Sign In

Ask the Experts What would you ask an IBM mainframe expert?
z & A > Ask the Experts > Share Your Mainframe Questions View modes: 
User avatar
Advanced member
Advanced member
mwestholder - 2/1/2013 4:04:49 PM
   
Share Your Mainframe Questions
Select submissions in this forum may be answered by IBM experts within the Destination z forum itself, through online articles or during upcoming 2013 educational webinars. In addition, those who post may be selected at random to win a prize.

User avatar
Guest
DanielCabrera - 2/14/2013 2:54:06 PM
   
RE:Share Your Mainframe Questions
Hi,

I would like how System Z integrates to Linux. I mean, can I run any Linux application on a system Z? Can I run an apache web server or a mail server application and have these running in a powerful hardware?

Thanks

User avatar
Member
Member
Wayne Rhoten - 3/7/2013 7:51:11 PM
   
RE:Share Your Mainframe Questions
A single System z machine typically is configured into multiple partitions. Each is called a logical partition or LPAR. This has been true for about 20 years and is being adopted on other platforms.
This is a form of virtualization, whereby the operating system that runs in each LPAR sees only that part of the machine. The rest of the physical machine is not visible.
There might be a dozen or more LPARs. System z supports these five operating systems: z/OS, z/VM, Linux on z, z/VSE and z/TPF. You can have any combination of these five systems in your LPARs. (The name "zLinux" is a slang term for "Linux on z".)
You also can configure one or more LPARs as an IFL, Integrated Facility for Linux. It is another layer of virtualization. In an IFL you can run hundreds of copies of Linux. Each Linux "thinks" that it has the whole machine. Of course in each Linux you could run Apache or a mail server or multiple programs at the same time.
The result is that you can run more than one thousand Linux virtual machines on one System z. "Regular" Linux machines on other platforms might typically run between 5% and 40% CPU utilization and that is as fast as they can go. You can run a thousand Linux virtual machines on z with >99% CPU utilization and it can be efficient.
With these several types of virtualization, you can provision a new Linux virtual machine in a few minutes instead of waiting for hardware delivery in a few weeks.

Another possibility is to use zBX, System z Blade extension. You can logically attach certain types of non-z machines (such as Intel x86) to z and run Linux in them. Those machines also can run virtual machines so several copies of Linux can run in each physical machine. These machines benefit from the proximity to z/OS so they can be managed together and benefit from economy of scale.

User avatar
Member
Member
DCabrera - 3/8/2013 10:58:29 AM
   
RE:Share Your Mainframe Questions
This is a good explanation.

Thanks Wayne

User avatar
Member
Member
sandeepkvv - 2/14/2013 4:35:18 PM
   
RE:Share Your Mainframe Questions
I'm working as Mainframe Storage Admin..
Seeing all the cloud and other things, how can we improve performance with more and more capacity being added ... I see FAST implementation in EMC boxes and few other technologies. But they are coming faster on open systems and after an year or so, it will be released.on mainframe.. Too bad for that.

How does Mainframe overcomes all these performance things ..

User avatar
Member
Member
Ken - 2/14/2013 6:37:30 PM
   
RE:Share Your Mainframe Questions
I would like to hear what IBM's future intentions are for z/VM-z/Linux and the z/BX. Do they plan on upgrading the blades to keep up with pSeries and xSeries technologies? How much of a lag should we expect? I do not believe they have announced any enhancements to zBX. Can we expect a z/BX only configuration to compete with VMware and Hyper-V clusters?
KLC

User avatar
Member
Member
sensuixel - 2/15/2013 1:33:46 AM
   
RE:Share Your Mainframe Questions
Hi,

My first question would be :

How would you code a DCB and DCBE in 31 bits in an APF-Authorized LOADLIB ?

And the second one would be :

In which language is DFSORT coded ? (Assembler, C ) ?


User avatar
Member
Member
Wayne Rhoten - 3/7/2013 7:56:11 PM
   
RE:Share Your Mainframe Questions
The DCB and DCBE macros are coded the same in an APF-authorized program and in an unauthorized program. The DCB must be below the 16 MB line. The DCBE can be below or above the line even if the program is running in 24-bit. Of course the program must at least temporarily be in 31-bit mode to build the DCBE.
DFSORT is mainly written in assembler but that should be irrelevant to all users.

User avatar
Member
Member
sensuixel - 3/8/2013 12:02:57 PM
   
RE:Share Your Mainframe Questions
Thanks for you reply.

I was asking about the DCB-DCBE since I get an S337 on
a 31-Bits reentrant APF program.

The S337 disappear as soon as my prog leaves the APF library, that's not a big deal either.

About DFSORT i can hardly imagine the complexity of the assembler code given the level of performance and the almost infinite options
of DFSORT.

Just being curious, what is the total number of line of code ?


User avatar
Member
Member
flupher - 2/15/2013 12:08:47 PM
   
RE:Share Your Mainframe Questions
As interest and confidence in cloud services continues to grow, would you anticipate this trend spurring the demand for mainframes, a platform proven to excel in both stability and security, areas of major concern?

1 2