Solaris is coming, Solaris is coming.....to System z

Wow....I felt the earth shake just now....IBM and Sun have announced a live demo of Solaris on System z. Experts at Sine Nomine – the same folks who were at the epicenter of the Linux on the mainframe movement – have brought the two companies together and showcased the possibilities at this week’s Gartner conference in Las Vegas. 

Click here to see David Boyes from Sine Nomine discuss the project from the Gartner event floor. 

Amazing. But why are they doing this? To make customers’ lives easier. One question I hear all the time is: Is IBM trying to take over all computers? Does IBM believe that a mainframe or a collection of mainframes can replace all the servers in a business? The answer to that is simple: NO. 

How do I know that? Because an IBM mainframe can never be the sole computer in any business and the reason is, it doesn't have it's own front end interface....well, maybe it does....the punch card and the 3270 terminal. But they really didn't take off the way IBM anticipated, so now you have ATM's, Kiosks, Web browsers, cell phones, pda’s, PC's with 3270 emulators....a cornucopia of front end devices as the human computer interface. In that regard, IBM's mainframe has to work as a master collaborator to make sure that it can interoperate with that wide range of front end processors and as such, has had to augment the 3270 data stream with slick new XML and web services built into it's systems.....but I digress....we are talking about Solaris on the mainframe....is this to be a demonstration system, a proof of concept, or an actual supported system? 

It’s a demo today, but it's intended to be a supported system – and it’s only a matter of time before we see it. We've learned already that customers may desire open source computing on z/OS, zVM and Linux for System z. They just aren't willing to service it themselves. Sure, they can download a myriad of tools from the internet, but they can't download a service contract. So they look toward distributors or vendors to provide service for those offerings before they'll put them into production. The same will be true of Solaris on System z. 

On this lovely mainframe, we also realize that code development is generally created on the desktop...an x86 platform and as such, typically favors native operating systems as the deployment platform as well. Those platforms are predominantly Windows, Linux and Solaris-based. But because of openness and portability, Solaris and Linux can be deployed on multiple hardware platforms now. Today, deployment of servers on the x86 platform has been considered Scale Out computing....just keep adding more and more server images, each dedicated to a single operating system and typically a single application or data base server. In many respects, these servers have the appearance of an appliance, because they are "single" function devices. Those servers might satisfy the needs of a lot of folks (e.g. clients), but typically excel at a single function. 

In the last couple of years, it seems like the next new thing is Virtualization and server consolidation...the ability to host multiples of these "appliances" in a single container and, in doing so, make the operational environment more green - use less energy, cooling, floor space, etc, but still meet businesses' service level agreements. Well, you'd think that virtualization just got invented. Nope - it's been around for over 40 years, with IBM's zVM as the cream of the crop in systems virtualization. We've already learned the power of virtualization when associated with Linux for System z. There have been a large number of deployments and over 90% of those deployments are on zVM. With Solaris on System z, 100% of the deployments will be on zVM. That's because the operational environment will take advantage of some of the native System z resource sharing and management tooling, in addition to offering the opportunity to manage each Solaris image independently. 

The virtualization available on System z through zVM has a number of distinct advantages over the Johnny-come-lately virtual servers on other platforms - the ability to run at 90% and higher system utilization without fear of failover for a very large number of operating system images; the ability to add or remove capacity on System z by turning on or off additional processors without suffering a service disruption and in doing so, meet tactical business processing needs on demand; the ability to leverage hardware and system memory to communicate between operating system images, which in turn reduces the number of system intrusion points; the compartmentalization of one operating system image from another one to provide an additional layer of security; the ability to use zVM services across Solaris system images for common auditing, disk and tape back up processing. 

The net of this is, it's the same code that you might be running on a different hardware architecture, but when executed within the zVM hypervisor, inherits much of the operational superiority from that environment, with no significant additional cost. And then there is the hardware benefits of the System z architecture...it had an OnStar like call home capability and autonomic healing capability long before the blue prints were written by GM or other server platforms. IBM's mainframe remote support facility provides electronic diagnosis of system failures and with redundant hardware built in, can switch over to the backup components and in parallel "call home" to dispatch a customer engineer to correct the problem. In the case of CPU failure, the z architecture will swap in a processor, transparent to any operating system running on its hardware to continue processing unabated. It's like changing the tire while the cars in motion and calling ahead to have a new spare put in the trunk, again without having the car stop. 

Let's continue that automotive metaphor. The mainframe is intended to be a super highway. There are folks that believe it's just a parkway...allows only cars and it's not heavily traveled. Today's announcement is just another occasion to demonstrate the super highway nature of the mainframe. It enables all kinds of vehicles to travel on its roads and the traffic is moving along very quickly. In fact, there are sensors in this mainframe highway to detect bottlenecks and provide re-balancing workloads to meet service goals, another value that Solaris on System z will be able to take advantage of as well. But let's not forget, the interstate highways are selective, they don't accept bicycles and pedestrians. There are other roads, running at a slower pace for those folks to travel on. Yes, they'll reach the same destinations, but it will take them a bit longer to get there. So in that sense, the mainframe is not going to "take over Solaris". There will be certain application and data serving workloads that will more naturally appeal to a consolidated effort on a mainframe and there will be other workloads that can continue to run independently on other server hardware or for that matter be virtualized in an x86 environment, quite possibly because they are stateless and don't need the resilience, security and capacity management that a mainframe brings to the operational environment. 

