FICON Modeling Service™
The ESCON to FICON Conversion Modeling Solution

Quick links on this page:

What is the FICON Modeling Service?

The FICON Modeling Service is a solution that helps you convert your S/390 datacenter from ESCON to FICON channels. The service uses extensive modeling to determine the best replacement FICON configuration that will address your needs for performance, robustness and economy.

Return to the Top

Why do I need the service?

The storage and availability of data is a key attribute to servicing today's growing reliance on I/T services. While storage media and subsystems have improved and propogated to address this data appetite, the connectivity between storage and server rarely receives the attention to performance deserved. This is because the complexity of addressing the diversity of servers and subsystems connected, the enormous quantity of performance data needed to evaluate bottlenecks and the difficulty of getting maintenance windows to implement changes makes infrastructure improvements difficult to plan and execute.

Your I/O subsystem infrastructure is the core component to a reliable and well performing I/T service delivery system. Without sufficient bandwidth in your channels, directors, and subsystem interfaces, your data center's growing need for data storage will drive your service delivery infrastructure into missed deadlines and unhappy customers.

Server and Storage vendors have responded to shortcomings in the ESCON™ architecture with the introduction of FICON™. FICON has the ability to improve the data delivery speed by many times the capacity of ESCON. ESCON can execute only one channel program at a time over fiber that transfers at a nominal 20 MB/Second. FICON allows multiple channel programs to execute concurrently over a signaling path that operates at 100 MB/Second (200 MB/Second for FICON Express™).

While the new FICON architecture is becoming available, designing a conversion plan to move to FICON continues to be a complicated balancing act. One can always over-build the FICON infrastructure, but this costs money up front in purchasing new hardware and later in a low Return On Investment (ROI) from under-utilized equipment. One can also under-build the fabric, but this leaves a most critical component of the data center service delivery system open to unacceptable levels of performance.

Finding the optimum balance between the expense of converting to FICON and the live-load performance of the resulting infrastructure is the key to a successful and seamless conversion plan!

The FICON Modeling Service is the solution that provides a flexible and superior performance modeling environment which helps you find the balance between costs and results.

Return to the Top

How is the service provided?

The FICON Modeling Service is an MVS, OS/390 and z/OS based service that collects and correlates the vast quantity of SMF and RMF performance data regarding:

  • Channel performance
  • Subsystem performance
  • Device performance
  • Current workload requirements
and derives easy to understand and highly reliable predictions of resulting FICON performance of any infrastructure design you choose to evaluate.

The FICON Modeling Service is not an RMF statistical model like most of our competitors. Instead the service combines device level detail and job processing requirements to derive an unparalleled degree of accuracy in its predictions. You'll see directly how a FICON-based infrastructure will effect your actual processing workloads.

But more than that, you have the flexibility with the modeling service to alter your planned FICON topology then run the current workload through the model and see how changes in what you buy and build will effect your final result.

Return to the Top

What will the FICON Modeling Service do for me?

The FICON Modeling Service answers your most pressing questions:

  • How many FICON channels do I need to purchase?
  • How many FICON interfaces does my subsystem load require?
  • Can I save on channel costs by using logical daisy-chaining to access my less active subsystems?
  • Which volumes are my Parallel Access Volume (PAV) candidates?
  • How large a director configuration do I need to service my I/O load?
  • Can I consolidate subsystems to avoid interface purchases and still maintain acceptable performance?
  • Where is there room to add new storage without jeopardizing current performance levels?
  • Can I relocate a volume instead of purchasing a new subsystem to address a chronic performance bottleneck?

Such knowledge can be a powerful ally when your job is to design a FICON conversion plan that provides the best possible ROI through a thoughtful balance between conversion costs and performance requirements.

Return to the Top