Last Updated: Tuesday, April 18, 2000 11:02 PM
JK
The HP-3000 series of computers dates back to 1972 with the first model, the HP3000/CX which ran the MultiProgramming Executive, MPE-C operating system with 64K bytes of core memory (addressed as 32K 16-bit words). Subsequent models (Series II, Series III) added additional addressing capabilities to include additional "banks" of 64Kb memory and each model brought out a new operating system (MPE II, MPE III). All models were and have continued to be object-code compatible (at least forward-compatible) with few exceptions (most notably the change in floating-point format from the CX).
Microprocessor-based 3000's were introduced with the models 30 and 33 concurrent with a changeover of the prevalent I/O architecture to HP-IB. Later models expanded the product line to the models 40, 44, and 64 and correspondingly the MPE IV operating system. Shortly thereafter, "disc cacheing" came into being and the controversial "MPE V" operating system. MPE V was eventually delivered in three flavors:
* MPE V/P for existing 3x, 4x, and 6x models which was marketed as an upgrade (40-42, 44-48, and 64-68) but involved no hardware (if the memory expansion was omitted). This provided disc cacheing software.
* MPE V/E which *did* require hardware upgrade (to support extended CST tables, among other things) providing disc cache and extended tables.
* MPE-V R for the series II/III to give MPE-V functionality without the disc cacheing option (they were considered too memory constrained). This was the last release of MPE for these systems (now unsupported).
HP experimented with a 32-bit version of the HP3000 CISC-based processor code named "Vision" but it never came to market; instead they opted for the RISC-based "Spectrum" machines where the current HP3000/9xx systems have their roots. Although radically different in architecture and in their instruction sets, the "Spectrum" series maintained an imbedded "Compatibility Mode" facility which (largely but not completely) kept the forward-compatibility object code concept alive. The underlying operating system (initially called HPE internally) became MPE/XL.
JK
HP3000 systems prior to "Spectrum" are now commonly referred to as the "Classic" systems, the 9xx series were initially called "Spectrum" systems (although currently "XL" or more recently "iX" is more commonly used). MPE/XL continued until the beginning of the integration of Posix support into the operating system, which was then abruptly renamed to MPE/iX as of Release 4.0 (?) although the only supported (push) release with realistic Posix features will be the MPE/iX 5.0 release.
I am going to update my classic HP system to an MPE/iX system. What special considerations should I bear in mind and/or expect?
JK
For the most part, you can simply use the provided utilities to migrate your accounting structure, restore your files, and run compatibility mode just fine. There are, of course, certain exceptions to this rule. The following list is *not* comprehensive (caveat emptor) but include:
* Anything using Privileged Mode (in-house, 3rd party, or HP-supplied).
* Anything calling COMMAND to invoke non-MPE/iX supported commands such as PTAPE, any cacheing-related commands, private volumes, and other specifics.
* Any code looking for PARM= or INFO= by tracing stack markers (depends on release of MPE and MPE/iX).
* Anything dependent on system logfiles.
* Anything dependent on LDEVs being < 255 or device type/subtype info.
When you start converting your code to "native mode" there are other considerations, such as alignment issues (persistant problem with any converted applications), switch stubs (if you have RL/SL linkages), and floating-point data conversion issues (the "Classics" used an HP format, while "Spectrums" use the IEEE format).
However, don't dismay, things are much better now (time-proven) than they used to be, but it is rarely if ever a "recompile and go" scenario.
JK
Yes - most all HP and third-party software must have a license upgrade to run on the new system, plus maintenance charges. You might consider upgrading to a (large) production CPU with no utilities and a small system for program development, compilers, debuggers, etc to save . Some third-parties do not charge (Robelle, 3k Associates, Adager, (?)).
JK
It will run in compatibility mode with switch stubs (note that you need the Pascal/iX compiler to compile your stubs or else work out something with your SE). There is also the SPLash! compiler from Software Research Northwest which can generate native mode code from SPL source (check the vendor contacts section).
