Enterprise Class Innovation: System z Perspectives
systemzblogger 2700017BYR 1,834 Views
Back when the the largest processor was a MIP (million instructions a second vs the 50K MIPS of just the z196 Central Processing Complex...) and a megabyte of memory on the 'mainframe' was a million dollars or so (versus about a thousand fold less now with memory advancements and the latest announcements), we saw not only the creation of distributed systems and strategies to optimize the effective use of those key resources, but also looked hard at individual user's cycle use. Of particular focus were systems programmers, developers, and 2 or 3D drafting. We know that CAD stations evolved, and that each TSO user (sysprog or AD) were looked at very closely since --and I remember doing these studies --- each user could easily be linked to say.... a percent or more of CPU usage as recently as the 80's. (Ouch!)
Fast forward to 2010 and the latest evolutionary strategy includes a strong focus on both cycle savings and productivity savings for users. For systems programmers, we have the z/OS Management Facility (which we have mentioned a few times) which is accessible via a browser. For developers on system z, there is the Rational Developer for System z (with Java or EGL). Besides being and eclipse based IDE, the workstation based tool does local syntax checking and -- with the new Unit Test feature -- also offloads still additional cycles. With the focus on effective use of system resources, and the emphasis on maintenance and operational support costs (since they represent a large portion of IT budgets still) these approaches should not be overlooked since they not only address both concerns of animate and inanimate resources (people and machines), but do so in an integrated way.
It is also nice to note that these are just part of a strategy to have tooling that are similar across multiple roles and platforms which address the challenges of models that involve extended development infrastructures for globally integrated enterprises that involve development centers located ...well... anywhere. (See near shore, off shore, on shore and models that are complicated to manage for shore! ) Groan....
..and these are just the first level tools since Rational also has strategies for communication across teams (Rational Team Concert for z ), for modernizing the UI and accessing legacy applications through the tool Host Access Transformation Services, and for advancing SOA services strategies that optimize component reuse by looking at what is out there today via Rational Asset Analyzer .
OK... I won't go off on portfolio reviews here, but will also mention that you should not overlook one of the most obvious areas for effective use of resources and that is: make sure you are current on software compilers since Big Blue has quietly been enhancing them on ALL platforms, but certainly System z...... and the impact over the last 5 years or so can add up to total improvements that may double or more performance across a range of language/subsystem/platform combinations.
So, don't get caught short with old tooling. Refresh, renew and refurbish. Change things out and up. Your mom may have told you to change out your old under things since you never know who may see them. I say change out your older system things before an audit shows you have been missing out on big savings and you have egg on your face to your management team. (How is that for mixed metaphors?) Oh, and do take time this labor day to take a break and set down that phone and e-mail system!
In case you have not seen the Fit for Purpose materials from your friendly IBM site or local team calling on you enterprise, there are a number tools available to help you determine 'best fit' for servers and workload. Based on insight from some external studies, they focus on workloads related to business intelligence and analytics, Web and SOA, traditional transaction, and suites like ERP, CRM, SCM etc. Best Fit has been around a long time with concepts like Balanced Systems (a nod to Ray Wicks et al), 'loved ones' ( and a tip of the hat to Seibo Freisenborg), and constant fiddling with levels and amounts of cache to keep data flowing to the big engines crunching away ---even when those engines where what we would now consider little guys.
Yep, some workloads need more or less qualities of service (QoS), or non functional requirements like availability, security, performance and so on. And some are more compute or I/O intensive, shorter or longer transactions, spread across infrastructures or limited to running on isolated or even specialized processors. This all makes sense to the technical mind, but it is a good idea to remember the power of inertia, decisions already made, and what one buddy called Fit for Politics. It is tough to make changes when decisions are often not re-examined or justified -- they get cast in stone or aligned with factions and can be perceived to be linked to career paths even. Don't forget that in you consolidation or movement plans for the z196 and zEnterprise we have talked about! And don't forget that part of the power of being able to QUICKLY move workloads across platforms in the new complex is that you can quickly try things out and over time learn to trust the idea of decisions not caste in stone, maybe not doing lots of analysis ahead of time and just freaking try it!! (Hey, there's an idea....)
On another 'how things have changed' I cleaned my office recently and found I tossed both round and square backup discs galore. Between more reliable cheaper drives, and backup schemes offsite I realized it had been awhile since I did that hours long data backup and labeling fun time. The other thing I found was a folder of magnetic shapes representing e-business server types. Ten years ago or so when the idea of creating these e-business infrastructures with customers was new I'd find a magnetic whiteboard (easier than you think for big companies) and slap these blocks on the wall with new names like web servers, application servers, portal, gateways for voice or files or B2B....and we would plan out the new world of opening up the enterprise to partners, suppliers and customers. Those concepts and server types are common place now, so I felt pretty secure in setting them free too.
...but maybe I should think about making a new set of magnet blocks with lines of business or service types, with event categories and collaboration options as discussions for the next wave of Smart systems start getting built?
systemzblogger 2700017BYR 2,827 Views
Ta-da! The zEnterprise is out!! I admit I hinted a while back, and yes a little bit last time, about continued evolution and here… it…. is… the System of Systems, the third dimension of not just making processors with faster engines or specialized function, not just growing the number of processors, but now pulling in other platform systems into the z complex.
After going to the teach the teachers session, I took my notes and summarized them for some internal calls with teams, so have been summarizing and netting out, boiling the ocean and relooking at materials. I won’t try to give an announcement here, but just touch on some highlights.
First, realize the vertical integration this reflects, the added dimension of creating the hypervisor of hypervisors (Universal Resource Manager) on top of the other platforms so that the zEnterprise system now wraps its arms around --and the amount of integration represented here. (I understand our friends at Gartner used the term: 'brilliant' in describing the Universal Resource Manager !)
Next, look at the concept of being able to manage, as in Service Level Agreements, not just workload, but security, availability and virtualization targets. (..and the implied amount of monitoring and reporting behind the scenes…)
Then, look at the absolutely killer numbers BOTH in the base complex of the z196 for the kinds of technology improvements we are used to seeing in new z generations, and also in the incredible impact on space, energy, and operating costs compared to an infrastructure before (yesterday!) the zEnterprise complex.
Note that in the z Blade Extension (zBX), that the Power7 and SAO optimizer blades are first, followed next year by DataPower and x-series blades. Take a hard look at what improvements to certain star queries-- that may be 80 fold (or more)-- might do to the concept of how you build your analytics and process systems. And….. if you have looked at the System z Management Facility, look at the new CICS Deployment Assistant that may reduce administrative time up to 80% !
OK, I am giddy… I promised not to repeat the announcement here. It is just so packed with improvements it will take us all time to fully absorb it. So, dive it, we’ll talk later.