The Latest Gartner® Magic Quadrant™Hyperconverged Infrastructure Software
Moderators: anton (staff), art (staff), Max (staff), Anatoly (staff)
Yes. But we use HA images in most of our setups; it should do work just great. What you can do is presenting a test HA image directly to the VM. NOTE: it is NTFS volume, be careful with mounting it anywhere else.Does your software have options to place image files directly on the disk without creating volumes and formatting them?
That should be sufficient. The settings though seem a little bit strange. What I would recommend is re-benchmarking everything in this way https://www.starwindsoftware.com/best-p ... practices/. Let me know if you have any questions.I used iometer and didn't format volume on my vhdx file. Instead of this I used cylinder numbers that should be equal file size about 250 Gb
I dont think that it may reduce latency . Anyway I made my experiment according to disk manufacturer parameters.What I would recommend is increasing queue depth while benchmarking this setup.
Try disabling caching or setting it to Write-Back mode.
It provides compute resource high availability.high availability provided by windows failover clustering.
Unfortunately, that is not possible as every setup has unique numbers.reference performance benchmarks for your software
Yes, that is I needIt provides compute resource high availability.
You can publish information for few reference architectures. It will very helpful.Unfortunately, that is not possible as every setup has unique numbers.
The iscsi layer is under vhdx file on the pictureOn your drawing, it seems that you missed iSCSI.
I need to discuss that with my team. Will reach out to you here with more details.You can publish information for few reference architectures. It will very helpful.
Yes, please.So, I will make network tests, create ram disks on my nodes, configure HA disk on it and then run performance tests, according performance benchmarking best practices.