Note that you *can* continue SPL applications (the SPL compiler and programs (non-PM at least) work fine under MPE/iX). And the switch stub requirement of Pascal/iX is only if you plan to use SWAT, the "Switch Assist Tool" to generate stubs; stubs can be written in other languages with considerably more difficulty.
SS
1. Beyond RISC! Copies available from Software Research Northwest (publisher), and Lund Performance Solutions. (Yes, it's dated, but still useful.)
2. IMAGE Handbook (? the one edited by Marguerite Russell (sp?))... it's still available (somewhere in the HP3000-L was a posting with her phone number)
3x = any Classic HP3000 in the 30 series (30, 33, 37)
30 = code named Koala
37 = code named Mighty Mouse
Micro/3000 = series 37 with microcode on chip, other enhancements
4x = (40,42,44,48)
40 = code named Cub
44 = code named Grizzly
5x = (52,58)
64A/B = code named Gemini
68A/B = 64 with two I/O bays
70A/B = 64/68 with bigger cache (highest end classic 3000)
(*B models had newer/improved power supplies)
9xx = any HP3000 in the 900 series.
922 = code named Silverfox (Low)
932 = code named Silverfox (High)
925 = code named Firefox
930 = code named Indigo
935 = code named Topgun
9x7 = Nova series boxes
948 = code named Silver Bullet -L
958 = code named Silver Bullet -H
949 = code named Shogun
950 = code named Cheetah
955 = code named Jaguar (also: PN10)
960 = code named Cougar (also: PN10C)
980 = code named Panther
918,928, = (sucessor of 9x7 boxes) code named "Wright Brothers"
968,978
99x = Emerald series boxes (multi processor options)
ER
BTW, I hear somewhere that the very first Spectrum was implemented in a S/48 cabinet.
RLE
You are correct. The Grizzly processor on an S/48 was microcoded to implement the Spectrum instruction set. It was called the LESS machine. I don't know what the acronym meant, but remember that less is more.
PAC
Some information on differences between processors:
S/40 - S/42 --- added disc caching, implemented in software.
S/44 - S/48 /
S/42 - S/52 --- Memory ctlr update to support 2 and 4 MB memory
S/48 - S/58 / boards. Also, updated CPU to support various
system table expansions and slight processor speed
enhancement.
S/64 - S/68 --- Added support of a 3rd I/O bay.
S/68 - S/70 --- Increased CPU (not disc) Cache for a rather
significant increase in CPU speed.
There were a couple of other things that differentiated the S/64A and S/68A from the S/64B, S/68B, and S/70. Most notable were a different SSDP (System Status and Display Panel) and a totally different power supply and distribution group.
JK
(As I remember) the offerings at the time were 40, 44, and 64. The 40 and 44 were essentially the same horsepower and had interchangeable boards, but the 40 was a square-ish box with no terminal connections (other than console); the ADCCs ran out to little metal boxes or your ATP came to a mini-cabinet. The 44 was the "desk" looking thing, the 64 was the "chest freezer".
MPE V started the cache business and had V/P initially (cache only), V/E (new processor and firmware for CST tricks), and later V/R (for the obsolete Series III and below). V/P was purely software, but you had to "upgrade" from 40-42, 44-48, or 64-68 to get it; consequently if you got the upgrade early (before the firmware was released) and deleted the memory board, all you got was a new FOS/SUBSYS tape and a new nameplate (for a considerable chunk of money). I know, I got two of them :-) Of course you later got the firmware (and V/E).
The next processor speed-up changed 42-52, 48-58 (for sure) and I think also 68-70. I've had a 40/42/44/48/52/58, dunno about the rest :-)
CB
Yes, 68-70 added CPU/microcode cacheing. We went through a few of those upgrades ourselves.
