Jonathan P writes:
I'm sure you're aware that LaCie has announced a couple of new SSD RAID drives, one of which, the 2TB Bolt3, is said to approach read speeds of 2,800 MB/s, which I assume will outrun the new MacBook's internal SSD.
I take it you're not a fan of LaCie (not sure), but I've been using their portable Backup Plus Fast and 4TB Rugged RAID drives for several years without a single hitch. I also have a few 4TB LaCie desktop Porsche design hard drives that have also been reliable. I realize those are too slow and too small for the work you do, and not really configurable to your liking, but I'm just a hobbyist. If the Bolt3 didn't run $2,000, I might consider it. At the moment, I shoot with a LUMIX GH4 and downsample to 1080p for delivery to YouTube. I really appreciate all the work you do.
MPG: As I understand it, only the LaCie Bolt is an SSD, the others are hard-drive based. My confusion stems from the most obtusely written press release I can ever recall reading. Why companies write such stuff is baffling. The point is to communicate, and the press release is a failure that left me puzzling through it.
I have little experience with LaCie products, because I can’t review what I can’t easily obtain. My time is always under pressure, and if a company doesn’t want to respond quickly or seek me out, then I generally cannot afford to pursue dead ends or to waste my time justifying my request according to some generic rules-based process at Company X. Recommending products that I have not personally tested is a bad protocol for everyone, so I avoid that. Accordingly, I cannot recommend the LaCie Bolt, but by the same token I am not dis-recommending it. The foregoing is an open offer for the folks at LaCie to contact me and send me a LaCie Bolt and their other products for testing.
Apple is claiming 3GB/sec for the internal SSD of the 2016 MacBook Pro, which is faster than the LaCie Bolt claim. I say “claim” because performance is much more complex than one number. The 3000 MB/sec = 2.86 MiB/sec or 2800MB/sec = 2.67 MiB/sec figures are sustained throughput—best case with very large transfers.
Take a look at actual behavior in the graph below, which shows throughput (MiB/sec) versus transfer size*. The Apple and LaCie offerings are going to behave according to this same pattern. The quoted figures are best-case for large transfers only. Yet in the real world, typically 99% of transfers are ~1MB or smaller!
Accordingly, factors like latency and how many requests per second can be handled figure into actual real world performance.
As I understand it, latency is inherently higher over Thunderbolt 3 than the direct-connect PCIe bus of the internal SSD. Moreover I expect the Apple internal SSD to be a very high performance design. Therefore I expect the Apple internal SSD to outperform anything and everything.
What the Apple internal SSD is likely to be less good at is industrial-grade longevity (over provisioning and such). So peak speed alone is a poor metric by itself: for professional use a device must meet or exceed and sustain a critical performance metric, and be designed for 3 or 5 years of taking a beating. Thus a slower offering may well be a far superior choice for some applications.
* MiB = 1024 * 1024 and MB = 1000 * 1000
Example of SSD performance behavior below.