Gen10 hosts affected (BL460c, DL380) with ESXi versions (6.5 - tested latest build 16576891, 6.7 - build 16316930). ESXi 7.0 (latest build 16324942) seems OK because Sensor Status is Uknown (for Memory Device). A recent feature in VMware ESXi allows you to use host SSD-caching as stopgap memory-management technique for memory-constrained servers. One of the more recent features that is available with ESXi. Easy to install ESXi (no hacking of drivers or VIBs) Break the 64 Gb RAM barrier so It's possible to go and have some nested ESXi and stay on single host. Well designed, robust case; Power efficient PSU; Power efficient platform (DDR4, Xeon E5-2630L) So that's why I opted for socket 2011 R3 and low power Intel Xeon E5-2630L v3 CPU which has.
vSphere 6 uses the well known memory reclamation techniques you may already know from previous versions:
- transparent page sharing (TPS)
- memory ballooning
- memory compression
- memory swapping
The memory reclamation technique that is used depends on the ESXi host memory state, which is determined by the amount of free memory of the ESXi host at a given time.
With vSphere 6 VMware introduced a new memory state, called 'clear state'.
Gen10 hosts affected (BL460c, DL380) with ESXi versions (6.5 - tested latest build 16576891, 6.7 - build 16316930). ESXi 7.0 (latest build 16324942) seems OK because Sensor Status is Uknown (for Memory Device). A recent feature in VMware ESXi allows you to use host SSD-caching as stopgap memory-management technique for memory-constrained servers. One of the more recent features that is available with ESXi. Easy to install ESXi (no hacking of drivers or VIBs) Break the 64 Gb RAM barrier so It's possible to go and have some nested ESXi and stay on single host. Well designed, robust case; Power efficient PSU; Power efficient platform (DDR4, Xeon E5-2630L) So that's why I opted for socket 2011 R3 and low power Intel Xeon E5-2630L v3 CPU which has.
vSphere 6 uses the well known memory reclamation techniques you may already know from previous versions:
- transparent page sharing (TPS)
- memory ballooning
- memory compression
- memory swapping
The memory reclamation technique that is used depends on the ESXi host memory state, which is determined by the amount of free memory of the ESXi host at a given time.
With vSphere 6 VMware introduced a new memory state, called 'clear state'.
So vSphere 6 knows five different memory states associated with one or more memory reclamation techniques to reclaim memory:
But which treshold of free memory is associated with which memory state?
ESXi uses a value called 'minFree' for the memory state calculation. minFree is a dynamic value and depends on the ESXi host memory configuration.
You can calculate minFree very easy for your ESXi host:
for the first 28 GB of physical RAM in the ESXi Host: minFree = 899 MB
+ add 1 percent of the remaining RAM to your calculation
Figure: „minFree calculation example vSphere 6:
In the example above the ESXi host has 100 GB memory:
for the first 28 GB RAM minFree = 899 MB, for the remaining 72 GB (100 GB – 28 GB) we have to add 1% to minFree: 1% of 720 GB = 720 MB -> minFree is 899 MB + 720 MB = 1619
Thresholds:
- high state: enough free memory available
- clear state: <100% of minFree
- soft state: <64% of minFree
- hard state: <32% of minFree
- low state: <16% of minFree
Esxi Host Memory Slots Online
If you want to know the memory state of one of your ESXi hosts you can use ESXTOP (extract from the vSphere 6 ESXTOP quick Overview for Troubleshooting' diagram):
Open ESXTOP and type 'm' for the memory tab. The host memory state is displayed in the first line on the right.
Esxi Host Memory Slots Free
You want to learn more about ESXTOP?
Esxi Host Memory
Take a look at the vSphere 6 ESXTOP quick Overview for Troubleshooting' diagram: