System iNetwork

We all (well most of us) saw the way IBM marketed Warp and really the lack of any real marketing over the years for the System i, but it's a great idea to market in this way, particularly for the System i, in my opinion. Let's hope IBM picks up on it — it really does have a nice sound to it.

"Possibly Flawed i5 Possibilities"

Any idea that might result in more exposure for i5/OS seems like a good one. We're in the small end of the SMB range, but we have an i520 for our big databases. We also have several rack-mounted xSeries and Dell MS servers (not attached to the i5).

Is it possible that IBM could create a BladeCenter that will allow both xSeries blades (for MS OSes) and Power blades (for i5/OS, AIX, Linux) to coexist? A box(es) like that could ultimately replace all the computing resources in our server room.

IBM will consolidate all xSeries, pSeries and iSeries servers to the BladeCenter brand. This would allow a level of parts and design consolidation that is too attracive for IBM to ignore. The major cost to IBM around iSeries/System i is the cost of engineering new CECs every 3 to 4 years. The BladeCenter S will be the physical packaging of the 515/525 replacement. It will come preconfigured with P6 Blades, Disk and I/O and will plug into the wall and just fire up. If not 2008 then 2009 for sure.

Sounds great, will it be LPAR'able, will it fit in a standard rack and not a special IBM rack? Will it need HMC? Lastly, once you can run an LPAR'd WIN-OS on it, that would even be more fantastic. Sounds like a great solution and quite possibly a $$ earner.

Now that VMware and its competitors are mature, accepted products, what is the point of a blade server? Blades look and sound slick. And as easy as it appears to deploy a blade server, doing the same in on a virtualized rack mounted server is easier and cheaper.

"...The major cost to IBM around iSeries/System i is the cost of engineering new CECs every 3 to 4 years. ..."

Ian Jarman has said there is no technical or performance reason why i5/OS cannot run on the p5. Not sure what a CEC is, but the fact that IBM continues to duplicate its efforts by producing both the i5 and p5 disproves the assertion that a reduction in production cost is important to the wizards of Rochester.

We sold our first IBM Blade Server ERP solution (replacing an AS/400) over a year ago and haven't looked back.

"Ian Jarman has said there is no technical or performance reason why i5/OS cannot run on the p5. Not sure what a CEC is, but the fact that IBM continues to duplicate its efforts by producing both the i5 and p5 disproves the assertion that a reduction in production cost is important to the wizards of Rochester."

Steve, IBM surely has plans that they don't share with someone who does not know what a CEC is. How do you know they are not planning on one platform -- the blade is one platform, and all suggestions are that AIX and i5/OS will run there.

"...Ian Jarman has said there is no technical or performance reason why i5/OS cannot run on the p5...disproves the assertion that a reduction in production cost is important to the wizards of Rochester."

CEC = Central Electronics Complex or the actual base unit. An i5 570 and a p5 570 are physically the same CEC. The control panel and the firmware are different. However, they are the ONLY pieces of the CEC that have any differences between an i5 and a p5.

With BladeCenter, the xSeries, iSeries and pSeries can share power, cooling, networking, IO and management. This would save the cost of having to engineer a new CEC every time there is a processor change. Only the Blade would have to be engineered and even there, there would be a great deal of commonality. I also know for a fact that Austin and Rochester were consolidated to save design costs.

I don't think anybody in the industry is waiting for a System i on a BladeServer form factor. Sure, it may be a challenge for Rochester to prove they can morph the system in any hardware or processor type (except Intel), but first, this is a natural capability of the MI layer architecture of OS/400 so no need to prove it one more time, and second, who in the industry is waiting for that kind of form factor?

Rochester should put some more effort into advancing the native development environment like the RPG ILE and CLP languages, which they have not touched for the past 10 years (free format RPG doesn't count like a major advancement). I mean, developing in Java, WebSphere, PHP, MySQL or even running Linux on this hardware has no purpose whatsoever if you can run it with the same performance and reliabilty on Intel hardware ten times cheaper. Without RPG and CLP, the iSeries has no reason for its existence anymore.

"...With BladeCenter, the xSeries, iSeries and pSeries can share power, cooling, networking, IO and management. ... I also know for a fact that Austin and Rochester were consolidated to save design costs. ..."

You're probably right. (I hope you are!) My point is that back in 2000, the rs6000 model 170 44p was just about identical to the as400 170. Despite that commonality IBM continued with the separate hardware lines. Even when LPAR was introduced and eliminated all technical barriers, the products were kept apart. If the p and i have been kept apart all this time, why will they now be consolidated?

To answer my own question, IBM might think that blades, which compete against partitioned servers, and as such have some fraction of the power of a server, are underpowered enough to not compete against the large i5s that they can still sell for high 6 figures.

"... the blade is one platform, and all suggestions are that AIX and i5/OS will run there. ..."

Trevor,

What might the pricing be? Will the blade hardware cost the same whether AIX or i5/OS runs on it? Will i5/OS be user based priced ($250 per) without the initial $3,000 charge for the first user we see on the i515?

"...Without RPG and CLP, the iSeries has no reason for its existence anymore...."

It still has advantages over Linux. The well organized program call stack, with good exception handling capabilities. Excellent debug facilities. The pre compiler is great -- you can effectively create your own programming language by modifying any of the built-in ILE languages. All you need is fast, low cost hardware that the mediocre IBM managers repeatedly refuse to make available.

"...Without RPG and CLP, the iSeries has no reason for its existence anymore...."

No doubt RPG ILE has advantages over Linux, unless you prefer to develop business applications in C. But IBM's strategy to introduce other technologies without paying much attention to the native ones is bluntly short sighted imho. First, for brand new application development, no one is going to consider running PHP and MySQL on an iSeries box because of its excessive pricing. So those technologies will only come to iSeries if they are used as an add-on to the existing RPG based core applicatons. But many companies are torn between the dillemma of either enriching their RPG core with in house developed additions created in Java, PHP etc. (meaning additional manpower to maintain the heterogeneous environment) *or* buying a complete out of the box new and non-iSeries application solution.

If IBM enriched RPG and gave it client graphical capabilities and seamless integration with MS Office apps (like SAP morphed the ancient Cobol language into the slik ABAP language), the future of iSeries would certainly look less bleak imho.

I do not get it!
I'm old, growed on SEU&RPG. Now I see probably the best dev. environment is .Net. I sow beautifull stuff (.Net) done with the i5. Why not enpower on it? Everybody will keep paying i5, pw1, RPG licensing. It is natural to keep the old stuff for maintenance . . . where is the problem? Horsepower? Come on! Security? It is our business! Or not? MS?....

Well, let's think out of the box a little bit ... or, maybe, really think inside the box (of chips). In point form for clarity:

1. Would i5/OS on Power-Blade be aimed at (by IBM) replacing big central i5 boxes . . . or, even decent-sized smaller ones? NO (in the next 3-5 years)

2. Are not Blades the "physical flip side" of virtualisation? YES - they are physical consolidation (a la "physical virtualisation" reducing complexity & management). The need is in the Intel/Microsoft "space".

3. Might a sophisticated Blade-S of the future, with everything working together, look like anything you might have seen before? Well, I have seen it, it was called the RS/SP with multiple nodes and high speed "back planes". So, that's where Blade-S might go in future, who knows . . . but, it can be done and there's significant experience of it in IBM (you know, the world's largest scientific computers and all that). It's also just a form of "grid computing" - depending on how many blades you string together.

4. IBM is becoming the leader in Blades (marketing) . . . and Power-Blades are doing very well too. Might IBM prefer Power-Blades to Intel-Blades? You bet they would!

5. If IBM could introduce a whole new application portfolio to Power-Blades that were not available to Intel-Blades, would that interest them (to help them sell Power-chips rather than Intel chips)? You bet it would! . . . and, that's where i5/OS and its applications portfolio might come in.

6. As a corollary, might IBM ever put i5/OS on Intel(Blades) - NEVER!

7. The Power-Blade is i) a strategic weapon against Intel (chips) and ii) a strategic weapon against Microsoft (applications) . . . in my humble opinion.

ugeerts is really on the right track. There needs to be easy way to make applications graphical in 'native way' to keep System i really alive.

WebFacing was nice try but not native (and now IBM SWG shot it down with WDHT price tag of 13K EUR on p10!). HATS... it's a toy nobody wants now because it just doesn't work good enough. PHP... great idea but it requires too much work -> takes too long time now. IBM should remember that those RPG applications running in houses are the ones that keep iSeries running and selling...
And in the same time that iron thingie... Blade is a nice concept but the true challenge is the pricing of iSeries... it should be a lot cheaper... 515 is a nice try but (in here it's not 7995 dollars to euros, hell no, it's 7995 euros(!)) it needs to be lot cheaper and the price has to include Application Development stuff and other needed softwares.

So in conclusion: IBM, iSeries needs your effort, not punishments like buggy and stripped WDSc7, mandatory WDHT runtime with incredible price on WebFacing and forced use of WebQuery (informations on WQ are short yet but believe me, it won't be only glory and joy for iSeries platform although product idea is great).

I think IBM, the Publications, COMMON and we, as programmers are part of the sorry state of affairs: the System-i's sharp decline in sales. We are backed into a corner with a system that is a series of archaeological layers.

I agree with Thomas Stockwell that it's a major accomplishment that even System 32 programs with minor changes can run on the native environment.

The problem is too big for any one person or group to get their arms around. Consultation is important but not necessarily consensus on all points.

We need a career and system conversion czar who will put through a rescue plan for most System-i constituencies. What's on our side is that IBM is the biggest server manufacturer! But on the negative side is the perception among key constituencies that the System i world is a patchwork of legacy technologies. These constiuencies are CIO's, Purchasing Managers and the students at Community Colleges and Universities who could be the next generation's decision-makers.

We are operating in an ocean of .NET and truly Object-Oriented program-development. I think that the successor to the System i should connect to EVERYTHING. I strongly believe that IBM should buy the company that bought ASNA and implement RPG.NET as one of the career paths the System i successors should offer. IBM, stop whining about spending money or effort and finish flagship offerings such as WDSc 7.0, DB2 and don't create perceptions that DB2 will not support or vice versa Lotus Domino!

I believe that with this economic slow-down that when the going gets tough, the tough gets going. We need Strong Leadership and VISION not piece-meal crumbs.

Chris,

Please post all of this in your response section. I would appreciate seeing the responses. I am sure many have disagreed with my posts but the very big white elephant is in the room with us. We need action.

Chris, do you?

-John deCoville

"ugeerts is really on the right track. There needs to be easy way to make applications graphical in 'native way' to keep System i really alive."

I would like to disagree with this statement. The only server for which there IS a native GUI is Windows - because that is all the OS is. Mac OS X is a GUI desktop on top of a Unix OS. Gnome and KDE are GUI desktops on top of Linux. X-Windows is a GUI desktop on top of UNIX. Why should System i be any different than any other OS?

On the other hand, the System i already has a 'native' HTTP server - to deploy HTML. It already has 'native' WAS to deploy java applications - and Tomcat is somewhat 'native', in that it runs under i5/OS. The 'native' 5250 data stream can be used by several third party tools that only work on 5250 - which is just System i.

If you are going to bet the future on having a native GUI for System i, what will you do when nothing changes? The answer is, use a third party tool which uses a 'native' System i interface, and move along. Or, write new applications that are true GUI using all the native WDSc tools that you have.

This is all available on the System i today. What are you waiting for?

This is silly, the System i is a server. There is no need for a server to natively support a graphical UI. That is the client's job. Until the System i becomes a client side machine it does not need a graphical UI or the overhead associated with it. BTW, the System i supports every significant client side graphical UI on the planet.

"... I am sure many have disagreed with my posts but the very big white elephant is in the room with us. We need action. ..."

John,
I agree that the average as400 pgmr can be seen as a poor excuse for a professional. Low adoption of sub procedures in RPG. Little knowledge of the power and utility of sql procedures as a programming language. Can't program in PC frameworks like .NET and Java which can be used very effectively to build as400 centered applications.

