Areca 1882, the new speed King?

[Anvil]

I was surprised to see the Areca 1882i being used in a "review" at tweaktown.

Not much details or performance is found in the "review" but it does feature the LSI dual core RoC.
(same as the 9265 is using)

Really looking forward to a review and some more info on the new series of controllers.

[Nizzen]

I would love to try it soon!

[Computurd]

yes im not sure what that review is...that is surely lacking some real info.

Very interested in getting one of these. Looking for the IX version though of course!
Anyone wanna buy a 1880 with 4gb cache?

[Tiltevros]

maybe it a engineering IR bios

I was surprised to see the Areca 1882i being used in a "review" at tweaktown.

Not much details or performance is found in the "review" but it does feature the LSI dual core RoC.
(same as the 9265 is using)

Really looking forward to a review and some more info on the new series of controllers.

[odditory]

Can't wait to see what a SAS2208 with proper firmware can do.

EDIT: and having read through that Tweaktown article it seems it was by someone that maybe doesn't know how to properly use or tune benchmark tools to fit a given test scenario. i can tell just by looking at the numbers- because I know the controller and drives are capable of way more. i'm not even sure what he was testing exactly or what the point was, the "Final Conclusion" makes it seem he isn't really sure since its all over the road. TBH the whole article is just a mess of arbitrary information.

If someone wanted to actually analyze the merits of the 1882 then the review would include a given set of SSD's and spinning disks tested on an 1880, 1882, LSI 9260 and LSI 9265; and include not just RAID0 but RAID5 and RAID6 performance to highlight differences in parity computing muscle between SAS2108/SAS2208 generations of RoC as well as any differences between Areca and LSI firmware running on the same RoC. Also, tune the benchmark tools to appropriate settings for both the SSD's and spinning disks so that you're shining a light on the capabilities of the cards rather than on the shortcomings of your test methods. Additionally, hghlight any performance diffs with Fastpath on LSI cards if Areca isn't offering a similar feature on the 1882. And lastly, baseline a single drive and then chart how the card scales multiple drives, and then comparing that to the scaling abilities of other cards in the group.

Of course all that would require some work and analysis, but a decent reviewer would have no problem doing it. My gut is telling me performance bump from 1880 -> 1882 is going to be underwhelming and average out to < 10% among all tests, despite the DC RoC of the SAS2208, but I'd love to be wrong.

[Computurd]

yes the real world gains are usually felt at high QD and parity raid sets.
With parity raid sets the dual core just shines. man its UBER


lets talk benching...that's what us wild eyed overclockers are looking for. Guarantee we will see some real results there with certain benches....its all about that DDR3 especially a IX version...*whistles*
Im betting they allow 6Gb DDR3 on this board

interesting testing ideas odditory, and you are spot on with your ideas. It would be very interesting with rebuild times in respect to parity sets, and parity set performance in general. From my experiences here that is where i noticed the largest differences.
now all we need is to find this reviewer with all this gear...lol i need an 1882 stat!

[INFRNL]

yes im not sure what that review is...that is surely lacking some real info.

Very interested in getting one of these. Looking for the IX version though of course!
Anyone wanna buy a 1880 with 4gb cache?

I'll buy it, but you won't like my offer . I think I need a good card on a poor mans budget. I currently have an SR-2 and I think I am very limited on SATA II or something is wrong

[Computurd]

happy to report that Areca is supplying a card for review Feel free to suggest tests and whatnot. I am willing to try anything

[alfaunits]

How about trying if it holds array as it should? Disconnet two of RAID6 drives before booting, boot, go into ARC BIOS, just confirm the array is Incomplete/Degraded, shut down, reattach the drive and figure how often it looses the array completely

[Wintel64]

happy to report that Areca is supplying a card for review Feel free to suggest tests and whatnot. I am willing to try anything

Support 8GB DDR3 Ram?

[Nizzen]

How about trying if it holds array as it should? Disconnet two of RAID6 drives before booting, boot, go into ARC BIOS, just confirm the array is Incomplete/Degraded, shut down, reattach the drive and figure how often it looses the array completely

Looks like only old Arecas have problems. Not many problems with Areca 1680 and 1880.

But you are the nr.1 Areca hater

[alfaunits]

The problems did occur on 1680 as well
So, my point still stands - and you have no say on Areca, as you never gave the results of the Areca tests I asked for several times. Once you obfuscated and didn't do the right test, the rest you ignored.

[Sandon]

How about trying if it holds array as it should? Disconnet two of RAID6 drives before booting, boot, go into ARC BIOS, just confirm the array is Incomplete/Degraded, shut down, reattach the drive and figure how often it looses the array completely

Never seen this happen on areca. Also it shouldn't be degraded if you boot up with drives missing after a power-cycle (the array should not auto-activate) so it shouldn't show any array during POST until you manually activate IE A normal bootup looks like:

http://box.houkouonchi.jp/areca_post1.png

And when even one drive is disconnected when the machine was powered down it looks like:

http://box.houkouonchi.jp/areca_post2.png

If you can't hook up the original disk back up you have to manually activate via:

http://box.houkouonchi.jp/areca_activate.png

Enabling the auto-activate incomplete raid set will not have this behavior but that is a crappy settings to enable (IMHO) as this can save you from degrading your array when moving to a new chassis or something and one of the drives doesn't have a good connection and IMHO is an awesome feature.

