GOMAINFRAME.COM
Your Subtitle text
Home Page
To information technology (IT) executives who just plain don't get it:
DISTRIBUTED COMPUTING ARCHITECTURAL DESIGNS CAN BE WASTEFUL AND INEFFICIENT. YOU MAY BE DOING YOUR COMPANY A GREAT DISSERVICE BY DEPLOYING POWER-HUNGRY, HEAVILIY UNDER-UTILIZED, COMPLEX TO MANAGE SERVERS IN DISTRIBUTED COMPUTING CONFIGURATIONS. AND YOU MAY BE WASTING MILLIONS OF DOLLARS TRYING TO DEPLOY AND THEN INTEGRATE VARIOUS VIRTUALIZATION/PROVISIONING/WORKLOAD MANAGEMENT SOFTWARE PRODUCTS FROM VARIOUS THIRD PARTY VIRTUALIZATION VENDORS WHEN YOU CAN ALREADY GET A COMPLETELY INTEGRATED, CONSOLIDATED, SECURE VIRTUALIZATION/PROVISIONING/WORKLOAD MANAGEMENT ENVIRONMENT TODAY ON A MAINFRAME .
STOP BUILDING DISTRIBUTED ENVIRONMENTS IF YOU DON'T HAVE TO! TAKE SOME TIME TO LEARN WHAT A MAINFRAME IS, HOW IT OPERATES, AND WHY MAINFRAMES SHOULD BE CONSIDERED A "STRATEGIC ARCHITECTURE OF CHOICE" FOR YOUR ENTERPRISE...
What Is a Mainframe?
A mainframe is a general workload server. It can run hundreds or thousands of different applications simultaneously at high utilization rates (mainframes often run in the 90-100% utilization range). Figure 1 shows a mainframe running many disparate applications simultaneously while operating at a high utilization rate.
Figure 1 -- Mainframe Architecture: Multiple, Virtualized Workloads Operating Simultaneously
How Are Distributed Servers Different?
By contrast, distributed servers have been designed to run one application very well (which is why distributed server vendors stress single application performance benchmarks so heavily). And some distributed servers have been designed to run multiple applications (Figure 2 illustrates bot scale-up as well as scale-within distributed server designs). Because the virtualization, provisioning, and workload management software on distributed servers is far less advanced than can be found on a mainframe -- it is comparatively harder to manage resources and balance workloads within your distributed server environments.
Figure 2 -- Scale-up and Scale-within Applications on Underutilized Distributed Servers
Distributed servers generally operate in the 10-20% utilization range (because distributed server managers and administrators leave lots of "headroom" for peak processing periods during the day). To fix this underutilization problem, you're spending millions of dollars to buy virtualization/provisioning software (available from a variety of software vendors); and you are spending a small fortune deploying and integrating this software. Further, you're spending additional IT budget educating your systems/network managers on how to use this new software. And, because distributed environments rely heavily on networking between servers, you're wasting money on expensive communications and network equipment. Finally, you are spending far more money than you have to on security, risk and compliance, and business resiliency. STOP DOING THIS!
There Are Some Legitimate Reasons to Build a Distributed Computing Environment
Sometime you don't have a choice when it comes to building a distributed computing environment. Examples include: 1) if you require Microsoft Windows (Windows doesn't run on a mainframe); 2) if you require Unix (Unix doesn't run on a mainframe); 3) if you workload needs to be processed at geographically dispersed sites; 4) if you cannot get the telecommunications connections that you need to access a centralized mainframe; 5) if your computing workload is too small for a mainframe (in this case, a scale-up/scale-within midrange server can do the work -- or multiple distributed servers can be used).
But, if you've got a large collection of disparate applications -- and if you want to cut your data center operational costs (including human labor-related management, power, cooling, real estate, and physical plant) -- YOU SHOULD BE LOOKING AT MAINFRAME ARCHITECTURE.
The Purpose of this Site
Mainframe architecture offers enterprises a myriad of ways to reduce operational costs -- while providing greater efficiency and easier management than distributed computing environments. But most IT executives tend to dismiss mainframe technology as:
1) too costly;
2) old technology, and,
3) hard to staff.
This site takes aim at these three common objections. It provides opposing opinions based upon research conducted on site at many mainframe customer sites. This site also advocates the use of mainframe technology wherever possible to help enterprises reduce the cost of operations while helping to improve overall operational efficiency.
It is my fervent hope that GOMAINFRAME.com will help you better understand mainframes. And, based upon this understanding, I hope that you conclude that mainframes should be part of your organization's IT strategic plans.
If you find this approach useful, please let me know at jclabby1@AOL.com.
