« February 2006 | Main | April 2006 »

March 31, 2006

A day to focus on later dates

This weekend, Daylight Savings Time takes hold in many countries — earlier than ever in the month of April. It will be the final April where the change will take place, in the US; beginning in 2007, as a result of a new federal law, DST kicks in during March. The shifting nature of that time change might pose a puzzle for some 3000 administrators. A command file can help in getting the 3000 on the correct time, and keeping it there through 2007's new cutover dates.

SETCLOCK is at the heart of the 3000's time change capabilities. The command has had some problems in the past, flaws that HP has always fixed in due time. Back in 2002, the command was patched in mid-year for MPE/iX, although HP didn't make an announcement about the fix in a timely fashion, so to speak. Applying the latest PowerPatch to 7.0 brings SETCLOCK up to date.

A command file written by Dave Powell of MMFab will keep the 3000 caught up with the changes in time changes, both this year, and in years to come.

Powell offered his command file in two flavors:

!#   Check for Daylight Saving Time change -- NEW
!IF  HPDAY = 1
!    IF  HPYYYY <= 2006
!        IF  HPMONTH = 4   AND   HPDATE < 8
!            SETCLOCK TIMEZONE = W7:00
!        ELSEIF  HPMONTH = 10   AND   HPDATE > 24
!            SETCLOCK TIMEZONE = W8:00
!        ENDIF
!    ELSE
!        IF  HPMONTH = 3   AND   HPDATE >= 8   AND   HPDATE <= 14
!            SETCLOCK TIMEZONE = W7:00
!        ELSEIF  HPMONTH = 11   AND   HPDATE <= 7
!            SETCLOCK TIMEZONE = W8:00
!        ENDIF
!    ENDIF

or, if you don't like white space....

!#   Check for Daylight Saving Time change -- NEW (condensed)

There's been a tradition of checking on how to shift the 3000's time in the weeks leading up to the change. The 3000 NewsWire has reported several strategies, including advice on understanding the difference between hardware and software clocks on the system. John Clogg explained:

The hardware clock is supposed to be set to UCT (AKA GMT), and the “software” time is derived using a time zone offset that is manipulated by the SETCLOCK command. It is not unusual to see systems where this relationship isn’t understood, where the hardware clock has never been set correctly, so the offset is hundreds of hours! The SHOWCLOCK command will show you the current status of the hardware and software clocks. In general, the SETCLOCK command adjusts the offset, and the date and time setup during a system start adjusts the hardware clock.

Stan Sieler added that a free FIXCLOCK command file at www.allegro.com/software/hp3000/allegro.html can help you fix the hardware clock date/time.

Then there's the nuance of using NTPDATE, or not, during the springtime DST change:

For the last couple of years during the time change, we have started also running an NTP time synchronization job at 3:00 AM every morning. The TZ variable is set to PST8PDT. Now as I understand it, NTP is adjusting the UTC time. But does it also adjust the time zone? In other words, are we duplicating effort with these jobs?


John Clogg replies:
    NTPDATE will not adjust the time zone. It will only accomplish the tiny adjustment to UTC time caused by the drift in your system’s hardware clock. There is no problem running both programs the same night, as long as your DST adjustment has completed when you run the NTP adjustment. Since the spring adjustment moves the clock forward, it will happen immediately. The autumn adjustment can take a couple of hours or more, so it calls for greater coordination.

12:34 AM in Hidden Value, Homesteading | Permalink | Comments (0) | TrackBack

March 30, 2006

Count on those massive disks really working

Yesterday one of the advocates on the OpenMPE board of directors posted a bit of a brag as well as a thank-you to HP. Donna Garverick, who'd lobbied HP along with hundreds of customers voting in the final Systems Improvement Ballot, showed off the results of her first Bigger than 3000 Gig volume under MPE/iX.

Garverick, who manages 3000s for Long's Drug, configured her new disk at 400GB as an interim step, just to check out the programming HP's 3000 lab did to accomodate the bigger disks. Advocates like Garverick and companies who'll have to put several more years onto their 3000s may need to use bigger disks; the HP enhancement lets the 3000 see up to a half-gigabyte of storage.

Before now, an MPE system couldn’t mount a disc drive larger than 300GB. Thanks to Jim Hawkins’ (et al) hard work, we can now. This request was a SIB item from a year ago (or so) and is very much aimed at homesteaders (who at some point won’t be able to find a 4...9...18...gigabyte drive).

Since I’m in the process of bringing up a new system (with a disc array attached), I told HP I could briefly configure this system with a large disc so that I could test their patches.  I built a 400GB LUN — which MPE sees somewhat larger (odd) and viola :-)

I rather figured since I was one of the people advocating for HP to do this work, the least I could do was be decent about it and give it a test drive.

Garverick had started her message to the community by posting a DISCFREE report (under the heading, "i can't count that high" showing so much free space that some people thought she'd erased her volume completely:

DISCFREE A.75.01 Copyright (C) Hewlett-Packard 1992.  All rights
                         WED, MAR 29, 2006,  9:01 AM


          |      Configured      |        In Use        |    

Device    |      35547040        |       4698992 ( 13%) |      30848048
( 87%)
Permanent |      26660256 ( 75%) |       4333216 ( 12%) |      22327040
( 63%)
Transient |      26660256 ( 75%) |        365776 (  1%) |      26294480
( 74%)

Device    |    1677721584        |        678928 (  0%) |    1677042656
Permanent |    1677721584 (100%) |        678928 (  0%) |    1677042656
Transient |    1677721584 (100%) |             0 (  0%) |    1677042656

Device    |    1713268624        |       5377920 (  0%) |    1707890704
Permanent |    1704381840 ( 99%) |       5012144 (  0%) |    1699369696
( 99%)
Transient |    1704381840 ( 99%) |        365776 (  0%) |    1703337136
( 99%)

01:57 AM in Homesteading, News Outta HP, User Reports | Permalink | Comments (0) | TrackBack

March 29, 2006

A quiz to teach about carriage control

The HP 3000 has Carriage Control (CCTL) capabilities that can come in handy when formatting output from applications. Robelle's SMUG Book entry says that CCTL

is the process of moving the print position (or "carriage") to the next line or page, as well as performing advanced functions such as overprinting and printing through the perforation. Normally the program specifies the CCTL code for a line through the Fwrite intrinsic. MPE has too many ways to do CCTL and modern printers are controlled via PostScript or Printer Command Language (PCL) anyway.

But CCTL has its users, and understanding how it works can be a key to resolving print problems with some of those aging applications you might have under your wing. Dave Powell of MMFab wrote up a quiz on CCTL behavior recently, one he shared with the 3000 community. He explained some nuances in his article, one he researched as part of his ongoing CCTL-to-Rich Text Format (RTF) utility.

By Dave Powell

Start with a simple program that generates a 132-column, fixed length report. Run it with output to a spool file, like:

file rptfile; dev=lp,1
run progname

Then run it with output to a permanent cctl file, like:

file  rptfile = permfile, new; dev=disc; cctl; save &
  ; rec=,32,f,ascii; disc=60000,32,0
run progname

Finally send the perm file to a spool file, like:

file newspool; dev=lp,1; cctl
print  permfile > *newspool

1: Are the two spoolfiles identical ?
2: What about the printouts ?


1:  The spool files should be identical except for trailing blanks.  The 1st spoolfile has all 133-column records for the actual report output (plus special zero-length records at the beginning and end).  The second file may have some shorter records because trailing blanks have magically disappeared. So, the second may be much smaller, and will look more like a “real” variable length file. Check the file size with:
calc finfo(“o###.out.hpspool”,”bytefilesize”).

Just because spool files are ‘variable’ doesn’t mean anything eats blanks your app sends.  But :print seems to eat them.  Test “:print SomeFixedAsciiFile > NewTempFile” without any file equations, and see what it builds.

