what does mean All flash profiles (AF-8 , AF-6 , AF-4)

[Bab]

I know we have 3 type of All flash profile

AF-8 , AF-6 , AF4

but according to follow pic :

 

 

 

 

 

 

 

 

1- Does AF-8  , 80K IOPS means the total size of capacity tier on each node is Maximum 12TB  and our max IOPS that can provide is 80000  ?

2- Does AF-6  , 50K IOPS means the total size of capacity tier on each node is Maximum 8TB  and our max IOPS that can provide is 50000  ?

3- AF-4  , 25K IOPS means the total size of capacity tier on each node is Maximum 4TB  and our max IOPS that can provide is 25000  ?

[IRIX201110141]

The AF-X referrers to sizing profiles.

The table you posted have nothing to do with the capacity tier. It describes the amount of cache tier (write buffer) when using medium level endurance (10 daily writes) flash drives and what to expect. When using AF the 10% rule of thumb isnt a good choice and its not necessary to select lllaaaaarrrge cache drives.

Regards,
Joerg

[Bab]

Thanks . I have read this document

https://blogs.vmware.com/virtualblocks/2017/01/18/designing-vsan-disk-groups-cache-ratio-revisited/

but actually now according that I want to now what is the capacity of my cache tier (write buffer ) when using all flash

for example in row1 in the table

 

That it saying while our workload is read intensive more than 70% of workload is read and about 30% is write that is better to have 400GB write buffer cache for AF-8 (because it is using two disk group so divided  2 for one disk group).

Actually I am confused for this part : I am not using ready node now what does mean AF-8 for me now imagine my workload is read intensive now I which one is the size of my write buffer 400G for AF-8 ?  200G for AF-6 or 100G for AF-4 ?

 

[IRIX201110141]

If you choose 2x100 you'll get something around 25K IOPS. Is this enough for your expected workload and all the backend io,de-staging and so on you can go with it.

Regards,
Joerg

[Bab]

Thanks but I asked about what does mean AF-8 for the first row I could understand from the first row when my workload in vSAN in read intensive more than 70% read and about 30% write the recommendation according to that table is :

for AF-8 = the size of write buffer disk is 400G

for AF-6 = the size of write buffer disk is 200G

for AF-4 = the size of write buffer disk is 100G

1- How can understand my structure is AF-8 or AF-6 or AF-4 ? I am not using ready node

2- I know iops is about performance for example 80k IOPS means 80000 IOPS now what does mean in that table

when write AF-8 , 80K IOPS what does it mean would you please give an example ?

[Duncan Epping]

AF-8/6/4 is just a name VMware gave to the profile of the configuration. AF=All Flash, and "4" / "6" / "8" just is what they would call the "size" or performance profile. It is nothing to really think too long about. It is like a t-shirt size.

[Bab]

I have read that document I think all of that doc can be summary in that table.

Ok . Now my workload type is write intensive about 45% is write intensive.  So according to that table my workload type is Medium writes , mixed workload . now . What is the size of write buffer disk? which size I have to select ? . I am using all flash

 

[Duncan Epping]

---

Mid-range workloads (Greater than 30% write but still random) – Can make do with less cache. 1.2TB for the most performance, and 400GB for lower demands. A slightly larger write buffer helps the SSD keep up with sustained de-stage speeds. 

 

---

 

[Bab]

Ok . So I edited the table for myself is that correct ?

 

 

[Bab]

Is that my table correct ?

[a_p_]

Please take a look at the title in the screenshot that you've initially posted.
It says "Total Write Cache Per Host"

André

[Bab]

So thanks.  I edited it .

Is this correct ? specially headers that that I replaced with AF-8 , AF6 , AF4

 

 

[Bab]

Is this table correct for AFA vsan write buffer size ?