December 12, 2018

Source code for MPE/iX: Security, by now

Blanket-Ad
Ten years ago this week the 3000 community was in a state of anticipation about MPE/iX. HP had an offer it was preparing that would give select vendors the right to use the operating system code. The vendors would have a reference-use-only license agreement for MPE/iX. No one knew whether the source would have any value, said Adager CEO Rene Woc.

Adager, the company whose 3000 products are so omnipresent they held a spot on the Hewlett-Packard corporate price list, believed there was potential for independent support and development vendors. What was far less certain was how far HP would let source go to solve problems for the 3000 community.

"Source code is important whenever these kinds of [vendors] have support from HP, which most of them do," he said in that month of 2008. But HP engineers can look at source, just as third parties will do, "and the answers won't come instantaneously. In the meantime, you have to get your business back on track, and I think that's what the customer is eventually interested in. It will be nice to have that additional [source code] resource — especially in the sense that it will not be lost to the community."

There was a chance that HP's source licensing terms would be too restrictive, "to the point where you say that you are better off not knowing, because then we're free to use all the methods we've worked with while we didn't have source." After getting a license to source, Woc added, "you might have to prove that you got your knowledge through a difference source than HP's source code. We will see."

That sort of proof has never been required. Not in a public display, at least. Source code, held by vendors such as Pivital Solutions and others, has been a useful component in workarounds and fixes. HP never gave the community the right to modify MPE/iX. This turned out to be a good thing, as it kept the 3000s stable and made support a manageable business for application vendors.

There was also the wisdom that the resource of HP's code would have to prove itself. At least it held a chance for rescue and repair.

The source code "is probably a security blanket," Woc said in 2008. "In that respect, it's good that it will be available, that they're starting to offer some things. We'll have to see what kind of conditions HP will offer in their license agreements." 

Having source access though a license did not automatically make license holders better providers of products and services, he added. "You cannot assume, even with good source code readers, that the solutions will pop up," he said. "A lot of the problems we see these days are due to interactions between products. So the benefit for the customer would be based more on the troubleshooting skills that an organization can provide."

"The basic resources [of source] won't make things better by themselves," Woc said. "It's a matter of troubleshooting." 

Posted by Ron Seybold at 02:18 PM in History, News Outta HP | Permalink | Comments (0)

Pivital Solutions: Your complete
HP 3000 resource

December 10, 2018

HP's 3000 boxes step closer to solid storage

SCSI2SD-V6-RevF-2T
Almost two years ago, an expert in HP's 3000 systems was working to use solid state disks (SSD) with the computer. John Zoltak was trying to link the server to microSD cards late in 2016. He checked in with us this week to report success on the project.

SSD on 3000 hardware from HP has been a dream for several decades. Imperial Computer had a solid state unit early in the 1990s that held a promise of faster IO transfer on MPE/iX. The cost was astounding compared to moving media and the capacity was a fraction of spinning disk drives'. Much later, SSD has become something of a desktop standard and is an active choice in enterprise servers, too.

The MPE/iX hardware from HP -- to us, something called an HP 3000 -- wanted to play from SSD, too. In his prior report, Zoltak was trying to copy one 917LX disk to a new disk on the server's SCSI bus. A 4GB drive is standard on a 917, so just about any microSD card would match that storage. Now there's a V6 edition of SCSI2SD, a combination of hardware and software that delivers SD storage to HP's 3000 iron.

The combination now works beautifully, said Zoltak, who's working at Fives North American Combustion in Cleveland, Ohio. "You want the V6 boards," he said. "The V5's are much slower. The V6 takes a full size SSD card and up to 128GB has been tested." Michael McMaster, the inventor based in Australia, has engineered the latest version of his product "as a complete redesign for the V6 boards, which use a completely different microcontroller." The device is for sale online at Intertial Computing. Today's price is $105 including 16GB of microSD.

The product employs a SCSI-2 Narrow 8-bit 50-pin connector. It does SCSI FAST10 synchronous transfers at 10MB/second. Zoltak is reaching way back into the HP 3000 hardware closet to test. He's attached the SCSI2SD to a Series 917.

"I have the board sitting on top of the system with a cable around to the back on the same SCSI as the 917's DAT and DLT drives. I did a reconfigure and a restore to the SSD. Seems to be fairly quick. While restore was running I used HP Glance and saw that the disk was doing about 65-70 IO's per second. This is not as fast as the Nike array it came off of, but then it was on a differential wide SCSI."

The bigger benefit is that the HP MPE/iX iron can rely on SSD instead of moving media. Disks are among the leading culprits in HP's 3000 failures in 2018. Tape is a close second. Storing and moving bits gets complicated while using the hardware that HP certified for storage with 3000s than a decade ago.

Newer storage reduces the risk of homesteading. This is one of the benefits of using a virtualized 3000, too.