The non-spool file created the 2nd time the pgm runs has all 133-column records, because omitting the record size in the file-equation causes the rec-len to be picked up from the program.

2:  What printouts?  The out-pri is 1.  Well, besides that... the printouts will appear identical, unless your eyesight is good enough to spot trailing blanks on paper :)
FWIW, most of our big print jobs happen this way -- works like a charm.

Powell has a command file available that reports info on a CCTL file -- the CCTL code, the first few bytes of the data, and the record length:


parm     _from   =   "";   _limit = 200                                 
parm     entry   =   'main'                                             
option nolist                                                          
if   "!entry"    =   "main"  then                                       
     if  bound (X_ERRORS)                                              
         deletevar   X_@                                                
     setvar  X_ERRORS  0                                                

     if  fsyntax('!_from')<>"MPE"  and  fsyntax('!_from')<>"POSIX"      
         setvar  X_ERRORS  X_ERRORS + 1                                 
         echo Parm-1 (file name to convert FROM) is required            
         echo and must be a valid file name.                           
     if  len ("!_limit") < 1  or not numeric ("!_limit")               
         setvar  X_ERRORS  X_ERRORS + 1                                 
         echo Invalid limit on #-of-records to read                     
         echo If not omitted, 2nd-parm must be a positive integer      
     if  X_ERRORS      >   0                                           
         showvar X_@                                                   

     setvar  X_FROM    '!_from'                                        
     file    XFROM =   !X_FROM,  OLD;    CCTL                           
     if  not finfo('*XFROM','EXISTS')                                  
         echo No such permanent file as "!X_FROM"                      
         setvar  X_ERRORS  X_ERRORS + 1                                 
         setvar  X_EOF     finfo('*XFROM','EOF')                        
         if  X_EOF     <   2                                           
             echo File '!X_FROM' does not have enough records          
             setvar  X_ERRORS  X_ERRORS + 1                            
     if  X_ERRORS      >   0                                           
         echo Old file '!X_FROM' cannot be processed                   
         echo format.                                                   
         showvar X_@                                                   

     setvar  X_LIMIT     min (X_EOF, !_LIMIT)                           


     xeq !hpfile  entry = ReadLoop   < *XFROM                           

     if  X_ERRORS  =   0                                                
         echo Done OK                                                   
         showvar X_@                                                   

elseif   "!entry"    =   "ReadLoop"                                    
     echo Line  Cctl-code            Data        RecLength             
     setvar  X_CNT       0                                             
     while   setvar  (X_CNT, X_CNT+1)    <=  X_LIMIT     do            
         input   X_REC                                                 
         setvar  X_CNT_      RHT ('   !X_CNT',4)                        

         setvar  X_LEN       LEN (X_REC)                               
         if  X_LEN           >   0                                     
             setvar  X_CCTL      LFT(X_REC,1)                           
             setvar  X_ORD       ORD (X_CCTL)                           
             setvar  X_ORD__     "!X_ORD"                              
             setvar  X_OCT__     "![OCTAL(X_ORD)]"                     
             if  X_ORD >= 128   and   X_ORD <= 192                     
                 setvar  X_DESC  "Adv" + "![X_ORD - 128]"               
             elseif  X_ORD >= 2   and   X_ORD <= 42                     
                 setvar  X_DESC      "SingleSp"                        
             elseif  X_ORD >= 192   and   X_ORD <= 207                  
                 setvar  X_DESC      "VFC code"                        
             elseif  X_ORD = 49   or   X_ORD = 51                      
                 setvar  X_DESC      "CondEject"                        
             elseif  X_ORD   =    0                                    
                 setvar  X_DESC      "Single"                           
             elseif  X_ORD   =   45                                    
                 setvar  X_DESC      "Triple"                           
             elseif  X_ORD   =   48                                    
                 setvar  X_DESC      "Double"                           
             elseif  X_ORD   =   64                                    
                 setvar  X_DESC      "PostSp"                           
             elseif  X_ORD   =   65                                    
                 setvar  X_DESC      "PreSpace"                        
             elseif  X_ORD   =   67                                    
                 setvar  X_DESC      "NoAutoPage"                      
             elseif  X_ORD   =  208                                    
                 setvar  X_DESC      "NoSpNoRet"                        
                 setvar  X_DESC      " "                               
             setvar  X_CCTL      '*NONE*'                              
             setvar  X_ORD       0                                     
             setvar  X_DESC      "{none}"                              
             setvar  X_ORD__     " "                                    
             setvar  X_OCT__     " "                                    

         if  X_LEN       >   1                                          
             setvar  X_DATA      str (X_REC+rpt(" ",16), 2, 16)         
             setvar  X_DATA__    repl (X_DATA, ">", "!>")               
             setvar  X_DATA  " "                                       
             setvar  X_DATA__    rpt (" ", 16)                         
         echo ![RHT ("   !X_CNT", 4)];&                                 
              ![RHT ("   !X_ORD__", 4)] &                              
              ![RHT ("    "+X_OCT__, 5)] &                              
              ![LFT ("!X_DESC          ", 10)];&                        
              ![RHT ("   !X_LEN", 4)]                                  

         deletevar   X_REC                                             

##-  just one more error-check                                          
     setvar  X_ERRORS  X_ERRORS + 1                                    
     echo Error - invalid entry point                                  
     showvar X_@                                                       

#    Cmd file to show an HP cctl report file's CCTL codes               
#    :printspf may be better for spool files, because it shows         
#    the p1 & p2 fields.                                                
#    Dave Powell,   MMfab   early 2004 and 03/2006                     
#    Parms:                                                            
#    1:  File name to read (must be cctl type, can be spool or not)    
#    2:  max number of records to read, even if the file is            
#        longer.  Its SLOOOOW, so use this.                            
#    For each line of the input file, it shows:                        
#    -   Line #                                                         
#    -   Carriage control code, if any, in decimal, octal and          
#        my description of the meaning.  Different codes that have      
#        the same effect may have different descriptions.               
#    -   The beginning of the data part of the record, padded or       
#        truncated to 16 bytes.                                        
#    -   The total record length (including the cctl byte).            
#        Spoolfiles will have records with ZERO length, hence no       
#        cctl code.  The 0-len records at the start & end have         
#        open & close commands in the 'function' word of the            
#        hidden area.   See :printspf and Appendix A of the            
#        Native Mode Spooler Manual.                                    

01:15 AM in Hidden Value, Homesteading | Permalink | Comments (0) | TrackBack

March 28, 2006

When slower development might open new doors

In the world of PC-based servers — one of HP's alternatives to the 3000 — customers are facing another delay in receiving Vista, the most significant upgrade to Windows since the XP release of five years ago. An article in yesterday's New York Times suggests that backwards compatibility is slowing Microsoft's Vista release schedule. It's the sort of thing that began to slow down HP 3000 development, just about the same time that Microsoft rolled out XP.

This kind of backwards compatibility is a marvel for the customer and developer partners to enjoy. But it puts tremendous strain on the technological bandwidth of the vendor that maintains the compatibility. One unexpected benefit to the 3000 community from HP's exit: MPE/iX might get simple enough to give to an outside organization to maintain. A new steward might preside over a streamlined product line.

The comparison between Windows' compatibility and the 3000's looks evident to Wirt Atmar, founder of AICS Research. The MPE/iX supplier of QueryCalc has moved into Windows software in a major way since its development of QCTerm, a free HP 3000 terminal emulator designed to run on Windows-based systems. Atmar said that HP's devotion to the 3000's backwards compatibility gave it something in common with Microsoft — and most of nature.

"As for nature, backwards compatibility is the norm," he said while we discussed the NYT story yesterday. "In fact, it’s the  only way that things can work. There are no major revolutions in code at any point in time. Everything is a continuous flow." For the record Atmar thinks the Times story is nonsense, and that Apple forced AICS to turn to Windows and away from the Mac.

