Can Netbackup 6.5.5 restore a tape blocksize of 256 with default setting?

[ckfsymantec]

hi

after deleting NET_BUFFER_SZ  262144, SIZE_DATA_BUFFERS 262144 and NUMBER_DATA_BUFFERS 16  and run a backup, i am still able to restore from a tape previously written in 256kb, i was expecting this restore to fail but it ran successfully. anyone has experieced this?

this test is run on dell tl2000 2xlto5, restore to san storage v7000

this is the media-server and client itself  (local)

i have only 2 months experience of Netbackup 6.5.5

im testing a block size change to maximize LTO4 performance

--------------------------------------------------------------------------------------------------------------------

concurrent restore lto4 64kb & lto4 256kb on 256kb setting

lto4 64kb - 2h 33m
23:07:50.500 [4516.4992] <2> drivename_close: Called for file IBM.ULT3580-TD4.001
23:07:50.500 [4516.4992] <2> tpunmount: NOP: MEDIA_DONE 0 0 0 0002L5 4000002 180 {100CB5E9-E642-40CF-AFA4-FDE743F331EF}
23:07:50.500 [4516.4992] <4> mpx_read_backup: successfully restored 1 of 1 requests, read total of 524165537 Kbytes at 57383.943 Kbytes/sec

lto4 256kb - 1h 15m
21:50:53.296 [5636.5752] <2> drivename_close: Called for file IBM.ULT3580-TD4.000
21:50:53.296 [5636.5752] <2> tpunmount: NOP: MEDIA_DONE 0 0 0 0004L5 4000004 180 {C6F3D0FA-02C8-4264-818E-B0C8AE331700}
21:50:53.296 [5636.5752] <4> mpx_read_backup: successfully restored 1 of 1 requests, read total of 524165504 Kbytes at 117044.885 Kbytes/sec

deleted file NET_BUFFER_SZ  262144, SIZE_DATA_BUFFERS 262144 and NUMBER_DATA_BUFFERS 16

netbackup change blocksize write setting back to 64kb (trigger dummy backup to confirm change)
09:11:16.406 [4708.4892] <2> io_init: using 65536 data buffer size
09:11:16.406 [4708.4892] <2> io_init: CINDEX 0, sched Kbytes for monitoring = 60000
09:11:16.406 [4708.4892] <2> io_set_recvbuf: setting receive network buffer to 263168 bytes
09:11:16.406 [4708.4892] <2> io_init: using 30 data buffers
09:11:16.406 [4708.4892] <2> io_init: child delay = 10, parent delay = 15 (milliseconds)
09:11:16.406 [4708.4892] <2> create_shared_memory: shm_size = 1966808, buffer address = 0x1380000, buf control = 0x1560000, ready ptr = 0x15602d0

logs during restore begin

12:47:04.687 [4728.4724] <4> report_client: VBRC 2 4728 0 veritas-master_1308033231 13 Backuplto4 0 *NULL* 0 1 1
12:47:04.687 [4728.4724] <2> mpx_setup_restore_shm: using 30 data buffers, buffer size is 262144
12:47:04.687 [4728.4724] <2> mpx_setup_restore_shm: child delay = 10, parent delay = 15 (milliseconds)
12:47:04.687 [4728.4724] <2> mpx_setup_restore_shm: shm_size = 7866840, buffer address = 0x1b50000, buf control = 0x22d0000, ready ptr = 0x22d02d0, res_cntl = 0x22d02d8

restore lto4 256kb on 64kb netbackup setting - 1h 15m
14:02:29.953 [4728.4724] <2> drivename_close: Called for file IBM.ULT3580-TD4.000
14:02:29.953 [4728.4724] <2> tpunmount: NOP: MEDIA_DONE 0 0 0 0004L5 4000004 180 {51E420E5-6D4F-4624-980B-F4DCE126331F}
14:02:29.953 [4728.4724] <4> mpx_read_backup: successfully restored 1 of 1 requests, read total of 524165504 Kbytes at 117052.674 Kbytes/sec

