Help choosing replacement Hyper-V host machines and connected storage
-
Again I am learning still
but I think we will get a few Scale people in here answering questions for ya soon. -
@JohnFromSTL said:
- Replication between servers in the event of host server failure
- Load Balancing
- Storage
- Hyper-V client machines (12TB - 14TB)
As far as connectivity is concerned I would probably use 10GbE between the Hyper-V host servers and the storage.
With only two nodes, shared storage should not come up as an option. There is both no way to afford it and it would defeat any value in having two of these nodes. Your budget doesn't allow for this option. The starting point for something like that would be an EMC VNX to be reliable enough to even slightly consider having two nodes.
As I see it, your choices are going down to a single server or going to two servers and using replicated local storage. Even that will be tough, but external shared storage both doesn't meet your goals not fit in your budget.
-
@JohnFromSTL said:
Due to our size and budget constraints we purchase refurbished equipment, which are typically 2-3 years old. Whichever solution I end up with needs to last 3-4 additional years.
Refurb is good. @xByteSean is around to help out with that. As is @ryan-from-xbyte
-
Dell PowerEdge R910 are monster servers but lack a lot of storage options. How is your CPU utilization? That is going to be the killer as those are quad processor systems. So even though they are two generations old, they are twice the size of normal servers. You are going to be looking at the Dell R730xd loaded to get the storage and the CPU that you need to come in around where you were before. That's not going to be possible, I don't think, in your price envelope.
-
Xbyte has a lot of great options for you and some great warranties to go with them.
-
@JohnFromSTL said:
@Minion-Queen said:
NTG can help with the Scale questions so ask away!
Any idea on starting price?
Did I miss your total memory needs? How much memory are you using added all up?
-
@JohnFromSTL said:
That's the fun part; I'm not given a true budget, but knowing how tight the pursestrings are $12-13k would be the max.
Total, for two compute nodes and all of the storage? That's closer to the price of one node. If you wanted a SAN to handle this (you don't, but just saying) you'd be around $35K - $40K for the SAN alone.
-
Just catching up on the thread now. Quick overview on what we do is that we combine servers, storage and virtualization into an appliance for running VMs (HC3 is the product name). Super simple to use product with HA out of the box. Starting price is $25,500 for 3 nodes of our HC1000 product (including 1 year of ScaleCare support). Happy to answer any other questions I can, so fire away!
-
@dafyre said:
That's right! I forgot you guys are Scale partners (that the right word?) now!
Yup, right term.
-
That is a huge amount of SQL Server workloads. Figuring out the CPU and memory needs for that will be the biggest part of doing capacity planning.
-
@scottalanmiller said:
@JohnFromSTL said:
- Replication between servers in the event of host server failure
- Load Balancing
- Storage
- Hyper-V client machines (12TB - 14TB)
As far as connectivity is concerned I would probably use 10GbE between the Hyper-V host servers and the storage.
With only two nodes, shared storage should not come up as an option. There is both no way to afford it and it would defeat any value in having two of these nodes. Your budget doesn't allow for this option. The starting point for something like that would be an EMC VNX to be reliable enough to even slightly consider having two nodes.
As I see it, your choices are going down to a single server or going to two servers and using replicated local storage. Even that will be tough, but external shared storage both doesn't meet your goals not fit in your budget.
And there are not many servers which have that amount of local storage.
-
@JohnFromSTL said:
@dafyre said:
I'd definitely recommend reaching out to the Scale Computing guys (are there any on here?) here or either on SW. For information about pricing. It would be new kit and would last you several years. I think they can do 2 node setups as well.
Thanks, I'll check them out.
Scale would very likely be an amazing fit for your needs. Easy to manage, highly reliable with built in high availability and auto-node recovery, etc. All of the compute and storage is built into the cluster. So it is a single price for everything, so no surprises down the road.
-
@JohnFromSTL said:
And there are not many servers which have that amount of local storage.
Correct, that is why I mentioned the R730xd, because it is one of the biggest local storage devices on the market.
-
@scottalanmiller said:
Dell PowerEdge R910 are monster servers but lack a lot of storage options. How is your CPU utilization? That is going to be the killer as those are quad processor systems. So even though they are two generations old, they are twice the size of normal servers. You are going to be looking at the Dell R730xd loaded to get the storage and the CPU that you need to come in around where you were before. That's not going to be possible, I don't think, in your price envelope.
Host01 hovers around 9-10% CPU and 55% memory utilization.
-
@JohnFromSTL said:
Host01 hovers around 9-10% CPU and 55% memory utilization.
Which CPUs and how many do you have in that host?
How much memory is in that host?
-
I currently have a Dell R720xd running Hyper-V 2012R2. It works really really well. Not sure on the price tag, but that is a nicely built system. Raid controller can do RAID 10, and RAID 6 in addition to the others... Add ram to meet your needs.
-
@dafyre said:
I currently have a Dell R720xd running Hyper-V 2012R2. It works really really well. Not sure on the price tag, but that is a nicely built system. Raid controller can do RAID 10, and RAID 6 in addition to the others... Add ram to meet your needs.
We have a monster R720xd with 128GB too. Sweet machine. R730xd has even bigger CPU options.
-
@dafyre said:
Raid controller can do RAID 10, and RAID 6 in addition to the others.
There are others?
That controller has a 1GB cache too, which is important, AND has CacheCade built in. The R730xd loses the CacheCade option.
-
@scottalanmiller
RAID 1, most notably... But I didn't want to mention RAID-5 spits on the floor. -
If the R910 is maxing out at, say, 20% CPU, then my guess is that an R720xd will do the trick to take over its load. The R720xd has two, faster procs than the R910. Not only are the individual procs faster, but by moving from quad procs to dual procs you gain a small amount of efficiency just from that one move. So faster procs and more efficient proc usage and then cutting the total number of procs in half.... seems like you will be okay.
