VSAN 6.2 Part 11 – Support for higher FTT with larger VMDKs
In the VSAN 6.0 Design & Sizing Guide, a caveat was placed around the size of a VMDK, and the Number of Failures to Tolerate (FTT) number. It reads like this:
“If the VMDK size is greater than 16TB, then the maximum value for NumberOfFailuresToTolerate is 1.”
I’m pleased to say that this restriction has been lifted in VSAN 6.2.
That limit was due to the metadata size. You might recall if you are a regular reader of this blog that each component can only grow to a maximum size of 255GB. With a 62TB VMDK, that a lot of components. Now if you factor in the Number of Failures to Tolerate, with a value of 1, you double the components, with 2 you treble it and with 3 (the maximum) with quadruple it. These all need to be tracked in metadata.
If the object configuration is too large, then we run the risk of the object configuration size being larger than the assigned metadata. That is why we added that Number of Failures to Tolerate limit restriction for large objects in VSAN 6.0 and 6.1.
However, VSAN engineering have implemented a compression method for metadata in VSAN 6.2. This now allows us to remove the limit. Now you can have very large VMDK with higher Number of Failures to Tolerate values, if you so wish.
5 Replies to “VSAN 6.2 Part 11 – Support for higher FTT with larger VMDKs”
It would be nice to have a possibility to create multiple datastores in one VSAN cluster. Eg. one datastore with all-flash disk groups and another with hybrid disk groups. Is it on the roadmap?
This is not the first time we have heard this request – however I cannot comment on roadmap or futures.
VSAN 6.2 design and sizing Guide mentioned the same things VSAN 6.0.
Is it document error??
Note: This is only true if the VMDK size is less than 16TB. If the VMDK size is greater than 16TB, then the maximum value for NumberOfFailuresToTolerate is 1
I’ve let the author know – thanks.
Comments are closed.