The ZCS(512) & BCS(520) Saga continues

Jon Toigo asked me to respond to a post on his blog .

Hi Jon:

I find PQ65’s comments very interesting, but I don’t understand the reasoning behind his comments. If BCS (520) is better then ZCS (512) and NetApp uses ZCS on their Nearstore products, doesn’t this mean that customers D/R and backup drives are more vulnerable to corruption? It would seem so, because NetApp recommends Dual Parity on ZCS drives. This seems to leave customers relying on a less resilient technology for their backups. How much less reliable are ZCS systems than BCS systems, and is it worth the risk? That is what my customers and I are trying to find out.

Can NetApp provide reliable, repeatable and verifiable data to show their consumers that the Nearstore products that use ZCS drives are as reliable as NetApp’s products that use BCS technology? Does NetApp keep its financial data on ZCS drives or BCS drives? Why not allow consumers to judge their cost to risk ratio by disclosing test results that can be duplicated and verified?

Clearly there are performance and cost advantages to each technology and drive type. NetApp could easily disclose accurate and repeatable test results, consumers could then make informed and economical decisions on where to store their D/R and back up data. And everyone would be a winner.

This entry was posted in Uncategorized. Bookmark the permalink.

1 Response to

  1. July says:

    Here’s my understanding. Network Appliance filers started with standard 512 sector disk drives. Parity inconsistancy, or data corruption while not common, were bothersome, perhaps having something to do with RAID4/WAFL.
    So they came up with BCS/520 sector disk drives and the
    performance was vastly improved.
    ——-Possble Interpretation—-
    Sounds like they solved a potential data corruption issue by going to BCS
    (i.e., greater reliability). Also, perhaps better
    system performance. Therefore it might be said they were claiming that with
    the new formatting, the filer and data have experienced greater performance
    and reliability.

    However, that’s not the same as saying the drive itself is somehow improved.
    The drive is still the drive and is as reliable as magnetic media can
    be–formatting doesn’t change that.

    So by going back to ZCS on the SATA drives, perhaps the data and filer might
    be less reliable and have less performance, but the actual drive is still
    just the same old magnetic medium. Formatting doesn’t change a drive’s
    reliability and I don’t see where anyone says it does. The SYSTEM on the
    other hand, might or might not be affected. I have seen claims about the
    system/filer/data performance and reliability.

    Everything I read indicates that
    Netapp said the system/filer/data reliability was improved with BCS. I don’t see in
    there where it says anything about the drives themselves being better or
    worse. Now, you can infer that Netapp might have believed something like
    that about the drives, and infer that that might have influenced their
    decision making when choosing the ZCS for the SATA, but I haven’t seen anything
    written anywhere to that effect. Are there documents that say anything
    like “We chose the SATA drives because they are cheaper and we formatted
    them ZCS because…..” ?

    NetApp might be infered to believe that Nearstore backup isn’t as important as
    production data and they were willing to sacrifice the data/filer reliability for
    price. Assuming it were true, and there’s no proof (just surmise), Mike disagrees, saying the backup is as important
    as the active data. That’s an agree to disagree type of argument. Taking that a step further, though, now that NetApp is also using these ZCS SATA drives
    on the production boxes, what does that mean? Now they are talking
    production data–is it less important now? Or have they done something in the newer versions of OS to accommodate for it?

    If they haven’t done the latter, it makes you wonder; Mike might be right about that. But
    to make the point Mike might want to separate NetApp’s claims about the
    “system/data/filer” reliability and performance from any claims (if they
    exist) about the “drive” itself. I think that’s where people are having
    trouble with the argument. A drive is a drive and everyone knows it and I
    haven’t seen Netapp claiming otherwise.

Comments are closed.