Kerberos Constrained Delegation with Hyper-V and Scale-Out File Server

When installing Hyper-V with a SMB 3 file share like a SOFS cluster, you will need to enable KDC to allow remote VM management without SCVMM.
Image the following scenario, you have a 2 nodes SOFS cluster, which hosts a high available share with a CAP called SOFS01. You have a 3 node Hyper-v cluster, running VM, hosted by your SOFS cluster. You installed a SCVMM VM for remote management:

If you want to import a VM in the Hyper-V manager you will notice the following error in the SMBClient Security log:

Pre-Server 2012 R2 you would have needed to allow the CIFS protocol for delegation in the Active Directory:

Since Server 2012 R2 you can easily configure this with Powershell. You have to have the AD Powershell tools on a server and you can configure KDC for CIFS with Powershell. After configuren SMB Delegation you can also use Get-SMBDelegation to confirm if the command was executed succesfully. It will give a list with all the server which are allow to use delegation for this SMB server:

How to monitor Storage Spaces with RDMA

Since I've started installing Storage Spaces, I discovered that monitoring and troubleshooting isn’t the easiest task.

RDMA Network Traffic

The first thing is to check if the RDMA network adapters are used or not. This can only be done by using the Performance monitor by adding RDMA Activity. In our case we can see that there is network activity on both RDMA Adapters:

To check on the client side if Smb Multichanneling is active and the network cards are RDMA capable use the Get-SmbMultichannelConnection:

Tiering Performance

When ending up on the cold tier, performance will suffer dramatically. To check the client-server traffic going to the tiers of a volume, the performance monitor needs to be used. When the Storage Tier optimization process is running you can see that the Read bytes/s are almost identical to the Write bytes/s:

By default each night at 1 o’clock the optimization process is running. When the optimization process is done (which is actually a defrag command) and eventlog entry is created with some statistics.

This statics show how many ‘hot’ data their is. At the bottom of the eventlog entry the Tier Size and the I/O hit ratio is shown. In the following case we can conclude that their is enough SSD capacity (1TB) for the 633GB of hot data. However their was ‘only’ a 82% hit ratio:

The 82% hit ratio isn’t a really bit hit ratio, I’ve seen worse cases, it also depends of the kind of actions performed that day on the storage.

In the following example we have around the same statistics but with a 97% hit ratio.