This post is a reflection post on possibilities on different scenarios whether you’re using a VMware vSAN certified storage controller supporting pass-through (or also IT Mode or JBOD mode… all this is the same name) or you’re using a storage controller which uses RAID0 only. I said only as it means that it is limited in functionalities and possibilities, but also some constraints. In this post we will details the differences in VMware vSAN Pass-Through vs RAID0 Storage Controller.
IT Mode (Initiator Target mode… did you know that?), pass-through or JBOD are the same words for the same thing. It means that each individual disk attached to the storage controller will appear as an individual disk to vSAN. Not as a RAID0 volume. Through the vSAN management interface, you’ll see the individual disks in this case. That’s the first point.
Another point is availability. While both RAID0 and Pass-Through modes are supported by vSAN, not both modes are doing the thing that you expect they do. (follow?). If I say that when you need to replace a failed disk in your host, you’ll be certainly pleased to find out that your storage controller, configured in pass-through can use hot-plug while the one configured for RAID0 has to:
- Reboot the host to configure RAID0 for that disk (you’ll add the disk to RAID0 volume).
- Attach the disk to vSAN (via the vSAN UI) if “Add Disk to Storage” is in manual mode.
So the big disadvantage is a need to reboot the host. Imagine that you have two disk groups in this host…
As you know, since several versions, vSAN allows migrating data from the individual disks (if they’re visible). In the case you have your storage controller configured or your storage controller only supports RAID0, then you basically have to migrate the whole disk group as the RAID0 volume appears through the vSAN UI as a single volume (a single-drive RAID0). Not as multiple disks attached to the controller via the JBOD, pass-through or IT Mode….
Here is the screenshot from the lab which shows the individual disks of the capacity volume. LSI 3008 is configured in pass-through mode (or IT Mode if you like).
Advantages one over the other?
- Need to reboot the host > add the new disk to the RAID0 volume.
- Cannot evacuate data from individual disks in the capacity tier
- All good here. Individual disks can be hot-replaced without the need to enter them into some RAID configuration. One thing to keep in mind. The disk cannot have any partition information.
- The pass-through mode allows the hypervisor to have greater control over the drives in the capacity tier group.
Deduplication and compression Gotcha
Note that it is not possible to remove disks from a disk group after deduplication and compression is enabled on the cluster. You should think ahead and add all the capacity that you need before activating the dedupe.
Also, if there is a failure of any of the disks in the disk group, a whole disk group will be impacted. (Not only the particular disk). You will have to wait for the data in the disk group to be rebuilt somewhere else within the cluster. Change the failed disk (go to bios, add the disk to RAID0 if your disk controller supports only RAID0 or if uses RAID0 instead of pass-through).
Then only you’ll be able to recreate the disk group and let the cluster to re-balance the components.
All this you did not know when checking the deduplication and compression check box. Did you? Those considerations should be taken into account from administrators perspective when planning to use the saving capacities of deduplication and compression on vSAN.
I’ve been always a supporter of vSAN technology, and I’m still. It’s good to see the fast adoption. By sharing the information to a largest number of readers I hope to give the best possible answers to folks which don’t have that information or which simply ignored that.
If I use a RAID 0 configuration, you’ll need to configure the disk as RAID0 before attaching it. And this usually happens through the BIOS of the storage controller. Or perhaps some vendors integrated the RAID0 management through remote control UI (iPMI, DRAC, ILO….) . Don’t hesitate to add your contribution in the comment section. We share as much as we can, so should you…
Last note concerning firmware and driver combination. Check HCL, check HCL, check HCL….. Also, remember to pick a controller with at least a queue depth 256. If not you’ll most likely introduce a bottleneck.
Last note from an old post: How-To check the queue depth of storage controller.
More posts from ESX Virtualization:
- What is VMware vSAN Caching Tier?
- VMware vSAN 6.5 Licensing PDF
- VMware vSAN 6.6 Announced
- What is VMware vSAN Disk Group?
- What is VMware Hyper-Converged Infrastructure?