Relative HP 3000 System Performance HP 3000 System Relative Performance Series 30, 33 0.5 Series III 0.6 Series 37, 37XE 0.6 Micro 3000 RX 1.3 Micro 3000 LX, GX 1.3 Series 39, 40, 44 1.0 Series 42, 48 1.3 Series 42XP, 52, 58 1.7 Series 64 3.2 Series 68 4.0 Series 70 4.4 Series 920 1.9 Series 922 3.2 Series 932 5.0 Series 948 10.7 Series 958 13.3 Series 925 2.9 Series 935 5.9 Series 949 11.7 Series 950 6.5 Series 955 10.0 Series 960 14.7 Series 980/100 22.0 Series 980/200 37.0 Series 980/300 49.0 Series 980/400 59.0 Series 917LX 10.0 Series 937 10.0 Series 947 10.0 Series 957 16.0 Series 967 20.0 Series 977SX 26.0 Series 987 32.0 Series 987/150 45.0 Series 987/200 60.0 Series 918 10.0 Series 928 14.0 Series 968 21.0 Series 978 Series 987/150 25.0 Series 939SK 28.0 Series 959KS/100 35.0 Series 987/150 45.0 Series 987/200 60.0 Series 959KS/200 62.0 Series 988 39.0 Series 959KS/300 86.0 Series 959KS/400 110.0 Corporate Business System 990 28.0 Corporate Business System 992/100 35.0 Corporate Business System 992/200 60.0 Corporate Business System 992/300 81.0 Corporate Business System 992/400 100.0 Corporate Business System 991 33.0 Corporate Business System 995/100 42.0 Corporate Business System 995/200 71.0 Corporate Business System 995/300 96.0 Corporate Business System 995/400 118.0 Corporate Business System 995/500 139.0 Corporate Business System 995/600 160.0 Corporate Business System 995/700 180.0 Corporate Business System 995/800 200.0
FMC
...a couple of months ago I discovered that HP have a hardware recycling department, which US callers may reach at 800-535-7933*. My understanding is that they can arrange for pickup and environmentally correct disposal of HP hardware.
*Note; the new phone # is reported to be 888 485-1849.
DC
Well, it's not a straight forward excercise, it will take (depending on the size of your system), several hours, and most of that time is downtime for your users.
Here is a (long, detailed) plan to follow:
Step 0. Validate NMMGR config
-------------------------------
:NMMGR
Select the following items from the function key options.
Open Config
Utility
Go To Validate
Validate DTS/LINK
Ensure that no errors are seen in the validate screen. Then press function
key 8, until you are back to the MPE prompt. If NMMGR went into sysgen
to cross validate then skip the following and go to step 1.
:SYSGEN
sysgen> io
io> hold
io> exit
sysgen> keep
keeping to group CONFIG.SYS
Purge old configuration (yes/no)? Y
sysgen> exit
Again ensure that sysgen did not give any cross validation errors.
Step 1. Create good CSLT without config of 2 old drives
---------------------------------------------------------
:sysgen
sysgen> io
** now to delete the drives from the config, lets assume ldevs 8 and 9.
io> dd 8 - delete ldev 8 from config
io> dd 9 - delete ldev 9 from config
io> hold
io> exit
sysgen> keep
keeping to group CONFIG.SYS
Purge old configuration (yes/no)? Y
sysgen> tape
** you will now be required to load a scratch tape, this will be your new
** Customised System Load Tape (CSLT). Reply to the tape prompt.
sysgen> exit
** We should now verify that the CSLT is a valid boot tape
:run checkslt.mpexl.telesup
Which level? 1
** Load the CSLT and reply to the prompt. Checkslt will validate the tape.
** Do not proceed with the next step unless the there are no errors.
Step 2. Create good full backup.
----------------------------------
We need to ensure that all files are backed up, as we will be doing an
INSTALL later and will need to restore all files from this backup set.
- Log all users and jobs off the system
:LIMIT 0,0
:ABORTJOB #S...
:ABORTJOB #J...
- If you have any networking software, ensure that the network is shut down
:NSCONTROL ABORT
:NETCONTROL STOP
:RUN NMMAINT.PUB.SYS,SHUTDOWN
- Now for the backup....
