In March, OCZ signed a definitive agreement to acquire Indilinx, the SSD controller company that pretty much dominated the value segment in 2009 before Intel got aggressive on pricing and SandForce took over the high-end. Indilinx was late on releasing its 6Gbps SSD controller prior to the acquisition and had all but fallen off the radar as most of its partners jumped ship to SandForce. OCZ scooped up the company with hopes of developing its own controllers and rising above the memory-manufacturer-turned-SSD-player crowd.

SandForce has obviously extended its reach since OCZ announced its intentions, partnering with many of OCZ's competitors in an attempt to expand its reach as well as lessen dependence on OCZ. As a customer turned competitor, SF like many of us knew that an OCZ SSD boasting a new Indilinx controller was in the works. Today we have the first details on that solution and it's called Octane.

Based on the Indilinx Everest controller, OCZ is touting incompressible data and sustained write performance as the two major strengths of Octane - obviously a jab SandForce and traditional SSD weaknesses. The controller is paired up with up to a 512MB DRAM cache, indicating that it is taking the more traditional route of storing both user data as well as page mapping tables in DRAM. There's no mention of power protection circuitry to flush the cache contents to NAND in the event of a power failure, but given the target market for these drives I don't suppose we'll see such a feature. 

The Octane will be available in two discrete versions: one with 6Gbps SATA support and one limited to 3Gbps (SATA 3.0 and 2.0, respectively). The 6Gbps version will ship with 25nm IMFT synchronous NAND, while the 3Gbps version will use 25nm IMFT async NAND similar to the Vertex 3/Agility 3 divide today. Sequential read/write speeds are competitive at 560/400 MBps, while 4KB random read/write speeds are quoted at 45K/25K. 

 

Capacities start at 128GB and will go all the way up to 1TB. OCZ is expecting pricing to fall somewhere within the $1.10 - $1.30 per GB range, which would put a 128GB drive at no more than $166 (in line with current Vertex 2 pricing) and a 1TB drive at just over $1300. Spare area is set at the standard ~7% you get from the GiB to GB conversion. No word on whether or not there's any additional NAND set aside for redundancy. 

The Everest controller features 8 NAND channels and up to 16-way interleaving. With 25nm NAND that means peak performance should be achievable with the 128GB model, with the larger drives just offering higher capacity and no additional performance. TRIM is of course supported. OCZ indicates support for both standard wear leveling and background garbage collection (presumably at idle time).

The first Octane drives will be available in the channel starting November 1st. I've already dispatched a huge list of questions to OCZ about the controller and the drive's architecture so I'll update this post as I get any new information. And of course, once we get a review sample I'll begin extensive testing.

POST A COMMENT

52 Comments

View All Comments

  • Paul Tarnowski - Thursday, October 20, 2011 - link

    While that would work, a better (and when power goes out cheaper) solution to the problem is using an UPS. Means that writes are finished regardless.

    Of course, it doesn't get around PSU failure. Hmm, I wonder if there's money in selling SATAPOWER-CAP-SATAPOWER connectors.
    Reply
  • DanNeely - Thursday, October 20, 2011 - link

    The consumer models of every other SDD with a ram cache have also dropped the battery/ultracap to cut costs. Given OCZ's habbit of making a zillion variants with each controller they use I'm mildly surprised they aren't launching a premium variant which adds backup power for a 100 or 200% marginal profit rate. Reply
  • josephjpeters - Thursday, October 20, 2011 - link

    Everest is a big deal. This will allow OCZ to offer next-gen SSD's at a price point no other company can match.

    The latency of this controller will certainly quiet down Fusion IO. They used to talk about IOPS until OCZ surpassed them and then focused on latency which it seems OCZ has surpassed them again.

    OCZ's challenge will be ensuring that this drive is reliable, but since the controller is owned by OCZ I doubt we'll see any issues similar to the "BSOD" debacle that occurred because they had to wait for sand force to implement their changes.
    Reply
  • sanguy - Thursday, October 20, 2011 - link

    OCZ is the same company that has the dubious honour of screwing customers on all 3 generations of it's SSD products.

    With them now in the drivers seat on the controller development I certainly don't view this as a positive - their business tactics will just continue but now the problem gets deeper with the rush to market.

    As for even mentioning OCZ going up against Fusion IO is a joke. Let me give you a hint : one has well established enterprise level support and qualification, one doesn't have any real support and zero qualifications (and don't say they have a support forum as if you ever voice a problem they rapidly lock and hide the thread as they don't want the transparency of all the issues).
    Reply
  • Sunburn74 - Thursday, October 20, 2011 - link

    That lack of transparency though can be applied to many many enterprise hardware providers when it comes to known issues (dell and their leaky capacitors, apple and pretty much anything that goes wrong with apple parts, etfc).

    The world is a "what have you done for me lately' world and if OCZ can deliver, businesses will do what they have always done: make a business decision.
    Reply
  • semo - Thursday, October 20, 2011 - link

    Lookup the 25nm transition fiasco. They sold drives with less capacity than was advertised on the packaging and they didn't even issue a recall. OCZ left it up to the consumers to find the fault themselves and complain to OCZ. Reply
  • retrospooty - Thursday, October 20, 2011 - link

    All SSD's have not only the formatted capacity misconception that HDD's have, but also the reserved area on top of that. Or was there an issue that went beyond even that? I googled it and found nothing, so I doubt its that bad. Reply
  • semo - Friday, October 21, 2011 - link

    It was that bad. If you bought an OCZ Vertex 2 in Oct 2010 and then another one in Jan 2011 then chances were that you wouldn’t be able to RAID them because the latter one would be GBs(!) smaller.

    "With twice the density per NAND die in these early 25nm drives, __usable__ capacity was also reduced when OCZ made the switch with Vertex 2."

    http://www.anandtech.com/show/4256/the-ocz-vertex-...

    http://forums.anandtech.com/showthread.php?t=21433...

    As I've been saying from the start, OCZ have tried very hard to burry this issue and unfortunately, it seems that they have succeeded.

    They can use all the terminology and excuses they want (RAISE and whatever) but the fact was, OCZ was misleading customers and NEVER issued a recall (the only correct way of dealing with the problem).
    Reply
  • JonnyDough - Monday, October 24, 2011 - link

    You can raid two drives of unequal size. Your raid volume would simply be reduced to the smaller of two drives... Reply
  • jjcrandall - Thursday, October 20, 2011 - link

    Wasn't this a big + of the SandForce drives? The nDurance stuff looks really interesting

    Bench this Anand!
    Reply

Log in

Don't have an account? Sign up now