Advice on replacing/upgrading Poweredge T320 & T310

Sorry... you answered my query about more than one SSD here:

"...You could probably get away with a couple of Intel server SSD's, like the S4510 - 480G..."

Now post no.5...

I've had a look for details of the RAID. In Device Manager, is states 'DELL PERC RAID Virtual Device'.

Does that shed any light on it for you, or at least confirm that it's Dell?

Edit to add: I see 2x DELL PERC H700 SCSI Disk Device stated under Disk Drives.

Questions welcome :)

2016/2019 ... no real difference. I'm assuming you can probably still buy CAL's for 2016, but you might as well buy CAL's for 2019, as they are backward-compatible with previous versions.

Domain Controller - Active Directory Domain Services is a role included with every Server OS. You install/configure it to manage a domain on your network - central management of users/computers, etc.

Hyper-threading - it can be disabled in the BIOS but on by default (and should be left on). It is simply a processor feature available out of the box that does not need to be configured or managed.

Which SSD's? or which controller? Controller will be shown in Device Manager, under Storage Controllers, and possibly hinted at under Disk Drives (PERC H700 Disk Device or something similar). The type/model of drive will not be shown in Disk Management (unless it is a non-RAID controller). You can get more detail in OpenManage Server Administrator (Dell hardware management program installed in Windows). Looks like you have a Dell PERC H700 RAID controller. (PERC is a Dell name: PowerEdge Raid Controller)

Dell RAID controllers have dozens and dozens of settings for managing RAID arrays, disks, and communication. Certified drives are drives with Dell firmware on them, tweaked to the exact set of settings the Dell RAID controllers expect from drives. Non-certified drives' settings may not match what the controller expects, so they may not be able to even negotiate a very basic scheme of communication and the drive may not show up at all on the controller, or they may not be configurable ... OR they may actually work or seem to work fine, if the settings are close enough to work with the controller. That's why putting desktop/laptop drives on a RAID controller is a bad idea - and they often won't work at all - because those drives lack the basic programming needed to obey, work, or "play nicely" with the controller. Worst-case scenario: the drives are recognized, are configurable in a RAID array, but then fall offline randomly and/or corrupt data. (Note: all OEM's do this - HP, Lenovo, etc.)

"You also suggested that we consider certified SSDs for mission critical workload. What would be an example of mission critical workload?"

These days, pretty much everything. I haven't order a server with spinning drives for over a year, even for large file servers (large-ish - 4TB of storage, all SSD-based). I probably would never again put a database on anything but SSD, be it QB, SQL, etc. You mentioned it is also a terminal server, that it acts as the workspace for several users ... SSD will make a big improvement over spinning drives in this type of environment.

"Finally - within the recommendation we've received with respect to a new replacement server, a NAS drive is included. Given that we now store all files in a cloud solution, would I be right in saying that there is no practical need for this, given that we would really only need an on-premise server to host our management software, and for print."

Never use or accept a NAS drive in a server. The credibility of any consultant that uses or suggests a NAS drive be used in server hardware should be questioned. NAS drives should only be used in NAS devices like Synology, Buffalo, QNAP, etc. Servers should have server/enterprise class drives. As for how this applies to your environment, I'm not sure I understand. Your server could be basic when only used as a DC/print and the DB ... a couple of 480/512GB SSD's in a RAID 1 would likely be plenty and could be the only drives in the system.

Yes ... just mentioned "a couple" of SSD's, as you'd want to put them in a mirror (RAID 1) for redundancy and to protect uptime of the server.

As for the processor ... contrary to what some people would have you believe, Dell does not use proprietary processors or memory. If they are marketed as "Dell", it will simply reflect an OEM-packaged CPU. It is still identical to any generic/retail version of the CPU, just in Dell wrapping with a Dell part number, designed to be sold by Dell for Dell equipment. No difference in code, etc.

Well...what can I say? That is a brilliant reply - so helpful! Thank you!

On the subject of the SSD brands/models - we can't really have them messing about, but here in the UK, the difference in cost between the Intel one you suggested and a Dell one is about £170 v £500 before tax (per SSD). It's something to ponder over.

I probably confused things with regards to the NAS. To confirm, there is no suggestion to use a NAS drive in the server hardware/box - I used the wrong terminology there. The proposal was to have a QNAP device alongside the server, but I can't see why, as we will not require any file storage. This is likely a misunderstanding on the part of the person who made the proposal, so I really need to revisit this point with them.

Therefore, am I correct in thinking that there is no practical need for the QNAP when all our files are currently off-site in a cloud hosted solution? We only need our software, database and print to be controlled by the server. That said, the software does generate PDF documents, but I can't imagine they will need a separate storage solution.

Everything else you've said makes sense and is very valuable food for thought.

On a separate note - you mentioned the terminal access; out of interest, can you think of a likely cause for why it can sometimes take between 5-10 connection attempts to connect to our software (on the server) via remote desktop? The actual connection for the remote desktop session for the server itself connects without any such issue. That may be a very open-ended question, but I thought I'd drop it in.

Thanks again for all your help with this.

Don't buy Dell parts from Dell … buy Dell parts from suppliers. I don't  know what is available in the UK, but places like serversupply.com, xbyte.com, and even amazon.com sell Dell parts but for much less than buying directly from Dell. That may help you to get the right hardware at the right price.

