Dell Quote... good price?
-
Hi folks,
Is £53,574 GBP (pre VAT) a decent price for the following kit from Dell? I'm use to using the kit, not ordering...
The list meets business needs/goals, so not going in to those details, i'm more concerned as to if the price is a good price for the hardware, of if Dell are trying to overprice the quote/rip us off...
- 2 x N2048 Switch with Stacking Cable and 2 x Redundant Power Supplies
- 1 x 42U Rack
- 1 x KMM
- 1 x KVM
- 7 x SFP+ to SFP+ 10GbE Twinax Direct Attach cables
- 4 x T630 Servers EACH having:
2 x Xeon E5-2667 v4 procs (8 core/16 thread)
24 x 32 GB RDIMMs (totals 768 GB RAM per box)
16 x 800 GB SSD using PERC H730P 2GB NV Cache
2 x Broadcom 5719 Quad Port
1 x QLogic 57810 Dual Port 10 GbE
Dual hot plug 1600W PSUs
iDRAC 8 Enterprise
5 Year Pro SupportThe going list price online is £100k + for all of this pre VAT, but when has the online price ever been the real price?!
Negotiation with a Dell rep has got me to the above price at top of post... but is that still too high? What would you expect that kit to have been worth really? Windows Server will be licensed separately on these (so not included)...
Just to explain what the plan is with these units:
- N2048 stacked (obviously).
- Each server has 1 x SFP+ to 1 x Switch SFP+ (total 4 x cables). This will be for Hyper-V Switch/VM Network access to LAN
- Each server to have 1 x 1 GbE to the LAN for host LAN access
- Each server has 1 x SFP+ to the next server SFP+ (A -> B -> C -> D) using 3 x cable (dedicated for hyper-v replica by edit of host files)
- Two of the T630 will be Raid 10 and will run production workloads (backed up on site and off site)
- Other two T630 will be Raid 5, used for hyper-v replicas from production (replica for some VMs), and used for dev team to run new VMs, to develop on, before moving the release to the production hosts (a cycle). If Raid 5 dies, its all backed up on site and off site, so we will just restore, hence not Raid 10.
So, a good price? Excluding Windows Server 2016.
Thanks,
Jim -
Raid 10 on the production hosts and backed up.
Raid 5 on the dev hosts (for more space) and backed up. That gives room for the replica VMs we plan to replicate, and room for dev VMs to build the next release before deploy. -
@Jimmy9008 said in Dell Quote... good price?:
- Two of the T630 will be Raid 10 and will run production workloads (backed up on site and off site)
- Other two T630 will be Raid 5, used for hyper-v replicas from production (replica for some VMs), and used for dev team to run new VMs, to develop on, before moving the release to the production hosts (a cycle). If Raid 5 dies, its all backed up on site and off site, so we will just restore, hence not Raid 10.
You're running SSDs so RAID 5 as a consideration is acceptable in many cases, unlike spinning rust (drives). Considering this, do you really need RAID 10 on the production servers? Could you save some cash by reducing the number of drives and using RAID 5?
Now that said, I don't recall the line where it makes sense to move to RAID 6 on SSDs. You're talking about 16 drives at 800 GB - @scottalanmiller probably knows if you should consider RAID 6 or not.Why do your backup servers have so much more storage than your production? At RAID 5, it's nearly double. Perhaps your change rate is really high, and you want a decent amount of restore points, I'm just asking a question.
Also, do your backup servers need SSD drives for the backup data? I'm not sure the cost savings, but if it's there, perhaps spinning rust would be the way to go? -
@Dashrender said in Dell Quote... good price?:
@Jimmy9008 said in Dell Quote... good price?:
- Two of the T630 will be Raid 10 and will run production workloads (backed up on site and off site)
- Other two T630 will be Raid 5, used for hyper-v replicas from production (replica for some VMs), and used for dev team to run new VMs, to develop on, before moving the release to the production hosts (a cycle). If Raid 5 dies, its all backed up on site and off site, so we will just restore, hence not Raid 10.
You're running SSDs so RAID 5 as a consideration is acceptable in many cases, unlike spinning rust (drives). Considering this, do you really need RAID 10 on the production servers? Could you save some cash by reducing the number of drives and using RAID 5?
Now that said, I don't recall the line where it makes sense to move to RAID 6 on SSDs. You're talking about 16 drives at 800 GB - @scottalanmiller probably knows if you should consider RAID 6 or not.Why do your backup servers have so much more storage than your production? At RAID 5, it's nearly double. Perhaps your change rate is really high, and you want a decent amount of restore points, I'm just asking a question.
Also, do your backup servers need SSD drives for the backup data? I'm not sure the cost savings, but if it's there, perhaps spinning rust would be the way to go?The 2 x 60TB NAS units are used for backups. One on site, one off site. We keep a lot of backups, and these do take up lot of room. I love restore points! Not only are servers backed up daily, we also backup all workstations each night using Veeam Endpoint Free. So, it adds up fast. Real fast.