Sincerely,
Joe Clabby
President, Clabby Analytics
jclabby1@AOL.com
(207) 239 1211 (USA -- Mobile)
DISTRIBUTED COMPUTING ARCHITECTURAL DESIGNS CAN BE WASTEFUL AND INEFFICIENT. YOU MAY BE DOING YOUR COMPANY A GREAT DISSERVICE BY DEPLOYING POWER-HUNGRY, HEAVILIY UNDER-UTILIZED, COMPLEX TO MANAGE SERVERS IN DISTRIBUTED COMPUTING CONFIGURATIONS. AND YOU MAY BE WASTING MILLIONS OF DOLLARS TRYING TO DEPLOY AND THEN INTEGRATE VARIOUS VIRTUALIZATION/PROVISIONING/WORKLOAD MANAGEMENT SOFTWARE PRODUCTS FROM VARIOUS THIRD PARTY VIRTUALIZATION VENDORS WHEN YOU CAN ALREADY GET A COMPLETELY INTEGRATED, CONSOLIDATED, SECURE VIRTUALIZATION/PROVISIONING/WORKLOAD MANAGEMENT ENVIRONMENT TODAY ON A MAINFRAME .
STOP BUILDING DISTRIBUTED ENVIRONMENTS IF YOU DON'T HAVE TO! TAKE SOME TIME TO LEARN WHAT A MAINFRAME IS, HOW IT OPERATES, AND WHY MAINFRAMES SHOULD BE CONSIDERED A "STRATEGIC ARCHITECTURE OF CHOICE" FOR YOUR ENTERPRISE...
What Is a Mainframe?
A mainframe is a general workload server. It can run hundreds or thousands of different applications simultaneously at high utilization rates (mainframes often run in the 90-100% utilization range). Figure 1 shows a mainframe running many disparate applications simultaneously while operating at a high utilization rate.
Figure 1 -- Mainframe Architecture: Multiple, Virtualized Workloads Operating Simultaneously
How Are Distributed Servers Different?
By contrast, distributed servers have been designed to run one application very well (which is why distributed server vendors stress single application performance benchmarks so heavily). And some distributed servers have been designed to run multiple applications (Figure 2 illustrates bot scale-up as well as scale-within distributed server designs). Because the virtualization, provisioning, and workload management software on distributed servers is far less advanced than can be found on a mainframe -- it is comparatively harder to manage resources and balance workloads within your distributed server environments.
Figure 2 -- Scale-up and Scale-within Applications on Underutilized Distributed Servers
Distributed servers generally operate in the 10-20% utilization range (because distributed server managers and administrators leave lots of "headroom" for peak processing periods during the day). To fix this underutilization problem, you're spending millions of dollars to buy virtualization/provisioning software (available from a variety of software vendors); and you are spending a small fortune deploying and integrating this software. Further, you're spending additional IT budget educating your systems/network managers on how to use this new software. And, because distributed environments rely heavily on networking between servers, you're wasting money on expensive communications and network equipment. Finally, you are spending far more money than you have to on security, risk and compliance, and business resiliency. STOP DOING THIS!
There Are Some Legitimate Reasons to Build a Distributed Computing Environment
Sometime you don't have a choice when it comes to building a distributed computing environment. Examples include: 1) if you require Microsoft Windows (Windows doesn't run on a mainframe); 2) if you require Unix (Unix doesn't run on a mainframe); 3) if you workload needs to be processed at geographically dispersed sites; 4) if you cannot get the telecommunications connections that you need to access a centralized mainframe; 5) if your computing workload is too small for a mainframe (in this case, a scale-up/scale-within midrange server can do the work -- or multiple distributed servers can be used).
But, if you've got a large collection of disparate applications -- and if you want to cut your data center operational costs (including human labor-related management, power, cooling, real estate, and physical plant) -- YOU SHOULD BE LOOKING AT MAINFRAME ARCHITECTURE.
The Purpose of this Site
Mainframe architecture offers enterprises a myriad of ways to reduce operational costs -- while providing greater efficiency and easier management than distributed computing environments. But most IT executives tend to dismiss mainframe technology as:
1) too costly;
2) old technology, and,
3) hard to staff.
This site takes aim at these three common objections. It provides opposing opinions based upon research conducted on site at many mainframe customer sites. This site also advocates the use of mainframe technology wherever possible to help enterprises reduce the cost of operations while helping to improve overall operational efficiency.
It is my fervent hope that GOMAINFRAME.com will help you better understand mainframes. And, based upon this understanding, I hope that you conclude that mainframes should be part of your organization's IT strategic plans.
If you find this approach useful, please let me know at jclabby1@AOL.com.