We began life as a 100 percent Mac shop in 1984, but 10 years later, we were essentially forced to change over to PCs, simply because by then Apple had gone through two redesigns  (6800 to 68000 and 68000 to PowerPC) in which they abandoned a significant number of previously working applications (and everything that we were interested in)  in order to rush ahead of Microsoft. Since then, they’ve repeated this  pattern either two or three more times, depending on how you count. I’m surprised that they have any customers left, to be honest.

Atmar went on to contrast Microsoft's compatibility devotion. "I hold Microsoft in very high regard, in the same way that I did [HP 3000 division] CSY, for their commitments to backwards compatibility. If  anything, Microsoft is much better at it than was CSY. For example, in XP, pretty much all prior DOS functionality was resurrected -- and hyperthreaded to boot. I thought that that was pretty damn cool."

The same kind of work, however, CSY could not perform fast enough — keeping the 3000 line in step with applications and systems more than two decades older. It probably delayed the N-Class 3000 releases (using the new PCI bus), which were delivered beyond their expected deadlines. This chilled sales during 2000 and 2001. HP announced its pullout of the 3000 market near the end of 2001.

HP will step away from the 3000 at some point, perhaps in 2009 — a move which, more than seven years after the announcement, will probably signal the end of the backwards compatibility for the 3000 line. No more development, no concern with preserving the value of customer investments made during 1990s. (HP has also used that preservation promise to keep recent N-Class systems hampered in horsepower, saying that customers who decided value propositions in 2002 and 2003 deserve to have their price points protected.) HP's ending could well be a beginning, though, of a simplified line of MPE/iX releases, able to be managed by an outside organization. The HP of this century does not promise such complete backwards compatibility.

The costs of such compatibility can be burdensome, to the point of threatening a product's ability to compete. Itanium-based servers might have been delivered far sooner if their x86 code compatibility wasn't so essential in hardware at first. This year, some Windows customers are wishing Microsoft would look forward, rather than back, to go faster toward the Vista of tomorrow. Others, who can afford to wait for new features, are delighted to preserve what works from the past.

01:39 AM in Newsmakers | Permalink | Comments (0) | TrackBack

March 27, 2006

The one thing to add to make a 3000 faster

Some say it's faster storage, and others claim it's extra CPUs. But the one item that any 3000 customer can add to their systems to improve performance is almost always memory. Unlike processors, upgraded systems or other things HP simply doesn't make anymore for the 3000, memory has always been available from third parties. How much to add depends on who you ask.

HP has weighed in on memory needs for MPE/iX on its Jazz Web site. In a white paper written a couple of years ago about investments in disc storage, HP's Walt McCullough said that the 3000 is designed to go to disk only when what it needs is not in memory. Therefore, he says

The greater the amount of memory for the HP e3000, the more efficient it will run (to a point). MPE uses available (server) memory as a cache to store user process information and also keeps a virtual copy of the file.

But add to what point? Ask Bill Lancaster, CTO of Lund Performance Solutions, and he'll give an answer that has more to do with a CFO than a CTO. "The best axiom about memory on MPE is to buy all you can. Very, very rarely is there such a thing as too much memory."

Lancaster explains that MPE's memory utilization

is best viewed from the perspective of how effective it is in acting as a cache. The two top metrics in evaluating this are Read Hit Percentage (the percentage of all read IO’s satisfied in RAM) and CPU Busy on Memory Management. You want a Read Hit Percentage in excess of 95% and CPU Busy on Memory Management less than 4-5 percent.

Users report that memory is "so cheap now it's really a no-brainer" to add it to HP 3000s. The "add as much as you can buy" rule of thumb might trump another guideline, one recounted by Dave Waroff of the San Juan School District: 64Mb/CPU + 16Mb (for the 3000's Transaction Manager) + 8Mb/batch job + 4Mb/session.

If you're unable to buy all the memory that you can fit into your 3000, and some more strategic measurement will help, then Lancaster's firm offers a good multi-platform solution to measure anything on a Unix, Windows, Linux or MPE/iX server. Meta-View Performance Manager includes Lund's Performance Gallery Gold suite. Meta-View has a Flash presentation on the Lund site that reviews system screens and product strategy. It's a product you can use to justify an optimal memory configuration and improve performance. Not bad for an add-on that is "so cheap now it's really a no-brainer."

01:33 AM in Homesteading, User Reports, Web Resources | Permalink | Comments (0) | TrackBack

March 24, 2006

Making that migration timetable yours

In our weekly podcast — a six-minute MP3 file — we hear from a early advocate of migration about how the job is taking longer than predicted back in HP's migration roundtables of '02 — but is being done on a 3000 level of service. That is, with an eye toward efficiency of code, ported by experts. The gurus happen to be from outside the 3000 customer's organization, but they are partnering with IT staffers who know the applications from the inside. That kind of partnership is extending migration timetables, so HP has expanded its migration timetable to match.

If you migration is not going to be finished in '06, don't worry. Do what it takes to make the timetable your own, based on your business needs. Oh, and to match the expected level of service your 3000 is giving your organization. Whatever you replace the 3000 with should last a long time — which might mean a migration will take longer than expected.

06:38 PM in Migration, Podcasts, User Reports | Permalink | Comments (0) | TrackBack

March 23, 2006

Controlling grep and securing FTP

What’s the correct invocation for grep from the CI?  I tried to use


but while it runs, I get a bunch of odd-looking errors rather than file names displayed.

Ray Shahan replies:


We use the allow statement on our HP 3000 to restrict what can connect to our 3000's FTP server:

ftp allow (and so on)

We also have our 3000 configured to use our network’s (Windows-based) DNS servers, using RESLVCNF. Can I change the allow statement in INETDSEC to use DNS names rather than IP numbers?

Matthew Perdue replies:

Yes. For example:
ftp  allow 192.168.0.* 172.16.0.* domainname.net 192.168.1.*

Be careful if you need to use the line continuation character, as the HP documentation is wrong! Use the “\” rather than the “/” in postion 80 as shown in the documentation.

06:32 PM in Hidden Value | Permalink | Comments (0) | TrackBack

March 22, 2006

Alternate universe spins support, system supply

HP's exit from the 3000 market has been pushed back at least two years, but that doesn't mean that a customer's non-HP universe got stalled in response. Third parties have been building toward a close of HP's 3000 business in 2006. The vendor's extension of its support, on a limited basis, just puts HP among the realm of aftermarket alternatives for 3000 and MPE/iX support and system supply.

Some third party suppliers have been preparing for this year ever since HP's last year of 3000 sales. In 2003 Pivital Solutions became an official HP 3000 reseller. The company remains in the market as a supplier of remarketed systems (the only kind available anywhere today) as well as support plans. Back in 2003, Pivital's CEO Steve Surcaci announced the vendor's long-term goals for joining HP's sales network with less than six months of selling time left on HP's calendar.

"We’re trying to position ourselves not just as a reseller between now and Oct. 31, but we want to make sure HP understands we want to play a long term role with these homesteading customers — and support is a major portion of that. If and when the opportunity ever presents itself from HP to authorize a select group of companies to manage support for them, we want to make sure they understand we want to play a major role in that."

In that realm of official resale outlets, Pivital sourced its 3000 stock from North American distributor Client Systems during 2003. Now Client Systems has become a direct-to-customer outlet in its own right, joining the HP Renew program to offer used 3000s.

"The great thing about the Renew program is that all [of the equipment] has been reworked and is warranteed as new," said CEO Casey Crellin of Client Systems. Crellin, who was the CFO of Client Systems during its HP distributor days, bought out partner Mike Murphy last year and has expanded the company's dealings with HP Renew.

