-1

Does anyone know what is the maximum physical hard drive space a single host of ESXi 6.0 can support? And no, this is not that dumb because Microsoft only supports a maximum storage of 64 TB.

I am planning to purchase 24 QTY 4 TB Seagate SAS drives from Amazon for my NORCO RFC-4224U case, with 24 hard drive bays, and an ARECA 1883IX-24 RAID card. If ESXi does not support 96 TB of total physical hard drives directly hosted in this box, what other operating systems or software (Openfiler, FREENAS?) can support this file size?

Thanks!

  • Can you clarify your RAID configuration? – Craig Watson Jun 4 '16 at 8:22
1

If you intend to present each disk individually through to ESXi (JBOD style) then one limit you could hit quite quickly would be the max VMFS5 volumes per host limit of 256. Of course most people would RAID these disks which would obviously get around this limit but VSAN prefers this JBOD style so could potentially get to this limit quite easily.

  • That helps. I actually want to individually present it to ESXi as unique HDDs. Any ideas if this will work (meaning I present the 24 4TB HDDs as unique 4 TB, therefore there will be 24 datastores with 4 tb in size)? – etnemo Jun 4 '16 at 13:16
  • Yes, that'll 100% work - you know VMWare Big Data Extensions? their Hadoop setup? well I once did a PoC of that and used a pair of HPE DL380's each with about 50 x 900GB disks and 'JBOD'ed' them - worked a treat, bit of a pain formatting and naming that many disks but you only do it once. – Chopper3 Jun 4 '16 at 18:29
2

As per this document on ESXi 6.0 Configuration Maximums, a single VMFS5 partition has a maximum size of 64TB.

There is no limit to the number of devices on the underlying storage as VMware will not be accessing individual disks directly - this will be done by your RAID card.

If you will be using RAID0 across all of your 96TB of storage (which is a very bad idea) then you will need to create multiple datastores on the array.

My personal recommendation is to use RAID10 (twelve two-disk RAID1 arrays, with data striped over them) - this gives a performance increase with the benefit of redundancy. A RAID10 array can cope with many disk failures, as long as you don't lose both disks in a single pair.

If you're concentrating more on capacity than performance, then two 12-disk RAID5 arrays will give a much higher final capacity, but at the cost of performance and resilience (each RAID5 will be able to cope with one disk failure, but with a much higher rebuild time).

For comparison:

  • RAID0 = 96TB
  • 2 x RAID5 = 88TB (each array is 44TB)
  • RAID10 = 48TB (12 x 4TB 2-disk RAID1)
  • 1
    Please don't R5 these disks - I can bore with why but R5 is essentially dead. – Chopper3 Jun 4 '16 at 11:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.