:FILE T;DEV=TAPE
:STORE @.@.@;*T;SHOW;DIRECTORY
** The DIRECTORY option is necessary! Don't leave it off.
** If on MPE/iX release C.45.00 (4.5) then use STORE /;*T;SHOW;DIRECTORY
** as STORE @.@.@ will not back up all files on this release.
** Use SHOWME command to determine your version of MPE/iX.
- Once backup has completed, we need to verify it.
:VSTORE *T;@.@.@;SHOW=OFFLINE;DIRECTORY
** The VSTORE will create a spoolfile, which should be examined for any
** errors. Do not proceed unless there are no errors (or you are prepared
** to lose the files that are in error state).
** You may wish to create a second full backup at this point, as during an
** INSTALL, all data is wiped from your disk drives, so you are relying
** totally on having a valid full backup.
Step 3. Shut down the system and INSTALL from CSLT
---------------------------------------------------
Now we are ready for the INSTALL. No need for a clean shutdown, so....
Enter (Ctrl) B on the console.
You should now be presented with a CM> prompt. The startup sequence may vary
from below, depending upon what model system you are using, but essentially,
you must boot from the alternate path, which should be your tape drive.
** Load the CSLT (created in Step 1) on your tape drive now.
CM> TC
Confirm restarting system? Y
Interrupt within 10 seconds? Y
Boot from primary path? N
Boot from alternate boot path? Y
Interact with IPL? Y
Now is a good time to physically disconnect the disk drives that you no
longer need. Just power down the disk drives and disconnect the cable
connecting them to the system. It is recommended that a qualified HP CE
perform this task, but it will be a chargeable excercise. Please contact
your local CE if you have any questions about this.
Your system should now have booted from tape and will be at the ISL> prompt.
We will now be doing an INSTALL. Remember, there is no turning back after
INSTALL is entered, it will load your system config from tape, all user data
on disk will be lost. You MUST have a valid full backup at this point!
ISL> INSTALL
Your system config and operating system will now be loaded from the CSLT,
once completed, you will once again be asked to choose a boot path. This time
we need to boot from the primary boot path.
Boot from primary path? Y
Interact with IPL? Y
ISL> START NORECOVERY
Your system should now boot and log you on as OPERATOR.SYS.
Log on as MANAGER.SYS, there will be no passwords, nor other accounts, or any
user files at this stage.
Step 4. Recreating operating environment.
------------------------------------------
:STARTSPOOL LP
:STREAMS 10
We need to first add our disk drives to the system volume set.
:DSTAT ALL
The disk drives will be in one of the following states:
MASTER: Contains volume set definition and data.
MEMBER: Contains data.
LONER: Volume with a valid label, but no master mounted.
SCRATCH: Volume with a valid label, but marked with no valid data.
UNKNOWN: Volume with no valid label.
In our case, ldev 1 will be a MASTER, all other drives will probably be LONER.
Any drives that are LONER, SCRATCH or UNKNOWN need to be added to the system
volume set - MPEXL_SYSTEM_VOLUME_SET.
To add them to MPEXL_SYSTEM_VOLUME_SET, the drives must first be in SCRATCH
or UNKNOWN state. So first of all, we need to SCRATCH any LONER drives.
:volutil
volutil> scratchvol x - x is ldev of volume to scratch
** Do this command for every drive that is in LONER state.
volutil> :DSTAT ALL
Re-check state of disk drives. There now should be none in LONER state. Now,
for every drive in SCRATCH or UNKNOWN state, we need to add it to the system
volume set...
volutil> newvol MPEXL_SYSTEM_VOLUME_SET:MEMBERx x 100 100
Add new volume on ldev x? Y
Do this command for every drive that is in SCRATCH or UNKNOWN state, replacing
'x' with the ldev number of that drive.
volutil> exit
:DSTAT ALL
Now all drives will be in either MASTER or MEMBER state (ldev 1 will be the
only MASTER).
Now you will need to restore your data from the full backup set created
in step 2.