Client Systems is now handling integrations and system upgrades of 3000s for HP Renew, after a couple of years of work delivering PA-RISC HP workstations through the program. HP takes in 3000s as customers trade them in to get credits toward Integrity or other HP Itanium-based servers. Warranty support comes from HP for the Client Systems remarketed servers.

While Client Systems and its used equipment arm Phoenix 3000 are listed on HP's limited Where to Buy 3000s Web page, right now the 3000 products are not listed on the HP Renew Web pages. But Crellin says the 3000s are available from his company. Resellers and HP sales personnel in Europe and Asia-Pacific can also use Client Systems as a 3000 channel, he added. "There's not much over there in the way of systems," he said. "It's mostly Stateside."

04:34 PM in Homesteading, Newsmakers | Permalink | Comments (0) | TrackBack

March 21, 2006

Houston RUG sets dates for conference

The only HP 3000 conference scheduled for 2006 has now released specific dates in November, as well as a location not quite inside Houston. The Greater Houston Regional Users Group (GHRUG) is employing the "Greater" part of its name, choosing a venue on Clear Lake, southeast of the city, as the site of this year's November 10-12 conference.

The conference will be held over a weekend, Friday through Sunday. But the Clear Lake venue — the University of Houston at Clear Lake — is only a mile from the Johnson Space Center, NASA's ground-tracking and training base for its space missions. Galveston, with a great beach along its seawall on the Gulf of Mexico, is only about a half-hour drive down I-45, too. Attendees will likely need a car, since the University is at least a $40 cab ride from Houston's Intercontinental Airport. The University is much closer to the Hobby Airport in-town, where low-cost carrier Southwest flies.

November provides good weather for the southeast Texas area, too — with much less risk of hurricanes than those September conference dates for the 2006 HP Technology Forum, also being held in the Houston area. And last year's record Nov. 10 high in Galveston was 84 degrees.

GHRUG was developing details on the conference as of last week, but the price for the event "will be kept very low, as is the cost for exhibitors," said GHRUG conference organizer Denys Beauchemin.

"The initial announcement brought forth a deluge of responses to yours truly," he added, "and so we are moving forward!"

Attendees can visit www.uhcl.edu for information about the location of the conference, and the GHRUG site for ongoing details. There's an all-purpose registration form for GHRUG conferences at the site, too. For comments, questions or suggestions, contact Beauchemin by e-mail.

03:23 AM in Homesteading, Migration, Newsmakers | Permalink | Comments (0) | TrackBack

March 20, 2006

Bringing experience to 3000 migrations

A request for migration advice drew broad and ample response on the 3000 Internet newsgroup this month. The caliber and quantity of the response shows the community is moving along on its transition path, a good deal more openly and frankly than in 2004 and earlier. Nearly everybody was caught by surprise at the pace of migration among the 3000 customers — except the customers, who knew how integrated the 3000 has become with their business success.

A lot of the users' reports about databases revolved around swapping in Eloquence. John Pitman, writing from the IT operations of Ryco Hydraulics, said his company is working through the move from IMAGE on the 3000 to the Eloquence database on HP's Unix servers:

We have started down the Eloquence/C path on HP-UX some time ago. Having many C programs already under MPE, that decision was pretty easy. We got a free copy of Gnu C, and once we worked out how the libraries worked, it got easy. Some of the IMAGE access routines had some OS conditional code added; now we have over 100 programs with only a few conditional compiles in them that compile and run on either system. This is what you should head for with whatever language you use, so you don’t have two source bases to maintain continuously while you convert.

Pitman said he's been working with test versions of the main databases, "STOCK.TEST” as well as “STOCK.PRODN.” Each program’s open routine for HP-UX demands a variable exist to specify which base to open, which for normal users is done by the logon menu, so we can have both bases coexist safely."

Users agreed that creating Eloquence databases is easy and quick, though Ryco hasn't moved its largest 4Gb database. "Marxmeier are also pretty responsive to requests for help with the product," he added, "and it's certainly much cheaper than Oracle and others." He outlines the complications:

The only traps we have met so far are the need for all DBGETs to read complete records, item list subsets are/were not supported — a feature we used heavily in our old Basic code. Due to the level at which we have the C calls come in, there is a problem using TPI sequenced retrieval, but I don’t think it will be too hard to correct.

John Hohn, a migration consultant and vendor of a database migration tool at Comp Three, noted that moving to SQL from IMAGE might seem daunting, but it can deliver opportunities as a reward for the hard work. Contrary to some advice, Hohn said doing a complete job the first time through has benefits not easy to see at first:

As long as you’re  in there re-engineering, do it right the first time; the downstream  payoffs (some known now; most to be realized as the industry continues to evolve) are enormous.

I think people are afraid of writing SQL, but it’s not that hard once you get used to it. There are many advantages to SQL; one that we found is, by rewriting our complicated reports to not use cursors (doing  intermediate summations with unions instead of programmatic  adding up of numbers inside cursor loops) we can complete reports  in seconds (yes, seconds!) that used to take 20-30 minutes on the  HP3K.

But the real reason to go with SQL are the tools — which will get better and better and remain cheap or free.

Dave Waroff of the San Juan Unified School District testified to the comfort in making Eloquence the first migration landing spot:

Eloquence is the most direct replacement for TurboImage outside of MPE. If you’ve got COBOL programs with VPlus, conversion can be a trivial effort (given tools that are available). JCL can be replaced with native scripting on any system.

This combination provides the easiest porting, and an environment most  similar to the original. It’s more comfortable.

Bookware's Peter Osborne even noted that his organization (bookware3000.ca) has moved 800,000 lines of 3000-based COBOL code and IMAGE/SQL access to Eloquence on Debian Linux. "This combo isn’t usually considered for larger operations," he said," as they like to be backed by large vendors.

Ray Shahan of School Specialty said that a migration of under a year "no matter what solution you go with for the migration... is a bit of an aggressive (umm, dare I say...unrealistic) goal." 

School Specialty is "looking to see if anyone else has outsourced and succeeded — so we can see if outsourcing our legacy system maintenance is another option we can add to the analysis regarding what to do about migrating." As for advice on where to go or what to do, he says: 

Listen to all of the sales people that you can, and about every product/solution that you can find the time for.  All salespeople represent some part of a good solution, but the trick is to weed through all the good to find the bad parts of each solution (and many salespeople seem to lack the understanding of the negatives of their product/service).

I mean no disrespect to sales folks (I consider them a vital necessity), but salespeople will almost always have a slanted view of their product/service, so you need to listen to all of them to get a better overall picture.  If you listen to only the first salesman, then you may pay for a dump truck when all you needed was a pickup, but the final result will actually be a shovel.

Finally, try to locate any user forums (like this list) for the products/services you are looking at to see what info you can find.

01:45 PM in Migration, User Reports | Permalink | Comments (0) | TrackBack

March 17, 2006

Listing of migration experiences produces Eloquence

During a fascinating week up on the 3000-L newsgroup, advice poured out this month from almost 20 customers, consultants and vendors about migration. In 40 messages, the 3000 community replied to Connie Sellitto's request from the Cat Fanciers Association:

CFA’s Board of Directors have expressed an interest in moving our main operations off the HP 3000 within the next 6-12  months. We have 12 TurboIMAGE databases, over 200 COBOL programs, all in-house written with source available. The few SQL-called routines were replaced with COBOL a few years ago.

Advice, suggestions, condolences — welcome!

Advice from the community rallied around Sellitto's note that "We also have a PC network with Web and SQL servers — could these be upgraded to handle a converted (Eloquence) database?" Eloquence received praise from its users who'd replaced apps using IMAGE databases with the Marxmeier Linux-Unix-Windows toolset. But the fact that the advice centered on choosing a new database reflects the primary concern for any HP 3000 shop migrating away from in-house applications.

