iostat LTO-2

[verbose5]

We just migrated our NB4.5 environment to a new STK SL8500. I'm not very knowledgeable about Soalris performance monitoring, and am looking for some tips about checking on my tape drives with iostat. Here's some current data from one of my media servers (sorry this forum won't preserve tabbed whitespace):

bash # iostat -xtdc
extended device statistics tty cpu
device r/s w/s kr/s kw/s wait actv svc_t %w %b tin tout us sy wt id
fd0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 0 474 7 22 8 64
sd0 1.5 1.9 89.8 93.6 0.0 0.0 8.0 0 2
sd1 1.5 1.9 89.5 93.9 0.0 0.0 8.2 0 2
sd2 0.0 0.8 0.7 5.6 0.0 0.0 17.4 0 1
sd3 0.1 0.8 0.7 5.6 0.0 0.0 17.0 0 1
sd21 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd409 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd410 7.1 0.1 444.1 0.7 0.0 0.0 2.4 0 2
sd411 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
sd412 7.1 0.1 443.7 0.7 0.0 0.0 2.5 0 2
sd425 1.9 18.8 4.8 608.6 0.0 0.0 4.1 0 3
sd426 1.9 19.0 4.7 610.8 0.0 0.0 4.0 0 3
st15 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
st16 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
st31 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
st32 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0
st63 0.0 120.4 0.0 5313.6 0.0 0.1 0.9 1 11
st64 0.0 73.7 0.0 4418.5 0.0 0.1 1.2 0 9
st65 0.0 95.1 0.0 4139.4 0.0 0.1 1.1 0 10
st66 0.0 81.1 0.0 5089.9 0.0 0.1 1.3 0 10
st67 0.0 25.0 0.0 1570.6 0.0 0.0 1.4 0 3
nfs1 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0

The st* entries at the bottom are my LTO-2 tape drives. What kind of stats should I expect to see at normal activity levels, and what values should get me concerned about performance?

[verbose5]

Actually, only the bottom 5 are LTO2's. The top four are 8mm.

[verbose5]

These drives are shared among all my media servers using SSO. I noticed that I don't see the same stats on the 5 drives two media servers have in common if I run iostat on different media servers at the same time. Anyone know why this might be?

1st media server (5 LTO2 in bold):
wbpbkup01 / bash # iostat -xn | grep rmt
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 rmt/0
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 rmt/1
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 rmt/2
0.0 0.0 0.0 0.0 0.0 0.0 0.0 0.0 0 0 rmt/3
0.0 131.5 0.0 5587.9 0.0 0.1 0.0 0.9 1 11 rmt/8
2.3 69.5 144.6 4168.5 0.0 0.1 0.0 1.2 0 9 rmt/9
8.8 89.7 565.4 3905.3 0.0 0.1 0.1 1.0 1 10 rmt/10
0.0 76.7 0.0 4807.7 0.0 0.1 0.1 1.2 0 9 rmt/11
0.0 23.6 0.0 1483.3 0.0 0.0 0.0 1.3 0 3 rmt/12

Second media server (all 9 are LTO2, 5 are the same as above)
bash # iostat -xn | grep rmt
2.1 3.1 66.7 99.4 0.0 0.0 0.0 2.9 0 2 rmt/0
2.0 3.0 63.9 94.6 0.0 0.0 0.0 3.0 0 1 rmt/1
0.1 4.3 2.8 140.2 0.0 0.0 0.0 1.0 0 0 rmt/2
0.0 5.2 0.0 168.0 0.0 0.0 0.0 1.0 0 1 rmt/3
0.1 7.0 1.6 227.0 0.0 0.0 0.0 0.9 0 1 rmt/4
0.0 2.4 0.0 77.7 0.0 0.0 0.0 0.9 0 0 rmt/5
0.0 6.3 0.3 202.6 0.0 0.0 0.0 1.0 0 1 rmt/6
0.1 9.0 4.5 289.0 0.0 0.0 0.0 0.9 0 1 rmt/7
0.3 17.7 9.3 567.0 0.0 0.0 0.0 0.9 0 2 rmt/8
0.3 13.9 8.6 445.3 0.0 0.0 0.0 0.9 0 1 rmt/9
0.5 28.6 16.1 915.5 0.0 0.0 0.0 0.9 0 3 rmt/10

[Timothy_Sanko]

Depending on the client throughput you will see greatly differing throughput on your media servers.

I have found that a 4 way 4 GB rp5470 can outperform a 4 way Sun v880 with 8 GB ram. It is the HBA that makes the difference there.

The WIntel platforms seem to work well also. The rp5470 and the
4 way DL580 seem to run comparably. The real issue has been the clients.