:FILE T;DEV=TAPE
:RESTORE *T;@.@.@;SHOW=OFFLINE;KEEP;DIRECTORY;OLDDATE
Load the first tape in your backup set and reply to the prompt. The restore
will create a spoolfile, which you can then examine to ensure all files were
restored. There may be a few files in the SYS account which do not restore
with error 'KEEP SPECIFIED; FILE NOT RESTORED'. This is ok, these files were
built by the INSTALL.
:RESTORE *T;COMMAND.PUB.SYS;SHOW;OLDDATE
Restore UDC Catalog file from full backup set. INSTALL creates an empty
COMMAND.PUB.SYS, this needs to be replaced with the copy on backup.
Now load a new scratch tape.
:STORE LOG####.PUB.SYS;*T;SHOW;OLDDATE;PURGE
This will store and purge old system logfiles (note that the current, in use
log file will not be stored and purged). We do not want old logfiles on the
system. If you require old logfiles for auditing purposes, then keep this
tape around so that you can use the logfiles again at a later date.
All data is now restored, however, if you have networking software, it will
not start until the next START NORECOVERY. So we now need to shut down the
system and reboot, this will also ensure that your SYSSTART file gets
executed at startup time (it was not on the system after the INSTALL, so has
not yet been executed).
(Ctrl)-A =SHUTDOWN
After 'Shut 6' message, do a (Ctrl)-B to get the CM> prompt. As before, the
startup script will vary from system to system, but essentially what we need
to do here is to boot from the Primary Boot path.
CM> TC
Confirm restarting system? Y
Interrupt within 10 seconds? Y
Boot from primary path? Y
Interact with IPL? Y
ISL> START NORECOVERY
System should now boot and be ready for use.
Finally, Stan Sieler (sieler@denkart.com) adds:
1) I'd not do a "Keep" and thereby lose my valid CONFIG.SYS.
Instead, I'd do: Keep NEW
and then, when cutting the SYSGEN tape, do:
BA NEW
TAPE VERBOSE
2) *ALWAYS* restart your machine via (control-B) RS, not (control-B) TC.
In reference to the ":RESTORE *T;@.@.@;SHOW=OFFLINE;KEEP;DIRECTORY;OLDDATE"
3) I usually do: FILE OFFLINE;DEV=LP,1,1
to guarantee that the OFFLINE listing stays around for awhile.
4) After restoring the system as shown above, I *strongly* recommend
putting the tape back online and doing:
RESTORE ; @.NET.SYS, @.DIAG.SYS, @.ARPA.SYS; DEV=1; OLDDATE; SHOW
Note: if the tape backref is omitted, RESTORE assumes an implied
file (username);dev=tape
On our machine, and many clients, both nine-trk and DDS are in class
TAPE.
The DEV=1 helps avoid system aborts 653 and phantom file labels
after the next UPDATE.
And in reference to the instructions "to boot from the Primary Boot path. CM
5) Again, RS not TC.
(This is HP's official stance, as well. One reason is that RS
"clears" memory, and TC doesn't. Also, on some older 950s a
TC may leave parity checking turned off, if I recall correctly.)
MB
:SYSGEN
&nbs
p;
sysgen> IO
io>md ldev=xxx
mode=AUTOREPLY
io>ho
io>e
sysgen>keep
Then take you
system down for a start norecovery.
RC
1) Run IOCONFIG.PUB.SYS
2) ioconfig: ADEV ####;PATH=NONE;ID=HPTCPJD
3) ioconfig: AC classname;LDEV=#### (optional)
4) ioconfig: EXIT
5) Add dev# and ip address in NPCONFIG.PUB.SYS
1000 (network_address = 128.128.50.1) #IS dept NMIS02
1001 (network_address = 128.128.50.2) #Computer room NMIS01
6) STARTSPOOL #####
TO DELETE PRINTER
1) STOPSPOOL ####
2) Run IOCONFIG.PUB.SYS
3) ioconfig: DDEV####
4) ioconfig: EXIT
The above is all it takes to add a network printer to the HP3000. The # signs in the NPCONFIG file are for comments.