After a couple of dozen reports arrived, Sellitto, who had been firmly in the homesteading camp before her board of directors had their say, echoed the database importance:

The comments from the List reinforce my belief that it’s more about the database than the hardware or even the application software. Our IMAGE databases are so designed as to give great access times, except when we’re trying to look for a value imbeded within a non-indexed field.

Tony Summers of Smith & Williamson Corporate Finance Ltd. saw the database choice as a simple one. "You could take the opportunity to use migration to convert to a RDBMS  — and as some vendors have stated  — in a way that is transparent to the application. However, I would advise against this approach  — if you’re using IMAGE, then use Eloquence. If you’re using KSAM, use an ISAM file system."

Summers was also frank about the amount of integration that a long-time HP 3000 shop will have to shoulder when moving off of the platform:

The first step is to take a rain check on your current applications  — and look at how closely wedded they are to MPE and/or how much interaction they have with other servers. Do you use lots of JCL, or direct use of HP intrinsics in the application?  How easy would it be to replace your screen handling with a new layer?

Do not underestimate the steep learning curve you’ll need even once you’ve chosen what appears to be a suitable set of products and hardware/operating system. Keep reminding yourself that with the HP 3000 you usually had all layers of the application in one box and supplied by one vendor: user interface layer (VPlus) application (COBOL) and database layer (IMAGE/KSAM).

One seasoned 3000 professional after another recommended Eloquence as a solution for the 3000 shop that's been built around IMAGE. When Sellitto mentioned that the CFA is already using an SQL database on some Web projects, Lars Appel pointed out a specific Eloquence feature that might be useful to populate another database.

Your “subset” example somehow reminds me that, at least in some use cases, it might be feasible to “replicate” selected parts of an Eloquence database to another type of data store by using the “forward logging” and “auditing” features...


The “forward logging” (somewhat similar to TurboIMAGE logging) captures database changes to logfiles, which can also be enabled for use in conjunction with the fwaudit utility, which in turn allows to extract (selected) subsets of the changes for review or further processing.

This could be used to populate/feed/update another database, SQL or whatever. Not a “real time replication”, but depending on the setup (eg frequency of logfile switches and extracts) something like once per day or once per hour sounds feasible.

OpenMPE board member Donna Garverick added that Eloquence puts familiar faces on the change to Unix or Windows. "I think for many MPE shops, the move to Eloquence makes a lot of sense," she said. "It mimimizes risk by letting the programming staff focus on actually getting the migration accomplished. Eloquence gives a wee bit of familiarity in a strange land."

06:41 PM in Migration, Web Resources | Permalink | Comments (0) | TrackBack

March 16, 2006

Making a case for change, sooner or later

Posteraerialgif_1 For many years Wirt Atmar, the founder of software supplier AICS Research, worked as a persistent advocate for the HP 3000's future. In 1996 Atmar organized The World's Largest Poster Project, a demonstration of what HP 3000s could do as well or better than any other computer. Atmar used an HP 3000 and an HP color printer to produce a poster that covered an entire football field in Southern California, bragging that MPE Users Kick Butt.

Atmar remains a resource for his HP 3000 customers; his company wrote a free terminal emulator, QCTerm, which can be had for just the bandwidth of a download. But this veteran of more than four decades in the computer business faces the future better than many of the 3000's more ardent fans. He weighed in recently on a recommended set of actions for the 3000 owner as well as for its administrators. While the 3000's value was extraordinary for many years, Atmar believes the magic has gone. He doesn't believe in migration of applications, however. That's not to say that 3000 owners and managers should stay put.

I strongly argued for the continuation of the HP 3000 for as long as I could  simply because it worked so well for our customers, but once HP killed the HP 3000, I immediately changed my recommendations. What I now tell our customers is that if you’re the business owner/manager, you will want to stay on the HP 3000 as long as you can, until you find something similar enough that you can  purchase and move your business operations onto with some ease.

But don’t try to  migrate your old application off onto something else. That’s a death trap, and it’s not worth it.

Atmar notes that developing new skills is essential for the 3000 professional. "But what also I tell our customers is that if you’re the data processing manager, you will want to get off of the HP 3000 as quickly as you can and develop your skills on something else. Computers are computers, and there’s nothing exceptionally magical about any of them."

He adds that calling migration of applications off the 3000 "a death trap" deserves about 10 paragraphs of explanation. He's willing to get specific about why, and offer up those paragraphs, if only one of our readers will ask.

04:31 PM in Migration, Newsmakers | Permalink | Comments (0) | TrackBack

March 15, 2006

RUG meet to connect ERP sites

Mark off the first Friday in April for a meeting of ERP customers, including those who want to remain on their HP 3000s as well as migrating sites. The Computer Aided Manufacturing User Society (CAMUS) will host regional meetings April 7 in Austin, Boston, Frankfort, Ind., Fremont, Calif. and Irvine, Calif. These locations will hook up with a five-way conference call at 11 AM PST to discuss the state of migrating off HP 3000 ERP applications, as well as options for staying put with a working application.

The two choices mirror the theme for this year's CAMUS annual user group meeting, now coming to San Francisco in July. Terry Floyd, the founder of ERP support and integration company the Support Group, said CAMUS wants to talk with 3000 ERP users, no matter which direction they're headed.

"If you want to homestead, we want to help you talk to the right people," said Floyd, who became a CAMUS board director last year. "If you want to migrate, we can help you, too." RUG meeting registration details, along with a form to sign up, are online at the CAMUS site.

Next month's RUG meeting will include discussion from the CAMUS board about the July conference. Floyd said that vendors of MANMAN alternatives might be helping to fund that summertime conference. "I wouldn't be surprised if there was a big banner over the conference that says, 'Sponsored by SAP.' People are going to that conference to talk about leaving MANMAN. There will be a lot of emphasis on migration."

On the other hand, Floyd says, an application like MANMAN represents the genuine value to a company using it, not the vendor's support of the hardware. "The application and its usefulness for your company is what determines value, not the hardware," he said. "Nobody can come up with a good reason why the application won't be running in 2012."

Floyd's business in the 3000 marketplace these days looks like that of many long-time vendors: Helping along customers who can't leave the platform, as well as serving up integration and consulting to enable migrations through his allied company ENTSGO. HP's extension of its support services through 2008 gives a 3000 MANMAN customer ample time to migrate, he said. "24 months is plenty of time," he said. "Almost anyone can get off MANMAN in 12 to 18 months."

With a little help, he added, a well-versed company could manage their own migration project. "It's nice to be self-sufficient," he said. "You should get some help, but if you're up to speed and reading what out there, you can do it yourself."

The Support Group will be assisting Lowrance Electronics in moving off of MANMAN, Floyd announced, and long-time MANMAN site Korry Electronics, a Top 10 MANMAN user, will also be leaving the application behind. "The other apps have finally caught up with MANMAN," Floyd said, "and now all the stuff the alternatives do is better."

CAMUS could be on its way to electronic-only status as a user group after this year, he added. Making a case for a CAMUS show in 2007 will depend on the attendence for this summer's conference. "There really isn't going to be a MANMAN conference in 2007, if people don't come this year," he said.

09:14 PM in Homesteading, Migration, Newsmakers | Permalink | Comments (0) | TrackBack

March 14, 2006

Opening up vote totals, then wallets

In our weekly podcast (5 MB MP3 file), we talk about the tempest over the details of the OpenMPE election results — and then move on to the more important matters ahead for OpenMPE. Putting the election results out quickly was a good move. Now the group needs to move on toward getting a budget assembled, to improve its visibility and impact. You might even be asked to pay to vote next year.