[ckfsymantec]

these are logs of 00041L5 backup

netbackup change blocksize write setting to 256kb
14:33:55.218 [2312.4652] <2> io_init: using 262144 data buffer size
14:33:55.218 [2312.4652] <2> io_init: CINDEX 0, sched Kbytes for monitoring = 60000
14:33:55.218 [2312.4652] <2> io_set_recvbuf: setting receive network buffer to 262144 bytes
14:33:55.218 [2312.4652] <2> io_init: using 16 data buffers
14:33:55.218 [2312.4652] <2> io_init: child delay = 10, parent delay = 15 (milliseconds)
14:33:55.218 [2312.4652] <2> create_shared_memory: shm_size = 4194696, buffer address = 0x1380000, buf control = 0x1780000, ready ptr = 0x1780180

backup to lto4 256kb on 256kb netbackup setting - 1h 15m
15:49:32.093 [2312.4652] <2> send_MDS_msg: MEDIADB 1 62 0004L5 4000004 *NULL* 6 1308033231 1308033231 1309242831 0 524165504 1 1 1 1 0 0 1024 0 2047526 0
15:49:32.109 [2312.4652] <4> write_backup: successfully wrote backup id veritas-master_1308033231, copy 1, fragment 1, 524165504 Kbytes at 116853.701 Kbytes/sec

[mph999]

The blocksize is used when the backup is written, not read.

The blocksize that the backup was written with is stored in the image db, for the backup.  Therefore, deleting the blocksize settings will not affect the restore, as you have proven.

Martin

[Marianne]

Have a look at this table in the Planning and Performance Guide  http://www.symantec.com/docs/TECH62317
 

[ckfsymantec]

hi Martin & Marianne, thanks for the answers.

im referring to the guide 307083.pdf, (the table u show is on page 125) and some recommended beginner setting read here in the forum when im doing these tests, hence the 16x256 256 settings, so i take it as the question has been answered as my previous understanding was the setting of blocksize is for both  read and write.

also, lets say if i change to another server due to hardware problem and after restoring catalog, will i still be able to retrieve this information from Report - unhide column to show the hidden block size information of backup tapes ?

this test server has been erased for other use, i should have taken a look at it first before the erase, instead of asking you this question, anyway i will look for it in my next test of Netbackup.

based on the test results, am i getting full performance from my hardware?
i havent tested backup/restore thru LAN client and other common setting like multiplexing of two drive d:\ e:\, multiple streaming, the combo, or many more which i havent discover.

performance of LTO tapes (native speed)
LTO3 max 80 mb/s = 4.8 GB/m = 288 GB/h
LTO4 max 120 mb/s = 7.2 GB/m = 432 GB/h
LTO5 max 140 mb/s = = 8.4 GB/m = 504.00 GB/h

[mph999]

backup to lto4 256kb on 256kb netbackup setting - 1h 15m
15:49:32.093 [2312.4652] <2> send_MDS_msg: MEDIADB 1 62 0004L5 4000004 *NULL* 6 1308033231 1308033231 1309242831 0 524165504 1 1 1 1 0 0 1024 0 2047526 0
15:49:32.109 [2312.4652] <4> write_backup: successfully wrote backup id veritas-master_1308033231, copy 1, fragment 1, 524165504 Kbytes at 116853.701 Kbytes/sec

524165504 Kbytes at 116853.701 Kbytes/sec

I make that 114 Mbytes/ sec (approx) - looks good to me.

You are unlike to hit a constant max read/ write of the drive.  From my experience of speeds I see, and thing over >100Mbytes/ sec is good for LTO4 or 5.  It is very often the case, that the infrastructure (disks, internal bus etc..) cannot keep up with the drives.

Marianne may think differently however ;o)   we will see ...

Martin