In the end though, I think this is blaming the victim, with IBM mgmt being the true villain. Arguably, old RPG is the best way to program the as400 because what with the system being overpriced and intentionally slowed down, programs have to be written with CPU efficiency as the top priority. I am currently working on an i520 that IBM has geared down to a 1/4 of its potential speed. Yet the system supports a very large warehouse operation and financial reporting. The SQL code I am write brings the system to its knees. IBM wants likely close to 6 figures to take the governor off and my client is willing to wait until next year. In the meantime, the RPG400 pgmrs are very necessary employees.

-Steve

I looked up Blade servers on IBM's site. I saw nothing coherent on it. If I was a customer I would not have the foggiest idea why I should buy a blade server from IBM with any OS on it.

I take it from above posts that it must be cheaper, hence the SMB talk, and more flexible, given the "load any OS on the same hardware" talk.

My understanding from the past was that the Power chip had an OS/400 side it switched to for OS/400. Is that all gone now with an OS/400 - i5/OS rewrite somewhere along the way? If it isn't, then the CPU has to be the same as in the iSeries.

But I would be surprised if all the engineering of an iSeries system can just be replaced with a "standard" hardware board similar to a PC.

Sure, some kind of iSeries system can probably be made to work from it, but what are we talking about? A market that doesn't even want to step up to the new 515? That doesn't even make any sense to me.

Also, I think the WHDT Webfacing - thin or rich client - technology pricing needs to be compared to Windows alternatives before being dismissed as too expensive.

I would like to see total costs compared from development through run time costs serving customers by those complaining about it before I would accept that Microsoft technology is a more cost effective business infrastructure for an SMB.

rd

ugeerts is really on the right track. There needs to be easy way to make applications graphical in 'native way' to keep System i really alive.

A clarification for above statement is required I think. Some interpreted this like if I proposed to give OS/400 a native GUI interface . This is certainly not the intention, although it would be nice to have an OS that would scale all the way from a client system to a near mainframe/lpar solution system. Windows (still) doesn't scale that high yet but Unix and derivaties like Linux do.

The problems with iSeries at the moment imho are :

1) The system today lacks "prestige". For non iSeries users, the image of OS/400 has detoriated from "prestigious-solid" in the past to "outdated-obsolete" today. Despite IBM's efforts, this detoriation accelerates as we speak.

2) The way IBM/Rochester communicates with his clients about software distribution, licensing, WDSC, fix packages, information center is from 10 years ago and is now barely catching up with the competition.

3) The premium price IBM/Rochesters ask for its systems is about 180 degrees opposed of what SMB's would be willing to offer. Only for large accounts, it is for the time being, still acceptable.

4) IBM's bizarre marketing of this product. No other manufacture would sell a machine with brakes on the wheels turned on all the time. To lift the breaks, you cough up an additional 6 figure amount. I'm referring here to the o/s governor.

5) The system perception by iSeries users (the companies CEO and CIO) is detoriating from "Solid 24/7 dependable" to "still-solid" but with a feeling "we are held hostage by our own RPG legacy code". IBM knows this damned well and exploits this fully to the last drop.

6) IBM may thank SAP on its both knees that the German company allowed to run SAP on iSeries DB2. Almost all top 500 companies who use iSeries on a global scale, are implementing SAP on the same boxes. After the conversion, which takes several years, all that remains for OS/400 is a single task (the SAP kernel) and a single user, leaving OS/400 completely underutilized. If SAP hadn't offered this capability, the problem IBM has now with SMB's, would also exist with the large accounts.

U.Geerts

Steve,

I agree with comment about IBM:

"Arguably, old RPG is the best way to program the AS/400 because what with the system being overpriced and intentionally slowed down; programs have to be written with CPU efficiency as the top priority. I am currently working on an i520 that IBM has geared down to a 1/4 of its potential speed."

I think the debate "Blade or no" needs to be shoved aside for "Overpriced or not?" or "Withholding performance or not?"

IBM, stop holding back on us! Give us "Bull-Necked" speed and plenty of capability! .NET, SQL do add overhead which can be monitored if you give us some more tools as MS has!

