR, and not trying to guess which component may fail next.
So to your original point - I would strike a balance. I'd buy a cheap server using volume hardware (so as to get a good price), backed up with a strong warranty. I'd build in resilience where the cost is not disproportionate to the risk it mitigates, and I'd make sure I have a watertight DR plan for when it does fail.
The way to make something bomb-proof is to minimise the risk / probability of failure, then reduce the impact of that failure. All hardware will have varying failure rates and as an end user, there are arguably only a few steps you can take to minimise the probability of a failure:
- Purchase hardware known to be more reliable (see comments of SSD's above)
- Purchase Hardware from vendors with a reputation for quality (hard to determine fact from marketing spin)
- Reduce the 'single points of failure' (RAID, Multiple PSU's, etc)
Each of the above usually come either at cost, or performance.
If the impact of the failure has been minimised, then you may not need so concerned about the probability of failure. As an example - I am currently working with a client on their DR environment. If any of their server HW fails, they can be back up and running within 10 minutes (in this case by using a cloud DR service, which I appreciate would not work for you). We have considered what the greatest impact on their data would be, and they are comfortable with it. As a result, the HW they use matters less, as their emphasis is on DR, and not trying to guess which component may fail next.
So to your original point - I would strike a balance. I'd buy a cheap server using volume hardware (so as to get a good price), backed up with a strong warranty. I'd build in resilience where the cost is not disproportionate to the risk it mitigates, and I'd make sure I have a watertight DR plan for when it does fail.
Gassing Station | Computers, Gadgets & Stuff | Top of Page | What's New | My Stuff



