Mainframe, MVS and zOS Discussion

Hi,

I'm new to mainframes but more I read about it, more or less it sounds like a Server but if it is a server, why it is called mainframes? Can you please your thoughts on this?

You are confusing the what it is with the what it does/how it is used

I have two PCs, and the FTP/HTTP servers are running on both of them,
they both could be called servers but they still are just two good old plain PCs

The mainframe has been called a mainframe for more than 60 years -- long before anyone even conceived of servers.

Thanks.

When we talk about data-centres and serves the context of server becomes clear but someone from open-system world asks the question what is mainframe compared to a server what answer can be given...this is where I am coming from.

A mainframe is a large computer system with redundant internal engineering, extensive I/O facilities, high utilization rates, and strict backwards compatibility with older software. Mainframes can run multiple server instances simultaneously (dozens or hundreds of them). Mainframes use proprietary operating systems (such as z/OS, z/VM, z/TPX) and the architecture of the hardware is very different than Windows / Unix servers.

To some degree you are attempting to compare apples to elephants. A mainframe is hardware and software; a server is software functioning in a particular way -- you can have Windows servers, Unix servers, mainframe servers. This may be where some of your confusion is coming from.

Thanks Robert but I have two questions on your post.

First why do you call it as "redundant internal engineering"? The meaning of which is "not or no longer needed or useful; superfluous". So you want to say mainframe itself is based on redundant engineering??

Second, you yourself said "mainframe servers", so in the context of your statement "mainframe" and "mainframe servers" are different?

Redundant as in much can be replaced without an outage, not redundant as in no longer needed. There's been a lot done to ensure there is no single point of failure in the mainframe; most mainframes can have a large percentage of the hardware fail and continue to execute flawlessly. Most sites configure at least two channels to each I/O device, so failure of a channel will not impact access to the disks or tapes associated with the I/O device.

Yes, a mainframe server and a mainframe are different. A mainframe can run transaction servers (CICS), interactive sessions (TSO), background jobs (batch), web services, Java, Unix Systems Services, communications servers with 100% utilization for days at a time without degrading performance. The mainframe can execute mainframe server software as one (or several) of the many tasks it handles simultaneously. The mainframe server software could be Websphere, HTTP Server, and so forth -- basically, any software that can accept client requests and return responses (which is the definition of a server). Since there is much executed on a mainframe that is not client-centric (started tasks and batch jobs for example), a mainframe cannot be considered JUST a server.

Here's the page for the latest IBM Mainframe, announced in January this year: http://www-03.ibm.com/systems/z/hardware/z13.html

If you have a look at the Datasheet (around about the middle of the page when I see it) and consider the operating systems that it can run, if nothing else, you'll see one of the many ways how where categorisation doesn't really give you much. With the doors shut, this is one physical box. It can be z/Linux (up to 8,000 instances if you dedicate the whole machine to it). It can be running other "unixes". It can be running Windows Server software. It can run four different (architectually different) Mainframe operating systems. And it can run this mix all at the same time, with a maximum 10TB (that''s 10,000GB) of main storage (RAM).

The performance advantages of having that all in one box, as opposed to across "wires" or "networks" is immense.

You can call it a Server if you like, but there's no other Server even remotely like it.