Hmm, good points. How safe is 16 x 800 GB SSD Raid 5, or Raid 6, over Raid 10?
-
@Dashrender said in Dell Quote... good price?:
@Jimmy9008 said in Dell Quote... good price?:
- Two of the T630 will be Raid 10 and will run production workloads (backed up on site and off site)
- Other two T630 will be Raid 5, used for hyper-v replicas from production (replica for some VMs), and used for dev team to run new VMs, to develop on, before moving the release to the production hosts (a cycle). If Raid 5 dies, its all backed up on site and off site, so we will just restore, hence not Raid 10.
You're running SSDs so RAID 5 as a consideration is acceptable in many cases, unlike spinning rust (drives). Considering this, do you really need RAID 10 on the production servers? Could you save some cash by reducing the number of drives and using RAID 5?
Now that said, I don't recall the line where it makes sense to move to RAID 6 on SSDs. You're talking about 16 drives at 800 GB - @scottalanmiller probably knows if you should consider RAID 6 or not.Why do your backup servers have so much more storage than your production? At RAID 5, it's nearly double. Perhaps your change rate is really high, and you want a decent amount of restore points, I'm just asking a question.
Also, do your backup servers need SSD drives for the backup data? I'm not sure the cost savings, but if it's there, perhaps spinning rust would be the way to go?We were looking at R10 for production to help minimise the chance of drive failures bringing those servers down, should they fail, we have replica VMs of the critical stuff on the second set of raid 5 servers to bring up. Though, if raid 5 is only 1% more likely to fail with SSDs over Raid 10, we would probably go with Raid 5 instead. What are the chances here?
-
@Jimmy9008 I have no idea on the difference in failure expectations, because the cost for SSD in RAID10 over RAID5 is so large and I have no workloads so critical that I had to consider that as the reason to use R10.
I would say that because you have live replication (Hyper-V or Veeam? Just wondering) running in addition to backups, that you do not need to worry about that anyway.
On to the point of your post, the cost seems reasonable for the specs you listed.
-
@Jimmy9008 said in Dell Quote... good price?:
Hmm, good points. How safe is 16 x 800 GB SSD Raid 5, or Raid 6, over Raid 10?
@scottalanmiller is the man that really understands this. So his answer will be best - but I think I read that he's busy this morning if not all day, so it might be a while before he responds.
@John-Nicholson or John Hooks might know. -
@Jimmy9008 For all new equipment, that seems to be a good price. Trying a different vendor/retailer will probably net you a higher price at this point because of how most OEM handle quotes. No idea if Dell is one of those or not because I haven't bought new in years.
-
I should also mention that the shear number of drives doesn't really matter here, it's the size of the arrays.
16 x 800 GB = 12,800 TB RAID 0
As this article points out http://www.theregister.co.uk/2015/05/07/flash_banishes_the_spectre_of_the_unrecoverable_data_error/, 12 TB on consumer drives have a near 100% chance to hit a URE while rebuilding in parity RAID.
I have forgotten if you consider the parity drives space as part of the space or consider remove it from consideration when looking at UREs.
The article does show that even consumer SSDs have 10x lower URE rates than business class spinning drives, and business class SSDs have 100x lower URE rates than business class spinning drives.
-
@travisdh1 said in Dell Quote... good price?:
@Jimmy9008 For all new equipment, that seems to be a good price. Trying a different vendor/retailer will probably net you a higher price at this point because of how most OEM handle quotes. No idea if Dell is one of those or not because I haven't bought new in years.
That's what I think, but wanted to check. Rarely get budget here to improve things. We have a budget and the board expect us to spend it all. If not, I don't get it again for a long time. I think its a fair price, but just checking. If somebody said nope, that's 10k too much... i'd go back to Dell asking for more off for example, then with the price dropped, add more hardware
-
@Dashrender said in Dell Quote... good price?:
I should also mention that the shear number of drives doesn't really matter here, it's the size of the arrays.
16 x 800 GB = 12,800 TB RAID 0
As this article points out http://www.theregister.co.uk/2015/05/07/flash_banishes_the_spectre_of_the_unrecoverable_data_error/, 12 TB on consumer drives have a near 100% chance to hit a URE while rebuilding in parity RAID.
I have forgotten if you consider the parity drives space as part of the space or consider remove it from consideration when looking at UREs.
The article does show that even consumer SSDs have 10x lower URE rates than business class spinning drives, and business class SSDs have 100x lower URE rates than business class spinning drives.
So going by that article, Raid 5 at this size will be fine for us and being raid 6 or 10 is just lost space.
-
If Raid 5 really is safe enough, i'd really like to drop a few of the SSDs and instead by 10 core or 12 core procs for each host and more on M$ licensing instead.