Sincerely,
Joe Clabby
President, Clabby Analytics
jclabby1@AOL.com
(207) 239 1211 (USA -- Mobile)
Viewpoints
The Mainframe as the Perfect Cloud
Clabby Analytics is a big believer in Cloud architecture. Cloud architecture is all about being able to find unused computing resources; virtualizing (logically pooling) those resources; and automatically provisioning (building-up/tearing-down OS/Application images) resources to make way for dynamically changing applications.
To us, the perfect cloud would have a centralized control point that could manage virtualization assignments, handle automatic provisioining, and manage workload assignments. This control point would also handle security, resiliency, SOA sessions, and logical resource management.
In our view, the ideal centralized control point for cloud computing is a mainframe. Mainframes offer best-in-the-industry virtualization/provisioning/workload management services; they have the industry's strongest security services; and they have industry leading meantime-between-failure (MTBF). And mainframes are ideal service-oriented architecture hubs (because service oriented architecture (SOA) is message-intensive -- and mainframes can handle a lot of message traffic internally on their large internal, high-speed communications busses -- lightening the messaging load and improving SOA performance).
For more on this mainframe-as-a-centralized-cloud-management-environment, please see this whitepaper on IBM SMCz or here for a YouTube discussion:
http://www.youtube.com/watch?v=vbHNU-QmksU.
And Then There Were Three...
Clabby Analytics believes that the server microprocessor market of the future will be dominated by three microprocessor architectures: mainframe, POWER (RISC) and Xeon 8-core and better designs. We think that UltraSPARC is in big trouble (seriously declining marketshare); we don't see a recovery path for AMD; and we think Xeon will usurp Itanium.
To us, POWER architecture will continue to increase its marketshare due to advanced virtualization services; the fact that it can run multiple workloads simultaneously (it is becoming a general workload processor like a mainframe); and it can be centrally managed (helping to drive down management costs).
We like the new Xeon-class processors because they're x86-based and can capitalize on a wealth of advanced software that already runs on x86 servers -- and because they represent a industry standard 64-bit challenge to Intel's Itanium-class processors (an architecture that we very much dislike).
To better cover these architectures, Clabby Analytics will open GOPOWER7.com shortly -- followed by a Xeon-focused Website (name to be determined).
The Mainframe as the Perfect Cloud
Clabby Analytics is a big believer in Cloud architecture. Cloud architecture is all about being able to find unused computing resources; virtualizing (logically pooling) those resources; and automatically provisioning (building-up/tearing-down OS/Application images) resources to make way for dynamically changing applications.
To us, the perfect cloud would have a centralized control point that could manage virtualization assignments, handle automatic provisioining, and manage workload assignments. This control point would also handle security, resiliency, SOA sessions, and logical resource management.
In our view, the ideal centralized control point for cloud computing is a mainframe. Mainframes offer best-in-the-industry virtualization/provisioning/workload management services; they have the industry's strongest security services; and they have industry leading meantime-between-failure (MTBF). And mainframes are ideal service-oriented architecture hubs (because service oriented architecture (SOA) is message-intensive -- and mainframes can handle a lot of message traffic internally on their large internal, high-speed communications busses -- lightening the messaging load and improving SOA performance).
For more on this mainframe-as-a-centralized-cloud-management-environment, please see this whitepaper on IBM SMCz or here for a YouTube discussion:
http://www.youtube.com/watch?v=vbHNU-QmksU.
And Then There Were Three...
Clabby Analytics believes that the server microprocessor market of the future will be dominated by three microprocessor architectures: mainframe, POWER (RISC) and Xeon 8-core and better designs. We think that UltraSPARC is in big trouble (seriously declining marketshare); we don't see a recovery path for AMD; and we think Xeon will usurp Itanium.
To us, POWER architecture will continue to increase its marketshare due to advanced virtualization services; the fact that it can run multiple workloads simultaneously (it is becoming a general workload processor like a mainframe); and it can be centrally managed (helping to drive down management costs).
We like the new Xeon-class processors because they're x86-based and can capitalize on a wealth of advanced software that already runs on x86 servers -- and because they represent a industry standard 64-bit challenge to Intel's Itanium-class processors (an architecture that we very much dislike).
To better cover these architectures, Clabby Analytics will open GOPOWER7.com shortly -- followed by a Xeon-focused Website (name to be determined).
Clabby Analytics, the host of this site, is an IT research and analysis firm. Unlike many other IT research firms, Clabby Analytics takes hard positions and advocates for the use of certain technologies. We have a clear bias toward:
- Scale-up/scale-within architectures;
- Centralized management; and,
- Energy-efficient architectures.
We encourage our readers to regard our reports as opinion -- and to seek out counter opinions. By blending the two, we believe that our readers will be better able to make informed decisions.
