System i - Releasing the Power to Innovate

[originally posted elsewhere, June 2008]

Some thoughts on what the recent Power Systems announcements mean to Logicalis and its System i customers
IBM's recent Power Systems announcement does two things. It completes the convergence of the System p (AIX) and System i (AS/400, iSeries) platforms. But, more interestingly, it may allow the System i's core operating system to free itself from the shackles of history and perception.

Until these announcements, IBM's Power based hardware platform came in two flavours: a System p, which ran AIX and/or Linux, and a System i, which ran any combination of i5/OS, AIX, Linux and Windows. You'd think this was a no-brainer from a virtualisation perspective ... but it wasn't, because, like for like, the hardware cost more on the System i platform. Also, vendors weren't comfortable with their AIX-based products running on the System i hardware platform - even though AIX was installed, configured, and managed exactly as on System p.

The defining characteristic of a System i was always the i5/OS operating system (or OS/400 - now renamed just ‘i'). The value i5/OS gives to a business is in its provision of a complete, upwards compatible, hardware independent, scalable environment in which practically any business application can be accommodated with very low TCO. The value is not in the hardware.

The System i architecture was far ahead of its time when it emerged in the 1970s. Single level storage, subsystems, and the hierarchy of microprocessors (to name just a few) were new ideas. The ‘iSeries Nation', as it was later christened, cottoned on very quickly and watched, amused and/or bemused, as the rest of the world caught up. But there's a lot of other stuff in the average IT estate. This other stuff works best with external storage. It runs on a blade. The System i didn't even go in a rack for many years. It's not perceived as easy to deploy in a data centre. Its hardware just doesn't inhabit the mainstream.

The new announcements recognise all this by dividing a common hardware platform (Power Systems) with common management software (PowerVM for virtualisation, PowerHA for clustering) from the operating systems (‘i', AIX, Linux) that run natively on it. IBM has further satisfied the need to move into the mainstream through major investment in external SAN connectivity for i5/OS and, most tellingly of all, by delivering a range of Power Systems blades that are capable of running i5/OS.

So that's the hardware dealt with ... what about that complete, upwards compatible, hardware independent, scalable environment for business applications?

Logicalis has been working with IBM midrange business systems since 1985. Then, we had a proprietary, integrated, resilient, secure and easily managed operating system running green screen applications. i5/OS in 2008 is an open, integrated, resilient, secure and easily managed operating system that can and does run any combination of workloads - from portal and Web 2.0 solutions through enterprise content management and data warehousing to complex ERP environments, large scale SOA deployments and high performance, high volume, high security financial systems.

That's the reality.

The perception is that i5/OS is an expensive, proprietary, dying platform of interest only to its diehard fans.

In general, people buy an application, then buy a platform for it. In the 1980s and 1990s, many vendors wrote applications for System/38 and AS/400 as the platform of choice. The crumbling of i5/OS' strong position with ISVs can be blamed on many things: the ridiculously well-kept secret that is the built-in i5/OS database; Oracle's and Microsoft's stronger, and much better marketed, messages; the cost of the hardware; the growing assumption that the platform was proprietary and dying.

The built-in DB2 on i5/OS provides comparable function to DB2 on other platforms and to Oracle. i5/OS' built-in security can provide an exclusionary access model, suitable to today's compliance climate. i5/OS' built-in work management can handle whatever mixture of workloads you throw at it, all within one logical partition. If ISVs could be encouraged to see the Power Systems hardware platform (rather than an AIX software platform) as their target, maybe some of them might look at i5/OS with new eyes and understand the range of functionality that is available out of the box. They might twig, for example, that an ‘i' partition, with its highly functional built-in database requiring minimal DBA support, might just be easier to implement, cheaper to run, and more easily maintainable than an AIX partition running Oracle.

If the newly rebranded ‘i' operating system is to be recognised for what it can achieve, we have to get the message out to ISVs that, however innovative their business application, i5/OS is potentially appropriate, cost-effective, and most importantly here to stay. IBM's very clear continuing willingness to invest massively in i5/OS and its hardware provides a strong message to ISVs and to potential new i5/OS customers. Let's just hope they listen!

Comments
Mandy Shaw, 7 months ago
It's a long story. Technology problems in the early 1990s forced many client/server application developers (including ourselves) away from the platform, in many cases never to return. To be precise, the problem was the lack of an effective ODBC driver - it's still depressing to contemplate how easily this could have been got right... Then Microsoft began its march to world domination. A key moment was J D Edwards, having been loyal to the AS/400 platform for many years, choosing Windows as the main platform when they re-architected their ERP system. The System i went from strength to strength technically. Every line of operating system code and microcode was rewritten for RISC, an enormous development project that was delivered on time, to budget, and with absolutely minimal impact on customers and full upwards compatibility. The layered architecture and the hierarchy of microprocessors were leveraged to support dynamic logical partitioning, Capacity Upgrade on Demand, Windows, AIX, Linux, and Java (my usual line is 'designed in the 1970s to support web applications in 2008'). The database became more and more industrial strength (while the average System i shop used less and less of its functionality). But it was too late. It has obviously not helped that the System i platform has been the consistent loser in internal IBM competition and politicking between the server platforms. But I do think the opportunity is genuinely different and stronger following the Power Systems announcements. I await developments.

... another comment ...