Topic: Cloud vs Dedicated Servers

[BlackBox]

It's been a few months with the new server, and thought I'd revisit this thread given current conditions....

Hard drive space seems quite limited, assuming the raid5 array of 600GB disks will be used for storage? I can't recall the formula we used but I want to say we took 2013's usage, accounted for 20% yearly growth and figured a 5 year hardware cycle.

For the CAD storage drive we now have this

Again, I'm not sure that I understand your meaning....

The RAID5 array proposed is more than triple what we're currently consuming on our existing file server (I work for a very small company; hiring me put the employee count into double-digits)... Even at your proposed 20% yearly growth, we would have to use this exact configuration (with no additional hot plug drives, NAS, etc.), and never archive, or delete any data created, for +/- 8 years to even reach capacity.

With the max extended warranty/support being 7 years, and the fact that we can expand RAID5 on the fly, both internally using the available hot plug drive slots, or externally via NAS, secondary server, etc. I feel very comfortable with our options for growth... I'm no expert at this, so kindly educate me where my logic may be wrong.

The storage capacity our now former 3rd party IT firm specified was horribly insufficient.

In order to spread the available storage capacity over the term of the warranty, we'd have to only consume 2.5% per quarter. At your recommendation, I estimated 5% quarterly (20% annually), which seemed much more realistic. However, I was not prepared for our actual consumption, which has been +/-17% in the past quarter alone.

For this reason I am exploring options now to increase our overall storage capacity, and backup infrastructure in kind.

When you get your new server, would you mind doing a disk benchmark on your RAID5 array? I went with a RAID 10 configuration for the performance and am curious how much faster it actually is.

I don't have the raw numbers in front of me (did the test a couple of weeks ago, finally!), and RAID 5 is terribly slower than the numbers you reported... To anyone considering the two, if your intent is performance focused, go with RAID 10... Something I may be doing in my legacy server upgrades if I stick with the 6 x 4TB SSHD drives (x2 servers) we're discussing in another thread.

Cheers

[ronjonp]

I don't have the raw numbers in front of me (did the test a couple of weeks ago, finally!), and RAID 5 is terribly slower than the numbers you reported... To anyone considering the two, if your intent is performance focused, go with RAID 10... Something I may be doing in my legacy server upgrades if I stick with the 6 x 4TB SSHD drives (x2 servers) we're discussing in another thread.

Cheers

I'd be interested in seeing those numbers when you get a chance. 

[BlackBox]

I don't have the raw numbers in front of me (did the test a couple of weeks ago, finally!), and RAID 5 is terribly slower than the numbers you reported... To anyone considering the two, if your intent is performance focused, go with RAID 10... Something I may be doing in my legacy server upgrades if I stick with the 6 x 4TB SSHD drives (x2 servers) we're discussing in another thread.

Cheers

I'd be interested in seeing those numbers when you get a chance. 

I'll start it again in a few... We you running from physical server, or within VM hosted on same?

[BlackBox]

When you get your new server, would you mind doing a disk benchmark on your RAID5 array? I went with a RAID 10 configuration for the performance and am curious how much faster it actually is. Also ... what raid controller did you choose?

I don't have the raw numbers in front of me (did the test a couple of weeks ago, finally!), and RAID 5 is terribly slower than the numbers you reported... To anyone considering the two, if your intent is performance focused, go with RAID 10... Something I may be doing in my legacy server upgrades if I stick with the 6 x 4TB SSHD drives (x2 servers) we're discussing in another thread.

I'd be interested in seeing those numbers when you get a chance. 

I'll start it again in a few... We you running from physical server, or within VM hosted on same?

I may have mistakenly left the default 1000MB selected in my previous test, whereas today I changed it to 100MB per your screenshot... This seems much, much closer to the numbers you reported (copied above)?  *confused*

Code: [Select]

  Test : 100 MB [E: 64.4% (1438.2/2233.4 GB)] (x5)
  Date : 2014/09/10 14:22:46
    OS : Windows Server 2012 R2 Server Standard (full installation) [6.3 Build 9600] (x64)

           Sequential Read :  1717.555 MB/s
          Sequential Write :  2066.245 MB/s
         Random Read 512KB :  1598.514 MB/s
        Random Write 512KB :  1815.414 MB/s
    Random Read 4KB (QD=1) :    64.653 MB/s [ 15784.4 IOPS]
   Random Write 4KB (QD=1) :    64.535 MB/s [ 15755.6 IOPS]
   Random Read 4KB (QD=32) :   292.127 MB/s [ 71320.2 IOPS]
  Random Write 4KB (QD=32) :   294.022 MB/s [ 71782.7 IOPS]


Code: [Select]

  Test : 1000 MB [E: 64.4% (1438.2/2233.4 GB)] (x5)
  Date : 2014/09/10 14:35:45
    OS : Windows Server 2012 R2 Server Standard (full installation) [6.3 Build 9600] (x64)

           Sequential Read :  1159.013 MB/s
          Sequential Write :   548.634 MB/s
         Random Read 512KB :   381.364 MB/s
        Random Write 512KB :   281.635 MB/s
    Random Read 4KB (QD=1) :     3.269 MB/s [   798.2 IOPS]
   Random Write 4KB (QD=1) :    11.761 MB/s [  2871.3 IOPS]
   Random Read 4KB (QD=32) :    17.248 MB/s [  4210.8 IOPS]
  Random Write 4KB (QD=32) :    27.238 MB/s [  6650.0 IOPS]


[BlackBox]

... And the boot/OS array; RAID 1:

Code: [Select]