Notes:
HP only supports "PCL" printers and JetDirect cards/boxes. The limiting factor with cards/boxes is that the MPE spooler uses SNMP status queries with the print card/box to get it's "status". Since most other "compatible" network-printer solutions don't support SNMP in exactly the same way, it won't (fully) work on other than Jetdirect hardware. The loss here is page-level recovery and status info. If you can do without that, you can add the following keyword to your NMCONFIG entry to turn SNMP off and allow you to print to non-JetDirect hardware:
snmp_enabled = FALSE
Also, for those of you trying to cheat and get non-HP (or non PCL printers) to work, some other valuable keywords you can add to a printer description in NPCONFIG are:
pjl_supported = FALSE
jam_recovery = FALSE
For example, even HP256x printers have been made to work via network printing with these keywords.
And for you unfortunate folks with firewalls:
To get your system talking to a Jetdirect equipped network printer through a firewall, you need to allow
Some firewalls (and some security policies) may have problems with allowing SNMP - in which case you can turn it off in the NPCONFIG file entry for the printer (snmp_enabled = FALSE), but no printer status checking will work.
See http://www.3k.com/ under the vendor/products directory.
See http://www.3k.com/ under the vendor/products directory.
See http://www.3k.com/ under the vendor/products directory.
There are also some newsgroups dedicated to computers for sale.
Are 3rd-party upgrades a "safe" option?
JK
As (memory), check into maintenance and guarantees too. If you are doing a box-swap, you're probably safe *IF* the original box was a 3000 (not a 9000) and is certified for HP maintenance and the vendor covers the box installation. If a board swap, be more careful. Be sure you are getting a *real* update from the vendor (CPU, PDH, and sometimes PSM firmware, etc) from a bona-fide 3000. If the equivalent parts came from a 9000 box you're again looking at time and materials, and possibly invalidating your service contract (there *is* a difference, albeit small, and HP has a low tolerance for swapping parts across platforms).
There are several third-party printer companies that are quite popular in HP3000 shops. In fact, many "HP" printers are actually manufactured by other companies and HP "puts their name on them". In many cases, you can buy the exact printer directly from the original manufacturer for quite a bit less than direct from HP.
In the past, third party disc drives on HP3000s were almost unheard of. HP claims that their drives (even the drives not made by HP), though appearing identical to drives from other manufacturers, have "sector atomicity" robustness the others do not. Nonetheless, third party disc drives are becoming more popular on HP3000s - including many RAID solutions and disc farms. Several of these are even recommended by HP (especially since they stopped making drives themselves some time ago).
Several HP3000-L members have installed FW SCSI drives (bought from local PC/Computer stores) and successfully installed them in their HP3000s - though the robustness of these drives is not thought to be of the same level as drives actually sold for HP3000s.
I want to add memory to my system. Are 3rd-party brokers OK? Is "used" memory OK?
JK
For the most part, yes. If you get HP boards, make sure they have been certified for HP maintenance; if you get 3rd party boards, check on their support options (typically FedEx overnight board swap). EMC, Kelley, and Datasys are (IMHO) reliable and easy to install (speaking first-hand, no sleight to the other vendors). But bear in mind if you have a suspected memory problem, you are best served by having adequate HP memory to be able to pull the "alien" boards and still be able to reproduce an error if you are calling HP for service, else you may get into "finger-pointing" relative to the cause of the problem. If the "alien" board is at fault, you're looking at time and materials billing. Likewise for any other 3rd party products you purchase. There are some good deals out there. There are also some questionable brokers.
CB
Note that there are a few brokers out there that will actually let you try the new memory out before you make any committment to buy. You get the opportunity to verify that the memory will help your system without having to "stick your neck out" getting a purchase order approved.
Also note that even HP buys some memory from 3rd party brokers for their internal systems.
SS
Short answer for the "cons":
0) HP usually calls theirs DDS, sneering at the lowly DAT. I think
it's to distinguish the computer certified tapes from the
audio tapes?
1) DATs have no significant error recovery mechanism.
If you encounter a tape error on a DAT, the rest of that file
is inaccessible. Subsequent files on the DAT might be accessible,
if your software is smart enough to issue the appropriate
"skip" commands PRIOR to encountering the bad tape area.