05:51 PM in Homesteading, Migration, Podcasts | Permalink | Comments (0) | TrackBack

March 13, 2006

OpenMPE opens up vote details

After a little prodding from a few HP 3000 customers and consultants, OpenMPE has voted to release the candidate-specific details of its 2006 election. The voting put two new directors onto the advocacy group's board of directors, a set of volunteers who do just about all of the work for the four-year-old organization. Three of the four incumbents earned back their seats in the election, which drew a record number of ballots.

Six volunteers ran for the five board seats, with the top five vote-getters winning in the election. The results reported by the board, and confirmed in independent observation by The 3000 NewsWire:

111 people voted
492 total votes were cast

Donna Garverick: 101
John Wolff: 91
Matthew Perdue: 83
Bill Lancaster: 82
Jennifer Fisher: 69
Steve Suraci: 66

John Wolff, the vice-chair of the Board of Directors, explained how the voting process works.

"When a ballot for OpenMPE Directors is cast on the OpenMPE.org Web site, the ballot data is automatically sent to three e-mail address at once: to the Corporate Secretary, to the Webmaster, and to the independent election observer. At no time do the ballots pass through any one person's hands."

"After the polls close, the results are separately counted by the Corporate Secretary as well as the independent election observer. Then they are compared. This year there was complete reconciliation between the two counts. The Webmaster also concurred with the results. The Board then meets and ratifies the results. Winning and losing candidates are then notified of the results by the Corporate Secretary."

This year's Webmaster was John Burke, who was on the board but not seeking re-election. The OpenMPE Corporate Secretary is Donna Garverick, one of the candidates in this year's election. The calls for opening up the election details came from Joe Dolliver, Craig Lalley, Jim Alexander and Martin Vaughn. 

We watched the ballots appear in our in-box here at the NewsWire, one by one. Nothing looked amiss to us, although the race for the fifth place was very close. We're of the opinion that nothing about OpenMPE should be withheld from the members or the 3000 community, at least within the realm of what the organization can control. (HP has its own idea about confidential discussions with the group, and HP gets to set the rules there.)

Personnel matters, one sensible off-the-record exception, haven't been an issue for the group yet — because no one except a CPA firm has ever been paid.

Money could turn up as an issue in the future of OpenMPE, a group that has been lightly-funded ever since its inception. In order to ensure the advocacy work continues, membership might rise from free to a $100 individual membership, according to one director. That is an issue that's still to be discussed by the new board, along with what OpenMPE might do to earn that $100. Some customers believe the group has already earned its keep by prodding HP to work out end-game details for the 3000 — decisions the vendor had not considered until the board of volunteers cast its light upon them.

04:56 PM in Homesteading, Migration, Newsmakers | Permalink | Comments (0) | TrackBack

March 10, 2006

HP struggles to keep up

In December HP made some commitments to its HP 3000 customers, but those like Paul Edwards are still waiting to see the vendor's work surface. Edwards, part of the OpenMPE advocacy group, discovered last year that his Certified HP Professional status had evaporated — at least for his MPE/iX certification. On Edwards' advice, HP poked around and learned that the HP 3000 certifications had been retired from HP's program.

Edwards believes that HP 3000 pros deserve a certification as long as HP 3000s are running, not as long as HP wants to fund its certification program. HP agreed in its December 20 announcement that extended HP support for two extra years.

The vendor has been struggling with making good on this promise, however. It's well past two months after HP's agreement to reinstate the certifications, but Edwards and others can't get back into the certification Web site.

HP's promise from Dec. 20 reads:

The HP partner community remains successfully engaged in meeting the needs of HP e3000 customers. In response to a request from that community, HP will continue to recognize the certification for those users and partners who currently hold an existing HP e3000 Certification, enabling them to continue using that certification in the conduct of their business.

Edwards reported last week, "I still have not received any communication from HP that verifies this or explains the details. I still can’t get on the certification site as a certified consultant."

It's too easy to read meaning into these kinds of delays about HP promises to the 3000 community. It's not a secret that HP is paring back the resources for the platform, even if the staff remains on the HP payroll, doing other tasks.

It may seem a small thing, a certification on a platform whose HP lifespan is ending. But the IT pros who supported HP and the 3000 deserve a better end-game than waiting through such delays to get back what they never should have lost.

HP has cut off the 3000 certified professionals so completely they can't access much of the certification Web site at HP with their 3000 IDs. Since it looked like HP's certification efforts for 3000 skills are over, Edwards wanted to take on the 3000 certification program, if HP will permit it.

For now, HP appears outmatched to keep up with some of its 3000 commitments. The IT pros can neither use their HP certification services, or rely on Edwards and his HP 3000 training venture to fill the gap.

08:55 AM in Homesteading, News Outta HP | Permalink | Comments (0) | TrackBack

March 09, 2006

OpenMPE board may consider opening vote totals

In the wake of its most popular election, OpenMPE has gotten a request to reveal the specific vote totals from the balloting which ended last Friday. One board member has replied, adding that the group may bring up the matter at this week's meeting.

Any restraint of information in the HP 3000 marketplace comes under scrutiny these days, particularly from OpenMPE. The advocacy group holds a series of telephone talks with HP's 3000 group — some of the most complete communication HP is sharing about the future of its 3000 operations. But OpenMPE had to agree, much to the displeasure of customers and former board members, to hold those talks confidential. HP wouldn't discuss its matters with OpenMPE otherwise.

If there's an issue with secrecy here, the problem doesn't appear to stem from OpenMPE. HP is the party that's insisting on being able to talk off the record. OpenMPE, like many an HP 3000 customer, hasn't much leverage to insist otherwise.

The 2006 election drew a record number of ballots, but members were voting for five directors among a six-candidate field. We hope the results from 111 voters don't call to mind hanging chads, denial of voting through inept registration, or biased election judges. (Especially the last one, since I was the only outside observer of the vote.)

But specific candidate vote totals from these elections — which would show how close the balloting really was — have been closed to the OpenMPE membership. Some board members believe that's been a policy of courtesy to the losing candidates.

Without giving out this information, it's just one more thing that a critic of OpenMPE can use to claim the advocacy group isn't open. Having seen the results in detail, but been asked to keep them to myself for now, I can say it doesn't seem like there's anything to hide here.

John Wolff, the vice-chairman of the group, reported to me in an e-mail that the group might revisit the election details question this week.

Our Board met on Monday to ratify the election results. This question came up and historically we have not released the individual vote totals for each candidate, just the results of who was elected.  I am not completely sure why we don't do it; perhaps to not embarass some individuals.  Of course, we would never release who voted for whom.  Since this was a Board-level decision, I do not want to authorize the release by myself. Personally, I am fine with it either way.  Maybe we could revisit the question at our next meeting.

OpenMPE has only had four elections, counting this year's. We've been the observer for three, here at the NewsWire. None of our observed elections included a report on specific vote totals. Perhaps this is one user request OpenMPE can deliver quickly, without putting it through the HP wayback machine.

07:16 AM in Homesteading, Migration, Newsmakers | Permalink | Comments (0) | TrackBack

March 08, 2006

Paying a CEO to play, not to go away

A pair of large HP shareholders sued HP yesterday, trying to get back part of the cash that Hewlett-Packard's board of directors paid former CEO Carly Fiorina to leave the company in 2005. The legal action points out a proper difference between on the job compensation versus the pay to go away. The 3000 customers who stick with HP over the next three years, following the vendor down the migration path, should consider what kind of pay policies get funded by system purchases.

In the case of Fiorina, an executive who demanded faster growth than the 3000 sector could deliver nearing its third decade of earnings for HP, her exit pay might have exceeded HP limits. A major shareholder, the Service Employees International Union, had sponsored a shareholder proposal to limit these golden parachutes to not more than about three times a CEO's annual pay plus bonuses.

