HP3000-L Archives

January 2000, Week 3

HP3000-L@RAVEN.UTC.EDU

Options: Use Monospaced Font
Show Text Part by Default
Show All Mail Headers

Message: [<< First] [< Prev] [Next >] [Last >>]
Topic: [<< First] [< Prev] [Next >] [Last >>]
Author: [<< First] [< Prev] [Next >] [Last >>]

Print Reply
Subject:
From:
Holger Wiemann <[log in to unmask]>
Reply To:
Holger Wiemann <[log in to unmask]>
Date:
Wed, 19 Jan 2000 00:17:27 +0100
Content-Type:
text/plain
Parts/Attachments:
text/plain (76 lines)
Hi all,

I very enjoyed this thread and cannot resist my experience regarding this
issue.

Holger


WARNING: Of cause all data are specific to the current test environment and
just should give you an impression what combinations of store/restore
parameters can lead to what elapsed time in this special environment; but
there should be no problem to adapt the results to other systems if you keep
in mind the change in CPU power and IO bus bandwidth.

Test environment:
-----------------
979-200, MPE/iX 5.5, pp C.55.07, 256 MB main memory, 2 x ST34371W FWSCSI
discs in system volume set, all backup devices connected to a single
dedicated host adapter card.
For the store to disc tests a Nike model 10 with 8 mechanics was used; it
was configured in an unsupported mode (all spindles are configured as
'Individual Disk Units') which means no disc has had any redundancy
configured within in the Nike; so, this Nike simply looks like a bunch of
slow FWSCSI discs to the MPE/iX box.

Test data:
----------
All store/restore tests were based on the same amount of data to work with:
3 copies of the group PUB.SYS in a special test account, spreaded over 2
ST34371W FWSCSI discs; the total amount of data was nearly 912 MB; the
restore tests were done by simply overwrite the existing data in the system
volume set.

All DDS tests were done with a HPC1537.

1. Using native-STORE (no TurboStore) -------------------------------------
-------------------------------------

1.a) best STORE results
-----------------------
DDS1: hw comp., STORE param. MAXTAPEBUF + INTER, 1108 KB/sec
DDS2: hw comp., STORE param. MAXTAPEBUF + INTER, 1103 KB/sec
DDS3: hw comp., STORE param. MAXTAPEBUF + INTER, 1807 KB/sec
DLT4000: hw comp., STORE param. MAXTAPEBUF,      2700 KB/sec
DLT7000: hw comp., STORE param. MAXTAPEBUF,      3958 KB/sec

1.b) best RESTORE results by tape created with
----------------------------------------------
DDS1: hw comp., STORE param. MAXTAPEBUF,            1054 KB/sec
DDS2: hw comp., STORE param. MAXTAPEBUF + INTER,    1052 KB/sect
DDS3: hw comp., STORE param. MAXTAPEBUF,            1303 KB/sec
DLT4000: hw comp., STORE param. MAXTAPEBUF,         2066 KB/sec
DLT7000: hw comp., STORE param. MAXTAPEBUF + INTER, 2723 KB/sec

2. using TURBOSTORE
-------------------
-------------------

2.a best STORE results
----------------------

DDS1: hw comp., MAXTAPEBUF + INTER + COMPRESS=LOW,    1126 KB/sec
DDS2: hw comp., MAXTAPEBUF         + COMPRESS=LOW,    1117 KB/sec
DDS3: hw comp., MAXTAPEBUF,                           1807 KB/sec
DLT4000: hw comp., MAXTAPEBUF         + COMPRESS=LOW, 3043 KB/sec
DLT7000: hw comp., MAXTAPEBUF         + COMPRESS=LOW, 4133 KB/sec
disk:              MAXTAPEBUF + INTER + COMPRESS=LOW, 4891 KB/sec

2.b best RESTORE results by tape created with
------------------------------------------------------
DDS1: hw comp., MAXTAPEBUF + INTER + COMPRESS=LOW,  1086 KB/sec
DDS2: hw comp., MAXTAPEBUF + INTER + COMPRESS=LOW,  1061 KB/sec
DDS3:           MAXTAPEBUF + INTER + COMPRESS=HIGH, 1355 KB/sec
DLT4000:        MAXTAPEBUF + INTER + COMPRESS=HIGH, 2371 KB/sec
DLT7000:        MAXTAPEBUF         + COMPRESS=HIGH, 3023 KB/sec

ATOM RSS1 RSS2