New Vertex 2 drives aim for servers

SandForce solid-state drive controllers have already made quite an impact on the desktop. Now, OCZ has taken SandForce’s best controller, the SF-1500, and used it to fashion a couple of enterprise-class SSDs aimed at server applications.

The new Vertex 2 Pro and Vertex 2 EX both have 2.5″ form factors with 300MB/s Serial ATA interfaces. The former uses “enterprise-grade” multi-level-cell flash memory and is rated for top transfer rates of 285MB/s for reads and 275MB/s when writing. OCZ claims the drive can sustain 50,000 IOPS for random writes. Meanwhile, the Vertex 2 EX uses pricier but more durable single-level-cell flash memory; OCZ says it has “ten times the program/erase (P/E) cycles with the same enhanced performance as the Pro edition.” The company also plans a Serial Attached SCSI version of the Vertex 2 EX.

OCZ’s announcement talks of even more enterprise-specific goodness:

Both series have a built-in supercapacitor that acts as a temporary power backup in the event of sudden power loss, enabling the drive to complete its task ensuring no data loss. Vertex 2 Pro/EX also provides superior ECC protection and unique technology that mimics RAID-like protection and recovery from flash block failures on a single drive.
Furthermore, Vertex 2 Pro/EX increases the longevity of NAND flash memory with intelligent block management and wear-leveling to extend endurance and overall reliability of the device. This feature automatically stores data in an AES-128 encrypted format to prevent data extraction to maintain security of confidential information, a necessity in an enterprise environment.

The company says it’s offering both products in 50GB, 100GB, and 200GB variants, all with three-year warranty coverage and a mean time between failure rating of 10 million hours. (That’s about a millennium, if my math is right.) The Vertex 2 Pro is also launching in a 400GB iteration.

Comments closed
    • sleeprae
    • 9 years ago

    OCZ? Enterprise? Hah! I have around 50 of the 120GB Vertex drives deployed to end users. I’ve had problems with at least 12 of them. Most of the time, they can be “fixed” by downflashing and reflashing firmware, but 6 of them have had to go back for RMA replacement. 3 of those failures were dead drives that resulted in lost data. For a long time, you had to go to their support forums to get recent “official” firmware releases that I continued to hope, in vain, would remedy the reliability problems. The RMA process is terrible from a corporate perspective–it’s clearly a company that targets nothing other than the enthusiast. There’s little that I’ve done professionally that I regret more than buying those drives.

    Once the Intel Gen2 drives dropped, I switched to them exclusively. Now with around 40 drives deployed, I haven’t had a single one give a single ounce of trouble.

    Now, that being said, I really like the idea of the supercap to ensure that data held in write cache is destaged to NAND. This may in fact be an excellent product–but OCZ has much to do to convince me they’re ready for the enterprise.

      • Ethyriel
      • 9 years ago

      Luckily others will follow with SF1500 based drives, which requires the use of a supercap. I know I’d feel a whole lot better getting a drive from Corsair, and the strong suspicion is that Seagate drives will be Sandforce based as well.

      • shank15217
      • 9 years ago

      A Vertex 2 isn’t Vertex 1, your reliability issues don’t apply to these models. Considering SF firmwares come directly from SF and are tweaked slightly by the drive makers most SF drives will have the same reliability. Some may use cheaper flash than others but the wear level feature and a generous 28% over provisioning pretty much insures that even the cheapest drive would last about 5 years or so, good enough for just about any IT scenarios. I wouldn’t buy the 7% over provisioned models, its not worth the headache. I am running 4 vertex limited drives as cache for a open storage device, probably doing a million IOPs a day, no problems.

        • sleeprae
        • 9 years ago

        I’m not trying to argue that my reliability experience with the Vertex should be applied to the Vertex 2. I didn’t say it directly, but I do absolutely refuse to use any first-generation SSD controller design after my experience with the OCZ Indilinx units. I don’t believe that any of my failures were with NAND–I strongly believe that each and every one of them was due to a controller failure of some sort. The SF is very impressive technology wise, but my experience with Indilinx has definitely made me gunshy with first generation designs. In contrast, I’ve had excellent success with Intel’s second generation units.

        The actual product was only part of the argument, however. OCZ the company needs to significantly revamp their entire support structure, including firmware release and RMA process, before I’d consider them enterprise ready. The processes make sense for the enthusiast focus of the company, but are among the worst I’ve seen from an enterprise vendor perspective.

    • dreamer77dd
    • 9 years ago

    Can you add these ssd with other devices like the RevoDrive

    §[<http://hothardware.com/News/Computex-2010-OCZs-Revo-Drive–Affordable-PCIe-SSD-Performance/<]§ will it scale to get more performance?

      • CasbahBoy
      • 9 years ago

      It depends on what you mean. I can say with near absolute certainty that you can have both a RevoDrive and OCZ SSD in the same host – but I can’t imagine how you would be able to make them both part of a single RAID array. My guess is that the RevoDrive board is simply a PCI-E connected SATA controller with a flash controller chip and flash memory mounted onto the same PCB. You can’t really RAID that when traditional SSDs are just flash controller chips and flash memory using whatever SATA controller is integrated onto the motherboard – if the RevoDrive had an SATA port on it however, then things would be a little different.

      • Firestarter
      • 9 years ago

      Only with some kind of OS based RAID configuration, because the Revodrive is actually RAID on a card, with 2 Sandforce based SSDs tacked on. So you’d need an OS that can tie RAID arrays and one or more drives together.