The proposal become part of HP's bylaws after HP paid Michael Capellas, the CEO of Compaq who helped engineer HP's takeover of that company, a cool $17 million on his exit. Fiorina got a package worth more than $22 million. It might rise to as much as $42 million, now that HP's stock is on its way back up from the record low prices of Fiorina's reign. In a bit of irony, new CEO Mark Hurd is helping to feather his predecessor's retirement nest.

And it's not like the shareholders are asking HP to adopt Google's thinking on compensation. Google pays execs Sergey Brin, Larry Page, and Eric Schmidt just $1 a year. They rely on the company's share price for their compensation, rather than millions a year and even more on exit.

This kind of corporate gratitude, which comes right off HP's bottom line, begins to matter when the company needs to cut back — as HP has done relentlessly since Hurd took his job just about one year ago. Among other things, HP is dialing back its investment in the HP 3000 support and development operations while this corporate compensation is striking new high-water marks.

Whether these two items are related depends on the customer's view of accounting checks and balances. Hurd's compensation for the first seven months on the job actually exceeded Fiorina's fiscal 2005 pay. HP paid Hurd $5.95 million: a $5.13 million signing bonus and $816,667 in salary for April through October. If you add options to the package, it topped $24 million.

Nobody's complaining about the CEO's pay on the job now, as Hurd earns that paycheck by delivering a better HP stock price as well as revenue growth to match a rise in earnings. He gets to stand at the head of the HP chain of leadership to announce last week that Itanium now has 7,100 applications ready for the chip that's at the heart of HP-UX futures, "more than twice as many as where we were at this time last year."

At this time last year Hurd was being wooed by HP, which had already paid out to terminate Fiorina. Those shareholders are trying to ensure more of the company's money stays with executives inside HP, instead of departing in a golden parachute.

HP has places to spend millions of dollars, spots that could affect the value of any platform that relies on Itanium or Integrity servers. "HP and Intel are co-funding specific, go-to-market initiatives with key ISVs in targeted vertical markets," Hurd said in last week's Webcast. He promised more details on these initiatives in the weeks to come.

We hope the details on the initiatives include reaching out to the ERP vendor community. At last check, just three apps popped up on the Itanium Solutions Alliance Web site under the heading of "discrete manufacturing." Itanium might have 7,100 apps — but if your vertical has few of them, it might as well be a million or none.

We have a good idea about how HP might fund more of those kinds of initiatives, but HP says that shareholder suit is without merit. The suit seeks to recover the extra money paid to Fiorina. The former CEO is making a living by exalting her HP accomplishments in paid speeches — which might be some evidence that she wouldn't miss the $4 million or so she'd have to give back   

04:40 PM in Migration, News Outta HP, Newsmakers | Permalink | Comments (0) | TrackBack

March 07, 2006

HP, Intel pitch woo to Integrity on Webcast

HP and Intel used their joint Webcast of last Thursday to promote the future of the Itanium processor and HP's Integrity server line, using marketing guru Geoffrey Moore as a softball-tossing host. The show was already fully subscribed over the Web as it began early on Thursday morning California time  — which is either a comment on the popularity of the content or a measure of the width of the purchased Internet band.

Now that the crowds are away, you can still watch the Webcast in a replay from the HP Web site, www.hp.com/go/hplive.

HP-UX users, as well as companies being taken to that environment like Summit and Amisys sites, should care a great deal about the future of Itanium. The processor remains the only long-term platform for HP's Unix, since the PA-RISC chips are being phased out from the HP server line. There's another end-of-life that HP has arranged for its enterprise customers, although the developers say this migration is simple so long as your application doesn't need to access native-level horsepower. Most Integrity customers report the servers deliver serious improvements over their PA-RISC predecessors.

HP's show was touted as a hard look at the enterprise issues, but the questions aired and answered didn't go far beyond what HP has already said. The tough questions, like those the HP 3000 community used to pose at conference roundtables, didn't seem to make the cut in the 45-minute show. And for an event which promised three CEOs, at one of them didn't appear for even three minutes.

Oracle's Larry Ellison checked in from Japan by video link to speak up on the promise of Itanium processors during the Web event. He believes the chip and servers built on it hold great promise for the computer industry and Oracle's database and application businesses. Two minutes and 25 seconds later, Larry was gone.

HP CEO Mark Hurd and Intel CEO Paul Otellini soldiered on without Ellison, led on by the leading questions from Moore. Perhaps being invited to moderate such a discussion includes some implicit rules about the hardness of the questioning. But the scripting looked obvious in some places. At one point Moore asked a question to which Otellini replied, "Well, it sounds like you answered the question with your own question, but let me spin it a bit."

Hurd said in prepared remarks
that HP is ready to introduce a new Integrity chipset "that delivers up to 30 percent higher capacity, improves memory latency and bandwidth by up to 400 percent." The chipset introduces new availabity features as well. HP's got to introduce something new to cover the spot where they had hoped to roll out Montecito generations of Itanium in the Integrity servers. HP reps were eager for release dates on that processor during last fall's HP Technology Forum. Intel said about a week later that Montecito will be delayed until mid-year. That could push HP's Integrity releases based on Montecito into 2007, unless HP's got a head-start on integrating the unfinished chip.

Near the end, Moore asked about a big share of the enterprise world which will have nothing more to do with Itanium. HP had outlined the Itanium Solutions Alliance, and Moore asked if "IBM will see the light and join the alliance?" Hurd chuckled and said, "I think that's a much better question for them than for us."

Intel said it's investing more in Itanium, not less. They added that they see great momentum in the market toward Itanium adoption. HP now accounts for about 65 percent of the Itanium system sales. HP said that Itanium is essential to the vendor's enterprise strategy.

The questions got even softer toward the end of the program, some evidence that whatever hard ones were sent might not have passed through the Internet gateways intact. Near the close, a question was posed that read, "What are some of the other major platform trends that are driving your computer investments?" Perhaps a customer or an analyst out there really pitched that one, since it fishes for something off topic. HP and Intel, two companies both doing a lot more business outside of the Itanium space, stayed on topic about their Itanium course throughout the Webcast.

05:08 PM in Migration, News Outta HP | Permalink | Comments (0) | TrackBack

March 06, 2006

OpenMPE vote sets record, installs 2 new directors

The OpenMPE board of directors certified election results this afternoon for the 2006 vote, a three-week campaign that drew a record number of ballots and elected new directors Bill Lancaster and Jennifer Fisher. Considering that homesteading customers have ever-less to do with HP these days, the turnout surprised us here at the NewsWire. Like in the past two elections, we got to watch the ballots come in, acting as independent judge. OpenMPE is all about interface with HP. We wondered why more people than ever who are staying with the 3000 seemed to care.

While we agreed to keep the vote totals for each candidate to ourselves, we can say the election was close. There were 111 ballots cast, out of a membership of 333 for the group. That 33 percent turnout is something that Interex user group elections didn't come close to achieving — at least not in the last 10 years of the group's history before it went out of business.

Elections for five spots with six candidates running aren't exactly dramatic races; only Steve Suraci, a board member running for a second term somewhat reluctantly, didn't retake a seat this time. Suraci, who runs the Pivital Solutions third party support company, put two years of service in for the volunteer group, which thanked him and departing director John Burke. Burke didn't run for another term. Both Burke and Suraci have been vocal about the depth of relationship that OpenMPE has with HP. But as an MPE consultant and a support provider, they represented important constituencies for all of the 3000 customers — both those homesteading, as well as the migrating market.

Those profiles can come from the same company, given enough time. That's why OpenMPE matters to migration, in our opinion. (We'll have more on that later this week.)

A look at the incoming directors tends to prove this point. Fisher works at Speedware, one of the North American HP Platinum migration partners. Lancaster's firm, Lund Performance Solutions, is also in the migration business, another of those Platinum partners. Lund has also devoted its resources to a homesteading business, working as a partner with the Resource 3000 homestead alliance.

