I have a SQL Server (2008 R2) WSFC on Windows Server 2008 R2 on which I am experiencing some storage latency between the OS and SAN. The SAN admin has noted that the SAN is experiencing very little pressure on the LUNs associated with the server noting that the highest response time is 8ms.
However, the OS and SQL Server is not getting that quick enough. I found a few questions referring to TCP Chimney Offloading and some issues that it may cause so I've been exploring the configuration on the server.
Following this guide, I see that at the OS level, Chimney Offload State is automatic
Querying active state...
TCP Global Parameters
----------------------------------------------
Receive-Side Scaling State : enabled
Chimney Offload State : automatic
NetDMA State : enabled
Direct Cache Acess (DCA) : disabled
Receive Window Auto-Tuning Level : normal
Add-On Congestion Control Provider : ctcp
ECN Capability : disabled
RFC 1323 Timestamps : disabled
However, the settings on the adapters in Device Manager show TCP Connection Offload (IPv4)
as "Disable."
OS: Windows Server 2008 R2 Enterprise Edition SP1 Network Adapters: HP NC551i Dual Port FlexFabric 10Gb Converged Network Adapter.
There are 8 of these FlexFabric adapters (which are the ones that I checked for chimney offload) but only 3 are enabled (one of which is to a "Private" network for the cluster). There is also an HP Network Team to a "Public" network.
What are the implications, if any, of having the chimney offload "automatic" on the OS but disabled at the card?