2) DATs can't be read past the current end-of-tape marker.
If you have a full DAT (say, 1.2 GB), and accidentally insert it
into a drive (write enabled) and write over the first 2 bytes...
poof! The entire rest of the data is "gone". (For the *really*
desperate, an unsupported desperation trick *might* suffice to
get about 90% of the data back, with a lot of expensive work.)
3) Lack of feedback
You can't look at a DDS while in use, and estimate how much tape
has been used.
You can't determine how many write-retries were successfully done.
(True, SCSIDDS offers a peek at both values, but with difficulty,
and not all DDS drives on HP3000s are SCSI.)
4) Slower per megabyte than reel-to-reel, I think.
5) Slow load/eject times (compare to thread-it-yourself reel-to-reel,
and to some auto-load reel-to-reel).
6) Jamming in drive
Pros:
1) capacity!
2) small size!
3) cheap!
You'll note that none of the pros/cons are HP3000 oriented, but are true on any system.
LSB
... we experienced consistently a fifty-percent wall-time reduction after switching from HP7980 reel-to-reel to the HP model 6400 4000 DC. That's the 120 meter variety with hardware compression. At first, we were continuing to have compression turned on in RoadRunner, but since we cannot turn off the hardware compression (5.0 push), it seemed redundant; we turned it off. The through-put improvement was a big surprise as the transfer rate is rated as significantly slower that 1/2" tape. No one can explain it, but it is not a fluke.
RLE
The DAT engineer said: "The bigger the better." Via HPFOPEN, MPE supports a 16K buffer, which is ok for DATs. STORE/RESTORE can go up to 32K. The basic criteria is to keep the tape streamin'.
StS
Yes! For DAT, the bigger the better. 32K will be good! Particularly for the compression drives, small blocks do not compress efficiently.
MR2
To prevent problems with your tape drive, keep the following list in mind:
* Are your tapes DDS certified? This is indicated by a DDS (Digital Data Storage) symbol on the tape.
* Do you replace your tapes on a regular basis? HP recommends tape replacement after approximately 300 uses.
* Do you clean your drives at regular intervals? Drives should be cleaned after 25 hours of use (Calculate this at approximately 2 hours for each tape use) or whenever the Cassette indictor light flashes the Caution signal (indicated by an alternating green light for 4.5 seconds and and no light for 0.5 seconds.)
* Do you replace the cleaning cassette at regular intervals? Each time a cleaning cassette is used, the date of use should be marked on the label. After 25 cleanings, the cleaning cassette must be replaced. If you fail to record each use of the cleaning cassette, after 25 uses the cleaning cassette may appear to be working normally, but it will be doing nothing. A working cleaning cassette will take about 54 seconds to clean the drive, a used up cleaning tape will take about 11 seconds.
* If you are having problems with a tape drive reporting that tapes are bad, and you suspect that the tapes are ok, HP recommends cleaning the drive three times in a row. If the problem still persists, place a call to have the drive replaced.
SD
FWIW, the "needs cleaning" indication is completely different for the HP6400 model 4000[DC] drives, as are most of the other indicators.
SC
Testing that we performed on backup technologies yeilded the following results:
A HP7980XC delivers backup rates of approximately 600-650 KBS. A DAT drive delivers backup rates of approximately 300-350 KBS.
The testing was performed on a standalone HP3000 960 dedicated to the test. All stores were performed with TurboSTORE. The data was a copy of actual production data and spanned multiple tapes.
The HP7980XC was attached to a dedicated HP-IB card to ensure that the tape would operate in stream mode. Our testing indicates that 2 HP7980XCs can be connected to each HP-IB card and maintain stream mode. Data storage rates were approximately the same whether the drive was in hardware compression mode or not.
The DAT drive was attached to a dedicated SCSI card. Several DAT drives were tested. The transfer rates were approximately the same but amount of data stored varied.
Time for reel changes were not included in the transfer rate calculations even though it impacts the backup time.
