in
Support Center

Issues during demo testing

Last post 07-22-2008 5:27 AM by zapf. 3 replies.
Page 1 of 1 (4 items)
Sort Posts: Previous Next
  • 06-11-2008 6:09 AM

    Issues during demo testing

    Hi everyone,

    I have been trialling the IT edition and have some issues and questions.   I would appreciate if anyone could provide help and feedback on this issue.

    Throughputs:
    I have been testing laptop and server imaging using USB 2.0, eSATA and network destinations.  I have been getting truely surprising throughputs and as a result I thought I would appreciate your experiences.

    USB 2.0 has a theoretical bus throughput of 60MB/s but people typically report 25MB/s.  With Shadowprotect I am seeing anything from 4MB/s to 20MB/s
    eSATA has a theoretical bus throughput of 375MB/s but practically people report 50-75 MB/s.   With Shadowprotect I am seeing nothing over 30MB/s
    Network transfers - I only tried this once and get 9MB/s as I describe below.

    From my reading the main limit to any transfer is the disk.  The Seagate Momentus 2.5" drive has a sustained transfer rate of 75MB/s so I'm a little lost.

    Now I have several questions:
     - is this what everyone else experiences ?  What do you use ?
     - is the throughput reported by Shadowprotect a current value, or average for the entire time of the backup?

    Basic Workstation Imaging

    I tried to image a basic Windows XP Pro workstation, and it failed "unable to lock volume" yet I specifically made sure NOT to lock the volume.  No apps were running.    I thought the whole point of this being based on VSS is that it runs on the fly without the need for a reboot.

    Secondly, I rebooted the machine and used the Vista based boot image on the CD and it takes FOREVER to start.  What gives !!! 

    Thirdly, I rebooted the machine and used the Legacy based boot image.  Trying to connect to a mapped drive took me nearly an hour of trial and error.  I eventually found I had to set the NIC to 100 Half Duplex to get a connection.  Once connected I got an absolutely horrible 9MB /s throughput !!!!

     I then tried to restore the 3GB image onto a 10GB disk and found out about ShadowProtects nasty little secret - it won't resize the image and restore onto a smaller drive !!!  Unbelievable.  I saw the post on the forums where everyone has been complaining, and I understand why.

    Lotus Domino Support

    I've read that older versions of Domino are not VSS complient.  Does anyone know if the latest v8 apps are supported ?

    Windows 2008 support

    I've got clients deploying Windows 2008.  Is ShadowProtectIT fully Windows 2008 complient ?


    Thanks
    Mark

     

     

  • 06-13-2008 8:29 AM In reply to

    Re: Issues during demo testing

    Through-put issues are always a little tricky.   Generally the thoughput rates that drive manufactures claim are read throughputs.   If they claim write thoughputs that are equivelent, then they are generally relying on caching to help with that.  If you nullify the cache, the writes are about half the speed of reads.  Most disks to a read after write to verify.

    ShadowProtect does both reading and writing with compression in between.  This means that there are 3 areas of possible bottlenecks. 

    The read:  Usually not the bottleneck on backup.  Local disk reads are usually the fasted IO.

    Compression:  Very CPU specific.  We recommend standard compression.  It gives the best throughput.  Issues that affect the throughput are CPU speed and number of CPU's.  The other is the data.  If data is highly compressible, then the amount of data to be written is less and therefore if the write is the bottleneck, then fewer writes will mean better througput.  Newer processors like the core 2 is very fast at compression.  Early P4's and ealier CPU's are not so fast.

    The write:  This is where most people find out how slow their disks, networks, NAS devices and RAID devices really are.   A large sustained write always competely removes the benefit of caching.  In fact, ShadowProtect tries the open all files as non-cached so that the cache managers are not burdened by our large sustained writes.

    The other strange fact that IT Edition has, is that IT Edition relies on Microsoft's built in snapshot technology volsnap.  When comparing backups with volsnap and StorageCraft's stcvsm, it is slower.   Not only is backing up volsnap slower, it also has to backup more data because of volsnaps snapshot tables.

    I don't know the domino answer but IT Edition is 2008 compliant.

    Hope this helps a bit, there is a tool called filegen that is attached to the following forum post that can be helpful for troubeshooting throughput.  See: 

    http://forum.storagecraft.com/Community/forums/p/754/3605.aspx#3605

    Scott
    STC

  • 06-13-2008 8:41 AM In reply to

    Re: Issues during demo testing

    In response to your other issues:

    If XP failes to backup hot, then there is an issue with the VSS framework on that system.  You can check the state of vss by opening a command prompt and typeing the following:

    "vssadmin list providers"  - This should show you the status of the snapshot technology on your system.

    "vssadmin list writers"  - This will give you the status of the vss aware applications on your system.

    If the status returned by both of these is not successful, then I would suggest trouble shooting this by searching for VSS issues on this forum and also with Google.

    Not sure what your NIC speed is supposed to be, if it is 100 MBit, then 9MB/s is correct and this is your bottleneck.  If it is GBit, then the and exact driver match for your network card is not available.  You might try adding them dynmically into the Vista based environment.

    Shrink is the number one requested feature, so stay tuned for an up coming solution.

    Scott
    STC

     

     

  • 07-22-2008 5:27 AM In reply to

    • zapf
    • Top 500 Contributor
    • Joined on 02-16-2008
    • Posts 4

    Re: Issues during demo testing

    Hello Mark,

      I use ShadowProtect day in day out.  Thought I will give you some ideas re thru put.

     

    Throughputs:

    USB 2.0 - 27~30Mbps (2.5" Samsung drive)

    100Nic - 9~10Mbps (Bottlenet is network)

    1000Nic - 70~80Mbps (Bottlenet is network, transfering between 2 system with 4 x SATA running RAID10 in each system)

    Internal 4 SATA disk RAID10 to Internal SATA 4 disk RAID1 - 90~100Mbps

     

    So from my experience, there is no limitation by the software I have yet found. It is all in the hardware you are running on.

    Cheers

    Zapf

     

     

Page 1 of 1 (4 items)
(c) StorageCraft Technology Corporation 2008