Speed comparison: comparing disk #, raid, CPU, encryption

[a13antichrist]

I don't know if some people just choose to not read, or what.

Thanks P3R, entirely expected results, and completely missing the point.

An 8-disk RAID-5 has 7 writing spindles plus a write penalty.
An 8-disk RAID-10 has 4 writing spindles. Thus we'd expect an 8-disk R10 to have approx the same WRITE speed as a 5-disk R5, plus a little bonus because of the XOR penalty. And that's exactly what P3R's numbers find.

With same number of spindles (expect maybe very low, where XOR penalty is a higher %), R5 is going to be faster than R10 at WRITING because R10 only uses half of its spindles for write throughput.

SInce I don't expect thta the original participants are likely to be in a position to re-rest the same components, I won't ask for a proper survery, as should have been done int he first place.

So I'll quote myself from above instead.

Besides, the whole point is that this thread is positioned to "refute the claim that R10 is faster than R5", which is just as misleading as the original "myth" that "R10 is faster". My only aim here was to point out that it is not as simple as "which is faster, which is slower", but that the context, usage, and characteristics of the arrays all play a role and there is no universal "faster" at all. Write Performance is not everything, and neither is Read Performance, and so it is disingenuous to test only one and then claim "the speed advantage is a myth".

It's quite dishonest to do -only- write tests --or-- only read tests and then claim one or the other is 'faster'. The arrays will always cost different amounts and have different capacities. So pick and declare your working frame first, and state the parameters you're testing. Otherwise you're just a hack. Like the original article.

[OneCD]

* topic split from viewtopic.php?f=45&t=94588 *

I've split this into a new topic so that @P3R may respond.

[P3R]

@a13antichrist,
You already made this point in the original thread 5 years ago and it's valid. RAID 5/6 doesn't always have better performance than RAID 10 as perhaps was at times claimed in the thread. The background to what I wrote in that 2014 thread was years of built up frustration with a lot of "experts" insisting on that RAID 10 was ALWAYS faster (regardless of use case) and particularly when writing!

The testing initially done by qpio made it possible to show all those RAID 10 fans that they are wrong and that RAID 10 doesn't have superior sequential write speed with the same number of disks so we can now have much better discussions about what RAID level to use for what applications. You're very welcome to write a technical article on the different characteristics of different storage configurations or give advice on storage configuration questions here but can't you let that 2014 thread go now? I only refer to it when the occasional (now very rare) RAID 10 "expert" tell us that RAID 10 is always faster than RAID 5/6 and for that purpose it work.

90-95% of users in this forum are using their NASes for light home/SMB file sharing with almost exclusively sequential loads and very little random access. Most are also inexperienced with the technology and only seek straight and simple advice with as little technical mumbojumbo as possible. The top priorities for these users usually are:

1. Maximum storage capacity at minimal media cost.
2. Maximum storage performance (regardless of that the gigabit network is still often their bottleneck).
3. Best reliability with the least drive redundancy penalty.

If someone have a different use case and different priorities than the above they may get a different advice than RAID 5/6 from me.