Linux on System z has been wildly successful in its ability to consolidate workloads. Does Solaris present a "weakness" in the force driving Linux ubiquity? No, quite the contrary. It's about flexibility and choice. In many cases today, Linux is evolving as a server of choice in the x86 world and collections of those servers can be easily consolidated to IBM mainframes running Linux. But in some cases, a UNIX workload, like Solaris, must first be ported over to Linux before it can take advantage of the virtualization and scaling capabilities of the mainframe. In addition, some operational tooling is different, so there might be a skills hit or a procedural hit to make a change. Well, just as Linux is Linux, regardless of where it's deployed, the same objective holds true for Solaris. Common code is a re-compile vs a port and the objective is to enable the same operations model, but offer some new capabilities, through zVM, that further reduce the operational complexity of running many, many Solaris images on the same mainframe. 

So back to the celebration....there's a new choice coming to town, Solaris on z. The benefits of the Solaris operating system as many businesses have grown to enjoy on other architectures, with the benefits of the operations model and virtualization capabilities of IBM's System z mainframe and zVM. Long may they both prosper to give businesses the choices and flexibility they need to build global system deployments that meet their business governance, privacy, security and resilience needs in solving problems along with the myriad of other options available for them to deploy on this modern mainframe.

by JimPorell November 28, 2007 in Future
Permalink | Comments (4) | TrackBack (0)

Give and take

Did a webinar the other week based on a whitepaper I wrote on give and take between the mainframe and the trendy new youngsters. Details here.

by pwarmstrong November 20, 2007
Permalink | Comments (1) | TrackBack (0)

the greaterIBM connection

CactusOne of the many innovations Sam Palmisano has spearheaded at IBM is the idea of reaching out to "alumni". The first initiative was a few years ago when he hosted a reception for a group of former executives of the company. A few were retired but most were in senior positions in other companies. That was just the beginning and now the idea of reaching out has been expanded -- big time. The number of past and present IBMers is probably close to a million people. Establishing communications with such a huge base can be nothing but a good thing for the company.

When I left engineering school and joined IBM in 1967, it was common to look for a job at a company and expect to stay there your entire career. Nobody thinks that way anymore. If you tell someone you were with a company for decades, they might ask "what's the matter, couldn't you find any other jobs?". Another change is that in the old days if someone left the company they were considered a traitor and barred from coming back. Today, there are many executives that left the company at some point, got some experience at one or more other companies, and then brought that experience back into IBM. Some have come and gone multiple times. The turnover has strengthened the company.

PeopleAnd now we have social networks. In the early stages there was a perception that social networking meant eleven year-old girls on MySpace. Now businesses are realizing that it is more likely forty or fifty year-old business people on Facebook and Xing and LinkedIn and Plaxo Pulse. The Internet has enabled everyone to be connected to everyone. Whether it is reading blogs, posting to wikis, updating status on Facebook, or making new connections through viral invitations, it is clear that a big company like IBM has a lot to gain by "connecting" past, present, and future IBMers to each other and with the company. IBM calls it "the greaterIBM connection". On Monday evening the company hosted a greaterIBM reception at the Metrazur at Grand Central Station in New York. More than four hundred attended. It was good to reconnect with some colleagues I had not seen for quite a few years.

Business ConferenceWill social networking payoff in business terms? Nobody knows for sure but in my opinion it is certain -- as soon as we see the New York Times run a front page story that social networking is a fad, in trouble or peaking out we will have confirmation that success is a sure thing. A short term inhibitor is that there are so many different social networks. As web standards evolve I am confident that we will have a world where people will create one profile and then be able to decide which part of their profile is accessible in which networks.

IBM sees the potential and is investing the time and resources to build a large and active network. The possibilities are endless -- collaboration on projects, networking to hire or get hired, crafting deals, referrals to and from IBM and its business partners. As a bonus, social networking is fun and good for morale. I look forward to continuing to be a part of the greaterIBM connection as it evolves. Upon e-tirement in 2001 after nearly four decades at IBM, I don't really feel like I left anyway! The stories that I have been writing since 1998 over at the patrickWeb blog fall into a number of categories. One section is devoted to "IBM Happenings". I am sure I will also be writing and linking at the greaterIBM connection along with others. Cross linking will increase the overall "connectedness". That's what the web is all about. I am really proud that IBM is taking networking and the blogosphere so seriously.

Related links
bullet the greaterIBM connection

bullet Greater IBM Wiki

by John R Patrick November 14, 2007 in History, Innovation, People
Permalink | Comments (1) | TrackBack (0)

Mainframe on YouTube - the sequels are in!

For all of you "Mainframe: Art of the Sale" fans, there are three brand new installments available now.

Mainframe_video2


Bob's protégé is ready for his first sales call - but can he close the deal?



You'll have to see for yourselves.  Click below to view:
Lesson Four
Lesson Five
Lesson Six

If you've never seen "Mainframe: Art of the Sale," you might want to start with these:

Lesson One
Lesson Two
Lesson Three

by Kevin Acocella November 13, 2007 in People
Permalink | Comments (4) | TrackBack (2)



The postings on this site are our own and don’t necessarily represent the positions, strategies or opinions of our employers.
© Copyright 2005 the respective authors of the Mainframe Weblog.