Thank you, Steve, for bringing this up.

--John

John, Steve is missing some information when he sprouts his 'wisdom'. We just received the new 515, and there is no limit on the CPU. With maximum memory, it will run at full CPU speed. He, of course, is talking about the 520 -- a server that is not part of IBM's System i future.

Besides, running a CPU at a lower speed was not a problem with the hardware, but a request from the customers. If I had a 170 with 100CPW, why would I need a 3800CPW speed processor? So, IBM carved up the processor, and made a slower version available for a lower price. This was a GOOD thing for all the customers I worked with.

And, since 600CPW was so much more than their 100CPW, there was no thought of writing RPG for CPU efficiency. Relatively, everything was already SO fast, there was no concern.

I think Steve is talking about System i customers who need performance, but deliberately choose the smaller configuration -- then blame it on IBM. Steve likes to blame IBM for a lot of perceived woes, all of them baseless.*

[*Note from Chris: Let's try to keep the posts focused on the topic in the post, rather than passing references to personal experience with others, online or in-person. I'm letting this last sentence from Trevor slide because it illustrates what I'm talking about here. I really don't want to a) get in the middle of personal issues, etc, or b) act as a censor. Either way, anyone's posts are better when they are more specific rather than sweeping and general, and most of us, myself included, have been guilty of a few general toss-offs. Specific to this issue, over the years, I've heard from a number of BPs who sometimes end up with customers who have been 'consulted' into buying an underpowered box for their needs so some consultant/BP could make the sale. But then some companies change their strategy and suddenly need more performance. So yeah, all these scenarios happen. But they also happen the way Trevor put it -- deliberately. Blame can go in a lot of different directions. Personally, I think IBM's issues with "i" most often come back to managing "legacy" revenue, which is a tough job. A lot of us here would happily give up our day jobs and go fishing all day, but there's a problem with retaining our salaries. IBM has the same issue, but that's a whole other complicated topic! Back to Trevor. . . .]

If you look closely at the current System i offerings, you will see there are some VERY fast servers. I think you will find "bull-necked" is thick and slow, and what we have is much much faster than that. Prices are coming down -- and blades will help that. All of the moves from IBM are in the right direction, and they are doing that regardless of the whining.

I say, thanks IBM -- for giving us a server that continues to be more powerful in every generation.

Trevor

"...Personally, I think IBM's issues with "i" most often come back to managing "legacy" revenue, which is a tough job. ..."

Chris,

Back in 2000 the RS6000 and AS400 roughly had the same performance and price. AS400 likely had higher revenue than RS6000. Since then, the p and i have taken two different routes. The i was milked for its legacy revenue and purposefully slowed down while the p was market priced and allowed to run as fast as the technology allowed. The result is the p has, what, 4x the revenue of the i?

I am posting to learn something. Why did the i5 mgmt team that presided over 8 quarters of falling revenue get to keep their jobs? Is IBM aware that its legacy revenue management plan has been and is a failure? Will i5/OS on the blades be priced the same as its system p equivalent - AIX + DB2?

-Steve

[*Note from Chris: I can't say that I understand the inner workings of IBM, but the System p has benefited by having a relatively direct competitor. Because other Unix options were/are available in the market, IBM management had little choice but to attempt to compete on price and performance. And competition usually tends to drive prices down. So while the hardware, development, etc, at IBM appear similar on the surface, the market dynamics are very different. . . . I'm working on an expanded post on this topic, btw.]

"...I can't say that I understand the inner workings of IBM, but the System p has benefited by having a relatively direct competitor. Because other Unix options were/are available in the market, IBM management had little choice but to attempt to compete on price and performance. ..."

Fascinating that the hidden hand of the competitive marketplace is better able, revenue and profit wise, to run a computer company than are System i executives.

Glad you are on the job Chris and look forward to reading you in the future. Would be great if you could track down the creator G. Glenn Henry and ask him what he thinks about being written out of the s/38 history books, why he left IBM, and what he thinks of the work done by his successors on the technical side of the house.

-Steve