Read "HP's 3000 boxes step closer to solid storage" in full

Posted by Ron Seybold at 01:44 PM in Homesteading | Permalink | Comments (0)

December 07, 2018

Memory and Disk Rules for Performance

Concentration
NewsWire Classic

By Jeff Kubler

You need to get management support for your efforts to keep your systems performing at their best. Memory and disk are two components of your performance picture under MPE/iX. Main Memory is the scratch pad for all the work that the CPU performs. Every item of data that the CPU needs to perform calculations on or updating to must be brought into Main Memory.

CPU used to manage Main Memory: The CPU must manage memory. It must cycle through the memory pages, marking some as Overlay Candidates (this means that new data from disk may be placed here), noting that some are in continued use, and swapping others out to virtual or what is called transient storage. Swapping to disk occurs when data is in continued use but a higher priority process needs room for its data. To accommodate this higher priority process and its need for memory space, the Memory Manager will swap the memory for the lower priority process out to disk. The more activity the Memory Manager performs, the more CPU it takes to do this. Therefore it is the percentage of CPU used to manage memory that we use as a measurement.

Page Faults per Second: A Page Fault occurs each time a memory object is not found in memory. The threshold for the number of Page Faults per second that can be incurred before a memory problem is indicated varies with the size and the power of the CPU. Larger machines can handle more Page Faults per second while a smaller box will encounter problems with far fewer.

An exceptional number of Page Faults should never be used as the sole indicator of memory problems but when observed should be tested with the memory manager percentage. If both agree, you have a memory shortage. There are some strange things that I have observed with Page Faults, so it does not stand alone as an indicator of memory shortage.

The number of Page Faults per second and the amount of CPU needed to manage Memory are always evaluated in conjunction with each other. That is to say the high Page Fault Rate will not be considered a problem if the Memory Manager Percentage is not above 4 percent.

Read "Memory and Disk Rules for Performance" in full

Posted by Ron Seybold at 02:10 PM in Hidden Value, Homesteading | Permalink | Comments (0)

December 05, 2018

One Alternative to $1 Million of 3000 Costs

Charon Portfolio
In a webinar this week, Stromasys made its case for how shutting down HP's 3000 hardware can reduce an IT budget. Using data from Gartner analysts and other sources, the company estimates that downtime costs companies $1 million per year on average. Any alternative to 15- to 25-year-old servers is a good shot at making the future more stable.

There still hasn't been a computer system built that will never fail. Hot-swap backups with automatic server failovers were never a big part of the 3000 datacenter experience. If you had to handicap which server was a likely failure candidate, HP's MPE/iX hardware would give you short odds of failure. In this case, short is not a good measure.

One million per year in losses is a big enough number to get the attention of a corporation's C-level. It's the same number, coincidentally, that Stromasys used this week to describe the costs of migrating MPE/iX apps. The text circled in the slide above "implies investments of $1 million+" for migrations.

These millions, lost through downtime or surrendered in datacenter budget, are averages. Smaller 3000 customers may not approach the $1 million in yearly lost revenues. Migration costs track closer to that number, but they're a one-time hit. The alternative is Charon, of course. During the webinar we learned that an additional HP market is coming online to use Charon. HP's Unix PA-RISC servers will be the latest Stromasys virtualization segment, according to Dave Campbell.

Read "One Alternative to $1 Million of 3000 Costs" in full

Posted by Ron Seybold at 12:13 PM in Homesteading, Migration | Permalink | Comments (0)

December 03, 2018

HP 3000 dream tracks close to virtualization

Railroad switches
An HP 9000 HP-UX virtualization product is in development. In that kind of design, a single Intel server with enough computing power (concurrent threads) could host both HP 3000 and HP 9000 virtualizations. HP had the same objective almost 20 years ago for its largest enterprise platforms.

Early in 1999 HP's Harry Sterling spoke at an all-day user meeting in the UK hosted by Riva Systems. Sterling, who'd retire before the end of that year, said a multi-OS server was within HP's vision for the 3000 and 9000 customers.

Sterling’s mentioned the possibility of running MPE, NT and Unix concurrently on the HP 3000 "sometime in the future." There was even the possibility of a “hot-swap” version of MPE alongside the production system. John Dunlop reported for us at the time.

The passing mention indicated that separate processors in one box would be able to run different operating systems. Sterling did suggest that a hot-swap version of MPE might be a valid use, so that there would be some redundancy with the live operating system.

This seemed to lead to the subject of more uptime. From these comments, it’s possible that HP is looking at allowing online changes to a hot-swap system and then just switching it over to achieve the so-called “magic weekend.” This is a system upgrade that occurs seamlessly and transparently to both the users and management.

That would be a dream not realized. Hot-swap didn't make it any further into the customer base than architect discussions. Sterling noted that in 1997 customers expressed concern about the future of the 3000. To counter that feeling and give the customers more confidence, he outlined in 1999 a five-year roadmap for the 3000.

Marketing was on board as well in that year that led to Y2K. It would take another 13 years before a multiple OS host for MPE/iX would emerge.

Read "HP 3000 dream tracks close to virtualization" in full

Posted by Ron Seybold at 01:28 PM in History, Migration, News Outta HP | Permalink | Comments (0)

Search