Then there's Birket Foster, the longtime chair of the group, who also runs an HP Platinum migration business in MB Foster. The fact that these professionals would devote their time to OpenMPE shows that the future of the 3000 is still in play — at least its HP end-game. Migration companies and homestead firms both have an interest in the near term of the 3000 customer.

For the record, the other board members winning a new term this year were Matthew Purdue, Donna Garverick and John Wolff. They join Foster, Lancaster, Fisher, Paul Edwards, Alan Tibbets and Chuck Ciesinski.

You can read more about the newest directors in their election profiles at the OpenMPE Web site. Lancaster did yeoman work in the late 90s and in 2000 to help the Interex e3000 Solutions Symposium get launched. Fisher has become important to Speedware's push in both the homestead market, where the company maintains its development tools, as well in the migration consultancy and tools offerings.

Service on the board can amount to several hours a week of prep and discussion on a conference call, or more if you're tasked with developing some policy or process. A lot of the work is getting to understand what HP is thinking, and trying to make a vendor act with dwindling resources devoted to MPE matters.

This year's turnout beat the 2004 results of 103 ballots, perhaps not sizzling numbers compared to the roll call of Unix and Windows users. But by our estimates, the 111 votes is probably about one-third of the number of migrations finished or in play by the leading migration suppliers. (That's not an easy number to estimate, because it reflects a lot of tire kicking and analysis by a user community just ready to roll up its sleeves this year.)

That number of migrations doesn't include self-migrators, the DIY folks of the community. Those companies include sites which are migrating at their own pace, not on HP's support schedule. They will need some assurance that MPE/iX can keep working beyond HP's plans (now the end of 2008) unless their migrations are quick, or well under way. OpenMPE can offer an advocacy rally point for a lot of the community — so long as HP is willing to keep acting after it listens to the board.

04:42 PM in Homesteading, Migration, Newsmakers | Permalink | Comments (1) | TrackBack

March 03, 2006

Be a part of a swelling turnout

The 3000 NewsWire is the official election observer for this week's OpenMPE election. As I write this, with less than four hours to go until the polls close, this year's voter turnout is setting a record for participation. We take this as a sign that customers are ready to involve OpenMPE in their transition plans — whether to rely on the organization to advocate a better end-game for those who are migrating, or enhance the platform for those who will homestead for at least several years more.

There is still time to have a say in the group's leadership. Six candidates are running for five seats; there are four incumbents. You must be an OpenMPE member to vote — that's free, and you can do it online quickly from the OpenMPE Web site.

Once you have your membership number in hand, then you can cast your ballot, until 8 PM EST, at the following Web address:


Look up your membership number here if you're already a member.


There are more than 330 members already. We'll have results on Monday, once the board has certified them.

03:18 PM in Homesteading, Newsmakers | Permalink | Comments (0) | TrackBack

March 02, 2006

CAMUS to update ERP users in April

The Computer Aided Manufacturing User Society (CAMUS) will host regional meetings in April, a way to connect with colleagues who manage manufacturing applications. CAMUS has a heavy tilt toward the MANMAN shop; several hundred companies, at a minimum, still rely on this application, often hosted on HP 3000s.

Terry Floyd, founder of the Support Group, is among the best in the world at tracking the needs and size of the MANMAN community. Now he's joined the CAMUS board of directors, so he has an even keener interest in gathering users who might be migrating from their 3000s, or trying to stay within budget and stay put.

Those users can attend regional meetings on Friday, April 7. RUG Day will take place in meetings in Frankfort, IN (CAMUS Midwest), Austin, TX (CAMUS SouthCentral), Boston, MA (CAMUS Northeast), Fremont, CA (CAMUS Northwest), and Orange County, CA (CAMUS Southwest).  "This may be the last year that CAMUS sponsors RUG events," a CAMUS announcement warned, "so it could be your last chance to meet with some of the MANMAN, MK and Maxcim users in your area." You can register online.

Floyd agrees that 2006 is a crucial year for ERP customers to decide if attending a user group meeting is a priority or not.

a cHe's not predicting that CAMUS will go the way of Interex, unable to pay its bills and forced to close up. "We think CAMUS has the financial viability to survive for at least five more years," he said. "Even with dwindling membership. I was glad to get inside of the books and see that we're not in any trouble at all."

But national meetings have been lucky to draw 500 users, "about half vendors and half users," Floyd said. "That's not the kind of base you want to build upon." CAMUS has between 400 and 500 users on its membership rosters, counting those who have gone inactive. Not all are 3000 sites; the OpenVMS world has a good installed base of those ERP apps. But a big chunk of that membership is using MPE/iX.

The user goup's announcement reported that RUG Day will include "a one-hour conference call between the five physical sites, which will include an address from Malcolm Miller, President of CAMUS, as well as presentations on the CAMUS Board’s intentions for 2006 and beyond. Learn about plans for the First Annual 'LAST CAMUS Conference' this summer in Northern California. After the successful 2005 CAMUS Conference, the theme will again be Homesteading and Migrating — because remaining users of MANMAN, MK, and Maxcim are about equally divided between the two alternatives and experience is growing with both."

11:13 AM in Homesteading, Migration, Newsmakers | Permalink | Comments (0) | TrackBack

Sure enough, it's full up on the Web

Interest must be high in HP's Webcast today featuring CEO Mark Hurd, Intel CEOPaul Otellini and Oracle CEO Larry Ellison. The Feed Room (a service HP is using for the Webcast) reports right now that it's unavailable. Keep trying, if you're interested, however; people bail out of this sort of thing, and the Q&A from Hurd doesn't begin until 12:15 Eastern (US ) time.

10:46 AM in News Outta HP, Newsmakers, Web Resources | Permalink | Comments (0) | TrackBack

March 01, 2006

HP tries to make waves over the Web

Ceos_2 Tomorrow morning (well, 8:30 Pacific US time, 16:30 GMT) HP will hold a Web seminar featuring its CEO, the head of Oracle and the CEO of Intel. As if that weren't enough computer business firepower, the show will be hosted by Geoffrey Moore, author of the seminal marketing text Crossing the Chasm, the book whose descriptions included Late Mainstreet, to sum up the HP 3000 position as of the late 1990s.

In order of cumulative exposure in the industry, you'd probably rank Oracle's Larry Ellison No. 1, Moore second, and Intel's Paul Otellini third. Newest to the power circle is Hurd, who picks his public appearance carefully. HP's not saying very much about the content of this Web broadcast except to describe it as tougher in tone that what you might expect. HP says it will be:

A compelling interchange on the future of enterprise computing streamed live over the World Wide Web. You won't want to miss it as these leading executives take enterprise computing issues head-on.

You will hear from three CEOs whose companies had combined 2005 revenues of more than $135 billion. What might be more worthwhile than the 40 minutes they speak, however, will be the postscript to the presentations: Mark Hurd taking questions over the Internet customers and partners who are watching. Well, a quarter-hour of them, anyway.

HP is streaming the event without any registration required, as far as we can tell. Add up the number of customers touched by these companies' products; you might get quite a traffic jam-up. Or not, depending on how the markets feel about such an event. It might be something like the "interview" Hurd did in Orlando at HP's Technology Forum. On that stage he handled questions lobbed from a key marketing executive VP. It was interesting to see him in action, but surprises were not part of the agenda.

They may not be on tomorrow's menu, either. But just about anybody could work their way to the top of the question queue; it's more of a lottery than the line-up which talk-radio producers use to get callers on the air.

There really are not any homesteading issues that a CEO would address in a conference like this, but there could be some insights on how HP is going to step up adoption of its Integrity/Itanium line. This might not be so important to the 3000 customer just stepping into migration today; several years could go by before they need to commit to Itanium hardware. But to Itanium they must go, eventually.

03:39 PM in Migration, News Outta HP | Permalink | Comments (0) | TrackBack