VMWare - Datastore, File System, Block Size for vESA?

[patelc3]

I have some VM running on VMWare ESXi, 6.0 and during business hours, when massive volume of email is sent out, the workqueue bloats as it does what it does.  Knowing how memory and disk performance is key for VM to run optimally, i found that the Datastores file system's block size is set to 1MB.  Reading thru the documentations, and knowing that a C600V requires 1TB of storage, shouldn't the block size be set to 8MB?  Also, for the datastore capabilities, what are the optimal settings for it?  Currently, Thin Provisioning is set to supported, Storage I/O Control...Status is Disable, Mode is set to 90% of peak throughput, Storage DRS I/O Metrics is Enabled and Statistics Collection is set to Disabled.

Hints? Feedback?  Help the one person that doesn't know VMWare that well? :-)

[marc.luescherFRE]

Scalability of VM

We run 24 vESA600 on VMWARE ESXi 6.5, and we still have days where out 6 inbound gateways can not keep up.

First make sure you DO NOT modify the settings requested by the deployment images (8/8 and 512 GB disk).

Network adapter  still the dear old NE1000.

Make sure you apply the following three settings to your config to adapt timekeeping.

monitor_control.disable_tsc_offsetting=TRUE
monitor_control.disable_rdtscopt_bt=TRUE
timeTracker.forceMonotonicTTAT=TRUE

Make sure you check VMWARE HA settings.

Sizing the Cisco sizing guideline says a VESA600 should be able to handle 12-15 messages per second. We have seen cases as high as 30 per second and as low as 8 per second, depending on # of URL's and attachment types. If we get more then 90% of load for extended periods of time we just deploy two more ESA in our cluster and go from there.