The negative side of a controller

[Tiltevros]

This is a Thread that is missing from this forum.

We are using the Best equipment out there. Sometimes the best that money can buy. We all speak about the Pros of our controllers or for our SSD or mechanic HDD.

But for our own expirience what is the Cons of them?? What problems do we see on them?

[SoulsCollective]

Increased bootup times. That's about it. You lose a PCI-E slot, but really, most modern boards have slots and to spare these days, and I don't consider that a con - it's just to be expected from using a product that sits in a PCI-E slot.

[Tiltevros]

its not only bootup times... Because controllers are used mainly in Servers with uptime 95%+ so u dont need to boot.
I would say for the LSI 9211 that is very very difficult to setit up properly for Optimum performance.. Factory settings Suxs.. very hot cpu when u use it for over 15mins. and and and.....

[SteveRo]

Too many cables if using a lot of drives.
power to all the drives gets to be problematic - too many connectors gets to be a reliability problem
advantage - pcie direct connect drives (ioextreme, ...)

[Tiltevros]

pcie direct connect drives (ioextreme, ...)

cost of buying

[AceNZ]

I think one big downside of RAID controllers is that they often don't result in the performance improvements that many people seem to expect. For example, light-use, desktop-type applications (with low QD) will rarely realize an increase in performance from RAID, and it might in fact end up being slower than using a built-in controller such as the ICH10R. Another way to say the same thing is that benchmarks often don't reflect real-life use, particularly for single-user desktop machines.

[F@32]

I think one big downside of RAID controllers is that they often don't result in the performance improvements that many people seem to expect. For example, light-use, desktop-type applications (with low QD) will rarely realize an increase in performance from RAID, and it might in fact end up being slower than using a built-in controller such as the ICH10R. Another way to say the same thing is that benchmarks often don't reflect real-life use, particularly for single-user desktop machines.

Oh , no, you didn't! Napalm called SWAT team to your place of residence Yeah, my problem is price/performance with these things. Would be sweet to have 2/4Gig RAID card with 4R0 x-25E though... Ideally direct PCIe card.

[Computurd]

@acenz...aot of that is setup dependent though, and usage dependent. there are gains they are just very few

my main gripe is the loading overhead time while card initializes. they really need to fix that it suxors

[AceNZ]

@acenz...aot of that is setup dependent though, and usage dependent. there are gains they are just very few

That's exactly my point.

I've talked with many people who have installed an expensive R0 setup, using all default / factory settings, and see no perf improvement for their day-to-day workload. For your average desktop user who just does web, email and occasional gaming, a controller board is overkill compared to just using the ICH10R.

[Computurd]

very true... i wouldnt suggest someone with that usage pattern to get a 800 dollar card and 8 ssd...but possibly a 180 HBA and 2 or 3 x-25's maybe....

[SteveRo]

I think one big downside of RAID controllers is that they often don't result in the performance improvements that many people seem to expect. For example, light-use, desktop-type applications (with low QD) will rarely realize an increase in performance from RAID, and it might in fact end up being slower than using a built-in controller such as the ICH10R. Another way to say the same thing is that benchmarks often don't reflect real-life use, particularly for single-user desktop machines.

If you can live with the capacity constraint - the "snapiest" low-qdepth performance for a desktop machine is probably had with an Acard 9010. At low qdepth, looks like the acard has something like a factor of 2 better 4k random read performance than the fastest SSDs (Vertex LE, Crucial C300, ...). The ioxtreme might also provide this kind of performance but not yet bootable.

[AceNZ]

If you can live with the capacity constraint - the "snapiest" low-qdepth performance for a desktop machine is probably had with an Acard 9010. At low qdepth, looks like the acard has something like a factor of 2 better 4k random read performance than the fastest SSDs (Vertex LE, Crucial C300, ...). The ioxtreme might also provide this kind of performance but not yet bootable.

Yes, the Acard definitely has a place.

My issue with them isn't so much the capacity / cost (although that's definitely an issue), as the fact that they're volatile; constant backups would be a big pain in the ass in my environment. Most desktop users I work with are lucky to backup once every few months. Going from that to once every few hours is a non-starter.

[[XC] Oj101]

What causes dedicated controllers to be so much slower than integrated controllers?

[lowfat]

Too many cables if using a lot of drives.
power to all the drives gets to be problematic - too many connectors gets to be a reliability problem

This. I use to have a 16 HDD array but got rid of it as I was sick of the cables.

[SteveRo]

What causes dedicated controllers to be so much slower than integrated controllers?

Probably multiple contributors - longer data path, additional processing, un-optimized drivers and/or f/w, marginal compatibility with either mobo or drives ...

[[XC] Oj101]

Sorry, I meant for initialisation. I've seen RAID cards add 40s to boot times while they start up

[One_Hertz]

Well it sort of has its own internal OS to load up.

The manufacturers also tend to have little care of the bootup times because the majority of their clients use RAID cards for servers which never reboot anyway so it is not a factor for most users...