Account team got me this response. RAID 5 is rough from a risk standpoint, i.e. one failed disk kills it all, not to mention RAID 10 has performance gains far beyond RAID 5 (and I don't need 6TB of space with RAID 5 let alone 3.6TB with RAID 10, but regardless, here it is.
Subject: RE: RAID Configuration for M5
Specific to appliances and TRCs...those are targeting voice/video admins with low skillset and low opinion level on HW BOM.
We also have to balance a number of (contradictory?) tradeoffs
1. "support" / how many different RAID configs across the set of TRCs (affects TAC MTTR et. al.)
2. Validating / maintaining performance for new SW release or new HW generation (including "qualification lag")
3. Usable GB space (runtime vs. for TRCs that are also appliances ... size of factory preload)
4. Latency and IOPS performance (big deal as so many of our apps are real-time and IO-driven)
5. Fault tolerance (with assumption of redundant VMs running redundant HW)
6. Ease of failed disk replacement and other customer/partner procedures (remember...voice/video persona not DC persona which is different)
7. Net HW cost vs. Customer willingness to pay
We actually used RAID10 in the 1st-generation Small TRC (C200 M2 TRC#1) with cheap SATA disks.
But in M3 and beyond, RAID5 became a better option in terms of ticking off more boxes above.
"Server/storage/DC persona admin" types dislike RAID5 and will debate what is a "better" or "best" configuration. Usually they prefer additional hot standby disks, RAID6 or RAID10.
If doing specs-based-infra, use whatever RAID you want as long as it meets our min specs around HCL, latency, GB, IOPS.
But we don't test on RAID6/10 nor do we provide prescriptive BOMs, so we don't make the same promise level as we do on our TRCs.
Should not be a surprise, but freaks out risk-averse customers.
If doing prescribed-infra, i.e. a TRC or a BE6K/7K appliance, then assumption is you are not opinionated on the HW BOM, and we've standardized on RAID5 for above reasons.