JO
We need to be a bit more specific about what rates and tape drives are wanted. RE HP DDS drives (note DDS is a subset of DAT) the 60 Meter and 90 M drives have a native transfer rate of 183KBytes/sec. The new 120 M drives have a 510KB/sec rate. The 7980's rate is 781KB/sec.
But for Backups using HP TurboStore (or other products) the capacity of the tape media, mounting and rewind times, channel capacity and usage, and compressibility of the data, and who does the compressing (hardware or software) must be considered.
For an actual datapoint, I have a 3000/947 which backs up about 17 million sectors (4.3 Gbytes) and it takes about 3.5 hours with software compression to a 90 M DDS. It uses about 2.4 hours to a 120 DDS with hardware compression. That is 1.25 GB/Hour and 1.83 GB/hr respectively. This may be somewhat limited by the tape and discs being on the same SE SCSI interface. Note that this is a low compression ratio as if I add only a few more files to the backup the 90 M DDS goes to two tapes.
For one reel STOREs the 7980XC is faster, but for the storage and convenience of one reel Full Backups, DDS with compression is King.
BT
Today, the system doesn't care if the console is switched on or not.
However, if the console is left hung via a control-s, console message
buffers
will eventually fill up and any process that tries to send a console message
will be suspended until the message can be delivered, at
least for a minute
or so.
MP
There are 63 what I call genmsg buffers. Every console message requires
a genmsg buffer and a TELL or WARN requires
a buffer per destination
terminal. When the system detects that all of the buffers are in use, it
starts a 30 sec. timer. If no buffers free up after
that time, the
system issues a hard-preemptive I/O to the console in an attempt to free it
up. Then, the system waits another 30 sec. If the
console is
still hung, then we abort every I/O that's using a genmsg buffer.
So, if the console can never complete I/Os, for whatever reason, then there is a 63 message/one minute timeout/abort 63 I/Os cycle that continually repeats itself. A system with a lot of console/TELL/WARN I/O traffic will seriously bog down as the result of a hung console.
BT
After that, the system gives up and starts tossing all its undelivered
messages into the system log file. So if you _have_ managed to hang up
your
console (turning it off won't do it by itself), you'll need to look in
your log files. I'm not sure what happens if you don't have console
logging
turned on -- the system won't crash, but it's not clear what, if
anything, will happen to undeliverable messages.
MP
Actually, the
system doesn't do anything special with the aborted I/Os. If console logging is
enabled, then the aborted console messages will be found in the console log
files. If console logging is NOT enabled, then these messages are
lost. If the aborted message was targeted at a non-console terminal, then
this message is always lost.
S/68 - S/70 --- Increased CPU (not
disc) Cache for a rather significant increase in CPU speed.
There were a
couple of other things that differentiated the S/64A and S/68A from the S/64B,
S/68B, and S/70. Most notable were a different
SSDP (System Status and
Display Panel) and a totally different power supply and distribution group.
JK
(As I remember) the offerings at the time were 40, 44, and 64.
The 40 and 44 were essentially the same horsepower and had interchangeable
boards, but the 40 was a square-ish box with no terminal connections (other than
console); the ADCCs ran out to little metal boxes or your ATP came to a
mini-cabinet. The 44 was the "desk" looking thing, the 64 was the "chest
freezer".
MPE V started the cache business and had V/P initially (cache
only), V/E (new processor and firmware for CST tricks), and later V/R (for the
obsolete Series III and below). V/P was purely software, but you had to
"upgrade" from 40-42, 44-48, or 64-68 to get it; consequently if you got the
upgrade early (before the firmware was released) and deleted the memory board,
all you got was a new FOS/SUBSYS tape and a new nameplate (for a considerable
chunk of money). I know, I got two of them :-) Of course you later
got the firmware (and V/E).
The next processor speed-up changed 42-52,
48-58 (for sure) and I think also 68-70. I've had a 40/42/44/48/52/58,
dunno about the rest :-)
CB
Yes, 68-70 added CPU/microcode cacheing.
We went through a few of those upgrades ourselves.