-
OBR5 using SSD's is safe enough, because SSD's don't have the URE risk (among other legacy concerns). Generally, when an SSD fails, it just stops working so the array is degraded. You install a new drive, the system resilvers and you're off to the races.
Decreasing the drive count may be possible, if it will fit your storage needs (by purchasing higher capacity SSDs).
-
@DustinB3403 said in Dell Quote... good price?:
OBR5 using SSD's is safe enough, because SSD's don't have the URE risk (among other legacy concerns). Generally, when an SSD fails, it just stops working so the array is degraded. You install a new drive, the system resilvers and you're off to the races.
Decreasing the drive count may be possible, if it will fit your storage needs (by purchasing higher capacity SSDs).
Understood
So for the price, is that an expected cost for the kit or reasonable?
-
@Jimmy9008 said in Dell Quote... good price?:
@DustinB3403 said in Dell Quote... good price?:
OBR5 using SSD's is safe enough, because SSD's don't have the URE risk (among other legacy concerns). Generally, when an SSD fails, it just stops working so the array is degraded. You install a new drive, the system resilvers and you're off to the races.
Decreasing the drive count may be possible, if it will fit your storage needs (by purchasing higher capacity SSDs).
Understood
So for the price, is that an expected cost for the kit or reasonable?
I don't want to comment on the price quoted as I honestly have no clue how much things are where you are purchasing for/from.
I would expect the drive cost to go up substantially if you go to higher capacity drives (which will take away from your budget for MS licensing etc)
-
@DustinB3403 said in Dell Quote... good price?:
OBR5 using SSD's is safe enough, because SSD's don't have the URE risk (among other legacy concerns). Generally, when an SSD fails, it just stops working so the array is degraded. You install a new drive, the system resilvers and you're off to the races.
Decreasing the drive count may be possible, if it will fit your storage needs (by purchasing higher capacity SSDs).
Of course SSD has URE. Where would you even get the idea that it doesn't?
-
@JaredBusch said in Dell Quote... good price?:
@DustinB3403 said in Dell Quote... good price?:
OBR5 using SSD's is safe enough, because SSD's don't have the URE risk (among other legacy concerns). Generally, when an SSD fails, it just stops working so the array is degraded. You install a new drive, the system resilvers and you're off to the races.
Decreasing the drive count may be possible, if it will fit your storage needs (by purchasing higher capacity SSDs).
Of course SSD has URE. Where would you even get the idea that it doesn't?
I haven't been living under a rock, SSD's do not have the same URE risk that winchester drives have.
-
@DustinB3403 said in Dell Quote... good price?:
@JaredBusch said in Dell Quote... good price?:
@DustinB3403 said in Dell Quote... good price?:
OBR5 using SSD's is safe enough, because SSD's don't have the URE risk (among other legacy concerns). Generally, when an SSD fails, it just stops working so the array is degraded. You install a new drive, the system resilvers and you're off to the races.
Decreasing the drive count may be possible, if it will fit your storage needs (by purchasing higher capacity SSDs).
Of course SSD has URE. Where would you even get the idea that it doesn't?
I haven't been living under a rock, SSD's do not have the same URE risk that winchester drives have.
That is not what you said. you said it doesn't have URE risk.
-
Literal horse today I see.....
-
@Jimmy9008 said in Dell Quote... good price?:
@Dashrender said in Dell Quote... good price?:
@Jimmy9008 said in Dell Quote... good price?:
- Two of the T630 will be Raid 10 and will run production workloads (backed up on site and off site)
- Other two T630 will be Raid 5, used for hyper-v replicas from production (replica for some VMs), and used for dev team to run new VMs, to develop on, before moving the release to the production hosts (a cycle). If Raid 5 dies, its all backed up on site and off site, so we will just restore, hence not Raid 10.
You're running SSDs so RAID 5 as a consideration is acceptable in many cases, unlike spinning rust (drives). Considering this, do you really need RAID 10 on the production servers? Could you save some cash by reducing the number of drives and using RAID 5?
Now that said, I don't recall the line where it makes sense to move to RAID 6 on SSDs. You're talking about 16 drives at 800 GB - @scottalanmiller probably knows if you should consider RAID 6 or not.Why do your backup servers have so much more storage than your production? At RAID 5, it's nearly double. Perhaps your change rate is really high, and you want a decent amount of restore points, I'm just asking a question.
Also, do your backup servers need SSD drives for the backup data? I'm not sure the cost savings, but if it's there, perhaps spinning rust would be the way to go?The 2 x 60TB NAS units are used for backups. One on site, one off site. We keep a lot of backups, and these do take up lot of room. I love restore points! Not only are servers backed up daily, we also backup all workstations each night using Veeam Endpoint Free. So, it adds up fast. Real fast.
Hmm, good points. How safe is 16 x 800 GB SSD Raid 5, or Raid 6, over Raid 10?
RAId 5 isn't bad. But RAID 6 for production would be excellent.