  Test : 100 MB [C: 20.4% (56.8/278.4 GB)] (x5)
  Date : 2014/09/10 14:10:29
    OS : Windows Server 2012 R2 Server Standard (full installation) [6.3 Build 9600] (x64)

           Sequential Read :  1711.044 MB/s
          Sequential Write :  2066.930 MB/s
         Random Read 512KB :  1596.115 MB/s
        Random Write 512KB :  1773.494 MB/s
    Random Read 4KB (QD=1) :    63.834 MB/s [ 15584.4 IOPS]
   Random Write 4KB (QD=1) :    65.175 MB/s [ 15911.9 IOPS]
   Random Read 4KB (QD=32) :   295.628 MB/s [ 72174.7 IOPS]
  Random Write 4KB (QD=32) :   298.133 MB/s [ 72786.5 IOPS]


Code: [Select]

  Test : 1000 MB [C: 20.4% (56.8/278.4 GB)] (x5)
  Date : 2014/09/10 14:46:11
    OS : Windows Server 2012 R2 Server Standard (full installation) [6.3 Build 9600] (x64)

           Sequential Read :   362.578 MB/s
          Sequential Write :   182.583 MB/s
         Random Read 512KB :   436.543 MB/s
        Random Write 512KB :   345.761 MB/s
    Random Read 4KB (QD=1) :     4.418 MB/s [  1078.6 IOPS]
   Random Write 4KB (QD=1) :    16.877 MB/s [  4120.3 IOPS]
   Random Read 4KB (QD=32) :     6.247 MB/s [  1525.3 IOPS]
  Random Write 4KB (QD=32) :    16.948 MB/s [  4137.8 IOPS]


[ronjonp]

Interesting .. Just to be sure we're testing apples to apples, your Raid controller is a PERC H710p Adapter RAID Controller w 1GB NV Cache and your drives are 600GB 15K RPM SAS 6Gbps 3.5in Hot-plug Hard Drive ?

Test : 100 MB [E: 18.5% (515.5/2791.7 GB)] (x5)
Date : 2014/09/10 11:55:09
OS : Windows Server 2012 Server Standard (full installation) [6.2 Build 9200] (x64)

Sequential Read :  2200.909 MB/s
Sequential Write :  2413.114 MB/s
Random Read 512KB :  1972.638 MB/s
Random Write 512KB :  2042.930 MB/s
Random Read 4KB (QD=1) :    88.306 MB/s [ 21559.0 IOPS]
Random Write 4KB (QD=1) :    77.442 MB/s [ 18906.7 IOPS]
Random Read 4KB (QD=32) :   420.508 MB/s [102663.0 IOPS]
Random Write 4KB (QD=32) :   344.331 MB/s [ 84065.2 IOPS]

Test : 1000 MB [E: 18.5% (515.5/2791.7 GB)] (x5)
Date : 2014/09/10 12:02:33
OS : Windows Server 2012 Server Standard (full installation) [6.2 Build 9200] (x64)

Sequential Read :  1816.431 MB/s
Sequential Write :   949.224 MB/s
Random Read 512KB :   509.119 MB/s
Random Write 512KB :   527.959 MB/s
Random Read 4KB (QD=1) :     6.781 MB/s [  1655.4 IOPS]
Random Write 4KB (QD=1) :    33.320 MB/s [  8134.7 IOPS]
Random Read 4KB (QD=32) :    28.267 MB/s [  6901.1 IOPS]
Random Write 4KB (QD=32) :    58.317 MB/s [ 14237.7 IOPS]

 

[BlackBox]

Interesting .. Just to be sure we're testing apples to apples, your Raid controller is a PERC H710p Adapter RAID Controller w 1GB NV Cache and your drives are 600GB 15K RPM SAS 6Gbps 3.5in Hot-plug Hard Drive ?

Correct:

Also ... what raid controller did you choose?

The RAID controller that is included with our PER620 server spec from Dell is a PERC H710P.

I've added the 100MB Edit, Copy data to my earlier posts, and will do same with 1000MB test result data when done.

So... What does each number mean exactly, given I'm using SAS and not SSD?

Given these numbers, how does that 'difference' make you feel about your RAID 10 configuration, given the loss of +/- half your available capacity?

Cheers

[ronjonp]

I don't have the raw numbers in front of me (did the test a couple of weeks ago, finally!), and RAID 5 is terribly slower than the numbers you reported... To anyone considering the two, if your intent is performance focused, go with RAID 10... Something I may be doing in my legacy server upgrades if I stick with the 6 x 4TB SSHD drives (x2 servers) we're discussing in another thread.

Cheers

I'd be interested in seeing those numbers when you get a chance. 

I'll start it again in a few... We you running from physical server, or within VM hosted on same?

Physical server.

Given these numbers, how does that 'difference' make you feel about your RAID 10 configuration, given the loss of +/- half your available capacity?

The performance difference is not a much as advertised but still faster on all tests. If not constrained by disk space, I'd do it again .

- Sequential Read : 28.1% -
 - Sequential Write : 16.8% -
 - Random Read 512KB : 23.4% -
 - Random Write 512KB : 12.5% -
 - Random Read 4KB (QD=1) : 36.6% -
 - Random Write 4KB (QD=1) : 20.0% -
 - Random Read 4KB (QD=32) : 43.9% -
 - Random Write 4KB (QD=32) : 17.1% -

[BlackBox]

I've added the 100MB Edit, Copy data to my earlier posts, and will do same with 1000MB test result data when done.

Both 100MB, and 1000MB test results have been added to my earlier posts, for boot, and data partitions respectively.

[Edit] - Looks like I'm using 6.3 Build 9600 vs. your 6.2 Build 9200, as well.