      I wonder when we’ll see the first SSD optimized RAID cards.

      • Firestarter
      • 9 years ago

      Also, you posted that because of this: §[<http://www.youtube.com/watch?v=6nzJxiRf5Kk<]§ right?

        • indeego
        • 9 years ago
          • Firestarter
          • 9 years ago

          I don’t see your point

            • dreamer77dd
            • 9 years ago

            just want everything to work together to make something go even more faster. make everything gpu cpu revodrive work as one for extreme performance. I dont know if it is possible but if a gpu can make my programs work faster well working together with a cpu why not.

            • Firestarter
            • 9 years ago

            Actually, there are quite a few programs that already do that. It’s never that easy as just tying stuff together though. CUDA is a nice example of how companies like Nvidia are trying to make their products useful beyond their nice: §[<http://en.wikipedia.org/wiki/CUDA<]§

            • indeego
            • 9 years ago

            I guess I just question linking to a blurb and video that give /[

            • Firestarter
            • 9 years ago

            Yeah same here, don’t know why he’d post a blurb at another site when TR has a news post about it. But I guess more coverage of OCZ’s Jessica is always better 😀

    • CasbahBoy
    • 9 years ago

    Is it known if Intel, Adaptec, or any other SSD/RAID controller manufacturer are putting work toward TRIM support in RAID arrays? Forgive me if this has already been answered or a reason already given for it not being worth the effort.

      • bcronce
      • 9 years ago

      Intel has, not sure about the others. Screw hardware RAID, all about ZFS software RAID :p

    • ew
    • 9 years ago

    *[<"ten times the program/erase (P/E) cycles with the same enhanced performance as the Pro edition."<]* Shouldn't this be reflected in a longer warranty for the EX version?

      • anotherengineer
      • 9 years ago

      Ya no doubt, 5yrs would be nice. I wonder how much the 100GB EX is going to cost?

      Probably over a grand 🙁

      • OneArmedScissor
      • 9 years ago

      No, because the warranty has nothing to do with memory’s write capacity, and everything to do with the fact that the manufacturers aren’t confident enough in their own products.

      They’re very quick to beat you over the head with marketing bullet points about how much more durable SSDs are, and yet, multi-platter, high speed HDDs, at much lower prices, have longer warranties. It’s quite cute.

    • Shining Arcanine
    • 9 years ago

    It seems that this should eliminate the need for journalling file systems, so the performance will be slightly higher than people could get with normal SSDs, because the journals can go.

      • ew
      • 9 years ago

      Journals help make large file system operations atomic. You’d still want a journal even with this drive.

      • bcronce
      • 9 years ago

      Journaling not only makes re-coving faster, but add other features that you want, like Transactions. Journals are also important for MFT data. It’s possible for a sudden power loss to hose your file system if the right thing is being changed and no journal.

        • Shining Arcanine
        • 9 years ago

        How do you get sudden power loss with a super-capacitor onboard?

    • wibeasley
    • 9 years ago

    How important is that supercapacitor? Aren’t all enterprise servers connected to a UPS?

      • cygnus1
      • 9 years ago

      UPS doesn’t help if your PSUs fail.

        • wibeasley
        • 9 years ago

        Won’t most of these have redundant PSUs? Are these likely to go into SANs, or just nodes?

          • CasbahBoy
          • 9 years ago

          I think the rule is generally “more redundancy is better,” especially when additional redundancy is cheap (the cost of an expensive capacitor to OCZ, and maybe an additional 15 dollars to the end customer.)

            • wibeasley
            • 9 years ago

            Makes sense.

            • shank15217
            • 9 years ago

            3 years is standard storage warranty, for a while drive makers brought that down to 1 year. There are other factors that effect warranty, such as cost to the manufacturers for a replacement, SSDs cost a shitload and their prices are going down at a pretty slow pace, don’t expect long warranties.

    • l33t-g4m3r
    • 9 years ago

    I initially read that as, “New Vortex 2 Driver”, and I was like whoa, now I have to try this.
    Then I looked closer, and was disappointed.

      • albundy
      • 9 years ago

      hahaha, i had to take another look at the subject line too. i guess its just the two of us that miss Aureal these days.

    • indeego
    • 9 years ago

    Now these just need 2+ years in the market before Enterprise will consider themg{<.<}g

      • 5150
      • 9 years ago

      …and a helluva lot better reliability record than their workstation brethren.

        • indeego
        • 9 years ago

        We’ve had one issue from all our SSD’s so far, and that was with an Intel 80G G2.
        About every 3-4 days we get:
        “The system failed to flush data to the transaction log. Corruption may occur.”

        Almost like the drive is being pulled or going offline randomly.

        Drive passes all diag tests, changed/reseated cables/ports, so we are just vigilant about backing it up at this pointg{<.<}g

          • CasbahBoy
          • 9 years ago

          If you’re still reading this thread indeego, how many are you or your company running?

    • StellaArtois
    • 9 years ago

    How long till the Intel Gen 3 stuff shows up? Wish they’d hurry up and get those out the door, might shake up the current pricing again. Still too pricey for me.

      • OneArmedScissor
      • 9 years ago

      The end of the year. It’s only been a few months since the 34nm drives were actually stocked in stores.

      I wouldn’t expect miracles of those, anyways. The 34nm drives being in such low supply caused prices to stagnate, as no one needed to implement their planned price cuts to compete when their prices turned out to be hyper inflated at retail. Considering how fast they’re moving to the next node, it will probably be the same thing all over again when those get here.

Pin It on Pinterest

Share This