That being said just yesterday I dropped by the Areca U.S. headquarters/RMA (its 15 miles away from where I live) to drop off another 50 ARC-1222's that have started failing intermittently (which is now 170 in total I have RMA'd). The ARC-1222 does seem to have a reliability issue when used in an enterprise environment (running heavy disk I/O 24/7) after 1-1.5 years. None of their other cards appear to have this problem and I doubt most consumers where the array runs idle 95% of the time has this problem either.

I talked to Benjamin after giving him the RMA batch and he mentioned the only difference with the ARC-1882 is that it has the dual core ROC instead of single-core. Curious on what advantages that gives for parity raid6 writes as currently my ARC-1880x seems to max out at around 1GB/sec write in raid6.

I have never seen areca no longer detect the array at all after you activate the array or the disks are hooked back up and we have over 600 servers with areca controllers.

[alfaunits]

Never seen this happen on areca.

Go to 2cpu.com -> Forums -> Storage
You'll see a hell of a lot of it.
What it should look like is, unfortunately, not what it does look like when we had Areca issues If I only had issues with one controller, I'd scratch it as a faulty one. If I had issues with multiple Areca controllers on the same system, I'd scratch it on a faulty system, or incompatible system.
However, not only does it happen with multiple Areca controllers on multiple (completely unrelated in any component, even the Areca controller) systems, it also happens to other people. That tells me Areca doesn't hold disks as it should.
I'm done using Areca for over a year now, but I keep pointing people at the issues, because there were many and I doubt it changed with the latest line.

[Computurd]

Curious on what advantages that gives for parity raid6 writes

certainly not raid 6, but here are some raid 5 write numbers with the same ROC, but of course this could be different with arecas implementation as compared to LSI.

http://thessdreview.com/our-reviews/...s-ssd-testing/

[Sandon]

Go to 2cpu.com -> Forums -> Storage
You'll see a hell of a lot of it.
What it should look like is, unfortunately, not what it does look like when we had Areca issues If I only had issues with one controller, I'd scratch it as a faulty one.

The only time I have ever had issues with the array disappearing and/or showing multiple raid sets (all inaccessible) is when at one point the volume states became a *FAILED* state which would mean 3 disks failing on raid6 or 2 disks failing on raid5, etc...

Also with the newer firmware (not necesarily cards) the cards are much better about detecting this situation and automatically reviving the failed volume set. I actually couldn't get it to fault the array if I had the disks hooked back in as it kept reviving the failed array. I was purposely doing this as I was writing up a wiki article for areca raid recovery (if say the wrong disk was pulled) for the company I work for.

Also Areca is one of the easier raid controllers to recover the array by re-creating it with the no init/rescue option. I myself have recovered a raid5 array with no corruption when the wrong disk was pulled when it was degraded.

The thing is. I have used areca, 3ware, and LSI and I have seen far more arrays lost to quirks on 3ware/LSI than I have on areca (we are talking 100+ machines of each brand).

Biggest problem I have had with areca is the ARC-1222's as I mentioned.

[alfaunits]

The thing is. I have used areca, 3ware, and LSI and I have seen far more arrays lost to quirks on 3ware/LSI than I have on areca (we are talking 100+ machines of each brand).

The problem with that is that we don't see people reporting issues with LSI - so we an't just go imagining how much it fails. We do see failures with Areca (myself much more than with LSI).
This is pure what-you-see-is-all-you-can-base-your-opinion-on.
You could be right - but the big bulk of complaints at RAID controllers are on Areca. That's all I need to see to avoid them.

[Computurd]

[Sandon]

Bleh it does annoy me that they came out with this new version like a week after I bought my ARC-1880x.

[odditory]

Well the 1880 has been out for more than a year - how long should they have waited on you?

[Computurd]

The 1880 is still a very viable card, man it is great imo!
There are really only a few key differences...the dual core ROC being a key one. Definitely going to help with parity sets with SSDs.
Another big one is the DDR3....running at 1333 and 1gb on the 1882x. the 1880 has 512 mb of DDR2-800.

going to be a bit of a challenge to draw straight comparisons from the 1880 and the 1882, as the 1880 is such a cache whore, it caches everything! no real way to disable caching totally imo...and with the 1880IX having a 4gb stick that is a bit unbalanced.

[Wintel64]

The 1880 is still a very viable card, man it is great imo!
There are really only a few key differences...the dual core ROC being a key one. Definitely going to help with parity sets with SSDs.

What is parity sets with SSD?

[Computurd]

sorry guess iu should phrase that differently... a raid set with parity (ex. R5, R6 etc)
Parity requires more overhead on your card, so more "grunt" with your ROC helps big time with parity.

[Nizzen]

No numbers up yet? Is something wrong!?
Maybe I share numbers faster than you?... You better hurry

[alfaunits]

You owe us some basic numbers that we asked ages ago, I am not holding my breath for you to give us any useful benchmarks, but rather just marketing ones