vSAN few questions

[Liksx]

Hello!

We have SQL DB and now this DB runs on ( SAS Disks )

We have big trouble when allot of people read from DB ( 6-7k IOPS )

now we wanna buy 3 or 4 PCIE NVMe intel SSD and create vSAN cluster for this mirror DB where people can read from DB and have alot of IOPS

but now i testing this case and have few questions....

i have 4 ESXI nodes where have 20GB SSD Cache and 40GB SSD Capacity

after configuring cluster and vSAN datastore ( with default policy or create another one where i chose FTT1 )

WHY when i using FTT 1 i see ~160Gb capacity?

i have 4 NODE and 40GB SSD Capacity ( 4 Node X 40GB =160GB - overhead )

i confused.....

i add another host to this cluster ( where reside DB ) and create in VM new VMDK160GB!! and i can do this!...

next i init this disk and create MBR partition in Windows

when i put in this partition 80GB files copy copy copy and VM Stuck..... then i get message about capacity on this datastore end etc.

my question why when i create vSAN ( with default policy too ) i see ALL capacity same as JBOD.... really confused

what i want?

i have 4 nodes installed 400GB SSD on it

if 1 nodes fails, anoter node automaticly rebuild vSAN

1 node witness

2-3 nodes mirror

4 node hotspare

i want have 400GB capacity....

if compare with RAID tech  summary capacity nodes - 1 nodes ( mirror ) - 1 nodes ( hotspare ) dont sure about witness BUT! anyway 400GB ( mirror ) 400GB ( hotspare ) i can't use but in vSAN i see all capacity

sorry for eng

[CHogan]

A few responses which might help.

1. Cache devices never contribute to capacity of the VSAN Datastore, only capacity devices do that. So 4 devices x 40GB = 160GB VSAN datastore as you point out.

2. There is no concept of individual hot spare devices. The whole of the cluster acts as a hot spare. If there is a disk failure, the components belonging to VMs on that disk are rebuilt on the remaining capacity in the cluster (assuming FTT > 0)

3. The policies are not associated with the VSAN datastore - they are associated with VMs. VMs on the VSAN datastore can have different policies. In fact different VMDKs belonging to the same VM can have different policies.

4. When you deploy a VM with a policy (default is the same as FTT=1) you can see its components distributed across the hosts and disks in the cluster.

5. You can create VMs that are bigger than VSAN datastore size because VMs are deployed as "thinly provisioned". They can grow over time. But if you consume all 160GB, then your VM will be suspended until new storage is assigned to the VSAN cluster.

6. I don't know how you are doing the copy, but it could be that it is inflating the VMDK. Now with FTT=1, and a VM size of 160GB, you will fill up the VSAN datastore since there are 2 x 160GB copies of the data.

Hope this makes sense

Cormac

[Liksx]

thanks for helpfully reply

mmmm all ok?

4 Nodes

each nodes have own Fault F1 F2 F3 F4

at this point i can create 30GB Disk ( 4 * 8 = 32 - overhead = 30.19GB )

but, Virtual SAN storage consumption = 60GB ( FTT1 )

ok go windows ->

put 1 file to this patition

we have vmdk some space + 360MB from this .iso on vSAN storage

Capacity! this vSan storage 30.20GB!

ok go next and go copy files again and again

andddddddd VM Stuck!

VM shuttdown imidettly lol..... ))

ok how much space i used??

in vShpere Client

in vCenter

in windows ONLY 10GB!

what i doing worng? thanks for help

i dont know really how need monitoring vSan storage space

[CHogan]

Is your C:\ on VSAN?

How big is your C:\ drive?

Remember that this will also be FTT=1.

[Liksx]

Question №2

in my prelab i using nested ESXI

in Guest OS resides on ESXi

in Nested vSan

where bottleneck? really nested ESXI can impact on perfomance ?

[Liksx]

no C:\ drive local storage only E:\ vSan

i have 5 Hosts

4 Hosts vSAN

5 Host where my SQL DB

i obtain single cluster and have 5 hosts on it

SQL have local storage and C:\ and other drive on it

only E:\ connected how vSan