No, other than backups (server and/or workstations), there is no need for a NAS (from all you have described) … backups can be stored on a $100 external drive; NAS is not needed. The only other thing I can think of you needing a NAS for is if you wanted to sync the cloud storage locally for faster access, but that could likely also be done from the server instead. 

"On a separate note - you mentioned the terminal access; out of interest, can you think of a likely cause for why it can sometimes take between 5-10 connection attempts to connect to our software (on the server) via remote desktop? The actual connection for the remote desktop session for the server itself connects without any such issue. That may be a very open-ended question, but I thought I'd drop it in."

So after connecting to the server with RDP just fine, you open the software and … what? The client running inside the users' profiles doesn't connect to the software server application or database? Does it give a specific error? Just times out? Happens to everyone/all users? Happens only on the terminal server, not anywhere else the software runs? Is the server AND the software fully patched and compatible? Is there a user limit being imposed on concurrent users? You said how many users? Could also be too much load on the server.

So, after looking at the RDP issue in more detail...

Just to clarify, when I said the connection on to the server works fine, I actually had a different RDP connection/launcher in mind - i.e. where you only log onto the server desktop. That works perfect. The issue we experience is related to a different launcher, which firstly makes you log onto the server - it automatically uses your network/server user name [edit to add: actually, I think it simply takes the PC user name], and then requests input of your password. If successful, this then opens up the software login, where you enter your individual password that is specific to the software.

The problem we experience is with the first part - i.e. logging on to the server. All users experience it. It attempts to log in, but after about 2-3 seconds will display 'Logging off', and you have to start again. It can take an average of 3 attempts, but sometimes (albeit rarely) it works first time, others will need 6 or 7 attempts.

It doesn't seem to matter as to the number of concurrent users logged in (but I will continue to monitor that), and we're definitely well within the maximum number of users permitted.

The obvious solution is to speak to the software people on one hand, or the IT providers on the other. We tried that and failed. It's something we just live with, but I thought I'd drop it in here as a matter of interest.

"Is the server AND the software fully patched and compatible?"
That is a good question, but I don't know the answer, unfortunately. I'll investigate more.

"Could also be too much load on the server."
This thought crossed my mind when you mentioned the terminal access. I will dig a bit deeper on this one too.

This isn't a critical issue really - more of an inconvenience, but we'll be changing or upgrading everything soon anyway. We might never experience the issue thereafter, but I'm just a bit curious as to the likely cause in the first place.

"Don't buy Dell parts from Dell … buy Dell parts from suppliers."
Yes - Google has been my friend with this. I've had a good dig around for price comparisons.

Just one final question on the subject of the SSD "playing nice" with the RAID controller. You mentioned that the non-Dell SSDs come with many different settings. Upon connecting the SSD to the machine, is configuration necessary, or would the RAID controller be expected to detect and accept the SSD automatically?

Thanks so much again for you time.

Drive and controller negotiate communication based on their settings. If they can agree, then they talk. If they can't agree, then the drive won't show up or won't show up as usable. No configuration is needed or really even possible (I mean you can force-flash drives with "incompatible" (to the drive) firmware or use a disk editor to change settings, but that is not usually an option to "lay" IT personnel), which is why the drive you use matters.

"No configuration is needed or really even possible..."

That has pretty much answered my question - i.e. they just work, or not.

Thanks again for all your help with this. I've got a lot of valuable information to help with choices going forward. I'm sure I'll be back to clarify a few points. :-)

If you can do it on a desktop, you can do it on a server. There are more population rules for memory on a server than a desktop, but the process is the same. 

OS install is fairly straightforward … 1) you could update the LCC (LifeCycle Controller) and install the OS using the build-in deployment tool, or 2) you could install directly from the OS DVD/USB and install the drivers after install (just like installing on a computer).

Configuring the Server OS is a little different than anything you would have done on a computer - Active Directory, Print Management, etc. - but still doable.

SQL is more specialized. You may be able to go through and do the install and import the database, but your software may have specific requirements for the install/configuration, so I would consult with them first (they may also have a service (could be paid or free) available to do the install for you).

From what you have said, I would probably task out the SQL install/setup, but the hardware and OS installs should be fine. Creating users/groups, etc. isn't too technical.

Back again... how tricky would it be to install new RAM, processor and 2x SSDs for someone who is not familiar with working on servers. I've messed a fair bit with PCs and various electronic items, so don't have too much fear with pulling things apart and putting them back together. I don't expect there to be too much stress with this, but you can correct me. 

If I can successfully negotiate that bit, is the Windows OS straightforward to install? Same for SQL?

I envisage, at that stage, that user accounts (with suitable permissions) would need to be created...and so on. I would expect to hand this bit over to someone who knows what they're doing. In fact, I still expect to hand the whole thing over to someone, but would consider doing some of the preparatory work, provided it's reasonably straightforward.

Many thanks!!

Hi again.... I'm back with MORE questions! Sorry!!

If I did upgrade the T320 as per your suggestions, for how many years would you expect the machine to keep going and operating to a decent/acceptable standard? I appreciate that's a bit of an unknown, but do you have a general feeling on it (assuming usage and requirements remains fairly constant)? Would it be a solid 3 year/5 Year/7 year etc investment?