Results 1 to 11 of 11
  1. #1
    Join Date
    Oct 2004
    Location
    Earth
    Posts
    419

    Anyone using LSI Cachecade SSD caching?

    Wondering what SSD raid configurations your using. Raid 1 or Raid 0, or what. Had it running in our SAN with Cachecade SSD drives set to Raid 0 (2 SSD drives) and we had one SSD drive fail this weekend. When it failed it took our RAID 10 volume offline. Would have thought if SSD caching failed it would degrade gracefully, but thats not what I'm seeing.

    Thanks!

  2. #2
    Join Date
    Aug 2007
    Location
    L.A., CA
    Posts
    3,654
    Its a RAID0, think of it as a single drive.
    EasyDCIM.com - DataCenter Infrastructure Management - HELLO DEDICATED SERVER & COLO PROVIDERS! - Reach Me: chris@easydcim.com
    Bandwidth Billing | Inventory & Asset Management | Server Control
    Order Forms | Reboots | IPMI Control | IP Management | Reverse&Forward DNS | Rack Management

  3. #3
    Join Date
    Oct 2004
    Location
    Earth
    Posts
    419
    Your saying you have a Cachecade SSD setup and your SSD drive (Do you have one or more?) is set as RAID 0?

    Was looking for real world experience and how you deal with the loss of the SSD drive taking out your SAN.

    Thanks!

  4. #4
    Join Date
    Jul 2007
    Location
    Ashburn, VA
    Posts
    1,312
    Quote Originally Posted by WebGuyz View Post
    Had it running in our SAN with Cachecade SSD drives set to Raid 0 (2 SSD drives) and we had one SSD drive fail this weekend.!
    Why in the world would you ever use RAID0 in a SAN? Should've seen it coming...

    Your SSD shouldn't have taken out your SAN. You should've been using RAID1 and swapped out the failed SSD.
    Preetam Jinka

    Isomerous - High performance web services for business and individuals.
    Bitcable Colocation, KVMs, cPanel hosting, Oracle expertise, and more.

  5. #5
    Join Date
    Jun 2006
    Location
    United Kingdom
    Posts
    1,763
    Interesting discussion actually. When we were shopping for storage solutions, I was told by Boston that STEC SSDs 'do not fail' and hence they were safe to be used in this sort of arrangement. I didn't believe them actually
    Have a server problem? Contact the experts

  6. #6
    Join Date
    Apr 2003
    Location
    Pittsburgh, Pa
    Posts
    198
    So the SSD disks in the Raid 0 were being used for caching in front of a separate Raid 10 array?

    If so, i wouldn't think it would take the whole thing down, shouldn't it just lose it ability to cache?
    TeraSwitch, Inc.
    Pittsburgh's Premium Provider of Cloud and Colocation Services.
    http://www.teraswitch.com

  7. #7
    Join Date
    Oct 2004
    Location
    Earth
    Posts
    419
    Thats what my thoughts were exactly. The RAID 10 drive was not accessable until the SSD caching was totally removed.

    Why even make available an option like RAID 0 for SSD caching if its going to make your Primary volume inaccessble when it fails.

    I will add that there were 2 choices for writing, Write Through and Write Back. I did choose Write Back. Not sure what would have happened if I chose Write Through. Thats why I was asking for real world experience with this product.

    In my mind it setting it to Write Back and have a failed SSD drive take the primary volume offline seems like a silly idea to even make it an option.

    I expect my vendors to protect me from myself.

  8. #8
    Join Date
    Mar 2002
    Location
    Philadelphia, PA
    Posts
    2,508
    My setup is a bit different, but accomplishes similar results (SSD buffering).

    RAID1: x2 Intel SSD's

    The intel SSD's are setup as Flashcache (kernel module) devices sitting in front of a RAID-10 SAS array.

    Using write-back mode (riskier), but we have redundant power including battery units on the RAID controllers.
    Linux junkie | steward.io

  9. #9
    Join Date
    Jul 2007
    Location
    Ashburn, VA
    Posts
    1,312
    Quote Originally Posted by WebGuyz View Post
    I will add that there were 2 choices for writing, Write Through and Write Back. I did choose Write Back. Not sure what would have happened if I chose Write Through. Thats why I was asking for real world experience with this product.

    In my mind it setting it to Write Back and have a failed SSD drive take the primary volume offline seems like a silly idea to even make it an option.

    I expect my vendors to protect me from myself.
    RAID0 writeback cache? You have no idea what you're doing.

    Quote Originally Posted by viGeek View Post
    Using write-back mode (riskier), but we have redundant power including battery units on the RAID controllers.
    SSDs are non-volatile. You don't need BBUs.
    Preetam Jinka

    Isomerous - High performance web services for business and individuals.
    Bitcable Colocation, KVMs, cPanel hosting, Oracle expertise, and more.

  10. #10
    Join Date
    Mar 2002
    Location
    Philadelphia, PA
    Posts
    2,508
    Quote Originally Posted by Bitcable View Post
    SSDs are non-volatile. You don't need BBUs.
    This is true. However this doesn't guarantee data is consistent. You're sending IO operations to a hardware raid controller with it's own subset of cache and IO scheduling. BBU provides safe guard on a very busy system that loses power with pending IO operations at the controller level. With a BBU we can ensure any pending operations at the controller level are satisfied.

    We use noop without nr_requests value of 0 (among other similar settings), so operations go more quickly to the controller.
    Linux junkie | steward.io

  11. #11
    Join Date
    Jul 2007
    Location
    Ashburn, VA
    Posts
    1,312
    Quote Originally Posted by viGeek View Post
    This is true. However this doesn't guarantee data is consistent. You're sending IO operations to a hardware raid controller with it's own subset of cache and IO scheduling. BBU provides safe guard on a very busy system that loses power with pending IO operations at the controller level. With a BBU we can ensure any pending operations at the controller level are satisfied.

    We use noop without nr_requests value of 0 (among other similar settings), so operations go more quickly to the controller.
    Didn't realize that. Very informative.
    Preetam Jinka

    Isomerous - High performance web services for business and individuals.
    Bitcable Colocation, KVMs, cPanel hosting, Oracle expertise, and more.

  12. Newsletters

    Subscribe Now & Get The WHT Quick Start Guide!

Similar Threads

  1. SSD VPS, Xeon i7, 1GB RAM, 10GB SSD only $8.99 USD/month
    By vpsdeploy in forum VPS Hosting Offers
    Replies: 9
    Last Post: 12-05-2011, 02:47 AM
  2. Replies: 0
    Last Post: 11-28-2011, 09:58 AM
  3. Replies: 0
    Last Post: 11-21-2011, 08:58 AM
  4. OCZ Synapse 32G/64G SSD caching drives
    By cwl@apaqdigital in forum Colocation and Data Centers
    Replies: 4
    Last Post: 09-23-2011, 09:51 AM

Related Posts from theWHIR.com

Posting Permissions

  • You may not post new threads
  • You may not post replies
  • You may not post attachments
  • You may not edit your posts
  •