![]() I will take a look the IOPs and report back within the hour.įrom one vdisk to another vdisk in the same guest: 550 MBps, The reservations did zip zero and nothing for the copy speed - exact same pattern of behavior. The one thing I was able to try was reserving CPU and RAM resources for the file\DB server that seems to be the cause of the problem. Thanks guys! Below I've replied to most everyone. What am I missing and more importantly, how do I get the file copy up to where I want it to be. I mention NIC settings that are ENABLED by default because it has been suggested that DISABLING the settings above might improve file copy throughput. On the VMXNET3 NIC in the guest the following are still ENABLED Set Large Send Offload V2 to disabled for both IPv4 and IPv6 Set Recv Segment Coalescing (IPv6) to Disabled. Set Recv Segment Coalescing (IPv4) to Disabled On the VMNET3 Nic in the guest set the following are DISABLED Netsh int ip set global taskoffload=disabled Netsh int tcp set global ecncapability=Disabled Netsh int tcp set supplemental custom congestionprovider=none Netsh int tcp set global timestamps=Disabled Receive Window Auto-Tuning Level: netsh int tcp set global autotuninglevel=disabledĪdd-on Congestion Control Provider: netsh int tcp set global congestionprovider=none Receive-Side Scaling: netsh int tcp set global rss=disabled Set Microsoft network server: Digitally sign communications (always) to Disabled.ĭisable TCP global settings run the commands from an elevated prompt:Ĭhimney Offload: netsh int tcp set global chimney=disabled Set Domain member: Digitally encrypt or sign secure channel data (always) to Disabled. In the Server Domain Controller GPO edit the following::Ĭomputer Configuration/Policies/Windows Settings/Security Settings/Local Policies/Security Options Once the problem is fixed then we can talk about what configuration is best practice.) (Just to remind everyone, I don’t need to be told that some (many? Most? all?) of the “fixes” listed below are undesirable. NOTE: Besides an instantaneous spike in Disk reads, none of the monitored parameters spikes to when a file copy is initiated.Ĭopy a 2GB file from one of the guest servers to another guest server in 30 seconds or less (and that’s slower than it should be). RamMap from SysInternals shows 50% RAM usage under heavy load much oĭisk Queue length rarely even 1 - usually at 0.5 Metafile use is at about 200K total for both active and used RAM Here’s the Operating Environment under heavy loadĪll host hardware reports to be healthy to vSphere.Īverage disk latency as measured at the host is a steady has been see to get as high as 36ms average of an hour of heavy loadĪverage CPU usage is < 20 % under heavy load.and rarely spikes over 50%įile cache configuration set using MS API setsystemfilecachesize MSSQL 2012 database is used on the primary applicationĪll AV (including Windows Defender) has been uninstalled - not ust disabled - on the file\database server. Windows updates for all guests are up-to-dateĭNS and DHCP are AD integrated and working and scavenging correctly Guest for file\database server has 1 CPU (I’ve not attempted giving it 2 CPUs yet)Īll NICs and switches in the network are GB (not relevant yet as copy test is guest to guest. Guest NIC is latest version of VMNET3 driver Guest vmtools is most recent version 10.1.7 Physical NICs NetXtreme BCM5720 Gigabit Ethernet PCIeĪll hardware is supported by VMWare v6.5 U2ġ Windows 2012 R2 Server as file server and database serverġ Windows 2012 R2 Server as an MFA server and home for administrative loginsġ Windows 2016 Server Essentials as the DC Other symptoms reported by the client include freezing of their PCs and losing connection to the server.ĭell PowerEdge T430, 1 6 core CPU, 32 GB of RAM, 6 disk hardware RAID 5 array. The software vendor, however, will not help until the file transfer of a 2GB file takes place in a reasonable amount of time - “reasonable” is up for definition but I think that even a steady 100 MBps would be acceptable. I need advice on what I should do.įile transfers start off at a reasonable speed between server on the same VMWare host and then drop to almost nothing - occasionally they will actually stop and the copy job “never” finishes.Īn update to a mission critical application where the problems manifest themselves the most was done about a month ago. I don’t need advice on what I shouldn’t have done. NOTE: Many of the “fixes” that I have implemented may or may not be best practices or even desirable. What am I missing? This has GOT to be simple!! I’ve been working on this for almost thirty hours how and need to give this out to the hive mind that is SW.
0 Comments
Leave a Reply. |
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |