I. STUDY WITHOUT MONITOR DATA
|
- Base model
In the simplest case, the customer installs our mainframe data extractor Xtract on his or her system, and processes the data of interest. The customer then sends to us, via e-mail or FTP to our site, the resulting .SMF file(s), together with a description of the system on which the data was collected and for which one is trying to obtain a base model. We work with the .SMF file(s) and the customer's description, produce a calibrated model of the current situation, and ship it back (via e-mail or FTP) to the customer.
In our experience, the building of such a base model is the most difficult part for most customers. Our customers can take advantage of our model calibration services to quickly obtain a calibrated Lead Time model for their system. Please note that the time required to calibrate a model increases with the complexity of the physical configuration as well as the number of controller images.
- "What if" scenarios
If the customer wants us to also perform his predictive "what if" studies, we need a detailed description of the scenarios under consideration. As deliverable, we produce the corresponding Lead Time models, as well as a report corresponding to scenarios submitted by the customer.
The charge for this service is based on the hours spent for a specific course of study. Clearly, the final charge will depend on the number and complexity of the scenarios considered. We can provide a preliminary estimate of time which will be needed. The engagement hours can also be bought prepaid at discounted rates in units of 4.
- Controller consolidation
Among our services, we offer controller consolidation predictive studies. Here, we will help you assess the response time you can expect when the I/O workloads currently spread over several smaller controllers are moved to a single larger control unit. Our deliverable includes relevant Lead Time models and a report detailing the expected performance after consolidation.
Pricing is based on the complexity of the configuration as measured by the number of controller system images.
Support, response time and temporary Lead Time license
Our typical response time for base model calibration and controller consolidation is 72 hours from the time the correct input files and the customer's configuration data are delivered to us via e-mail or ftp and accepted by us. While working with Xtract, you are entitled to free technical support on Xtract issues directly related to the job. You are also entitled to three free technical support calls regarding the results of the study and Lead Time use. In all cases, you get a 2-month license to use the Lead Time I/O modeling software. This allows you to do additional work, if desired, with the models produced.
Data source and related information
- Assuming the controllers are currently under system 390, z/OS, our mainframe companion software Xtract should be used to process the RMF data for the period of interest. The preferred method is to have Xtract installed on customer site. Xtract is available for a free download from our Web site. The Xtract User's Guide is available in pdf format via public ftp from our Web site. Xtract is also packaged with all versions of Lead Time, including the free trial version. When included with Lead Time, Xtract object modules and related files are installed in a subdirectory /Xtract of the installation directory. The Xtract object modules need to be transferred to the mainframe and link edited there.
Note that the current of Xtract includes Xtgen, a Windows-based application which generates the correct JCL based on a few simple dialogs about the configuration and the desired time period. Additionally, the file samplejcl.jcl provides an example of JCL for a "generic" data extraction based on device numbers. This file is available from our Web site via anonymous ftp. Xtract should be set up so that is produces one .smf file per controller image (LCUGRP=NO). Xtgen will automatically generate the correct JCL to accomplish this.
To reach our public FTP site: ftp://ftp.pallas-international.com/.
We will use the .smf files to build the model of the base configuration. It is important that, as much as possible, cache activity records be present in the input data set for Xtract. The .smf files together with the configuration description is the input we want from the customer.
- For each controller, we will need controller configuration information such as number of Host Adapters and channels, number of Device Controllers (on the back end), total number of physical drives (excluding spares), as well as the type of physical drive used. If available, logical to physical mapping information would be useful. If applicable, we need to know which volumes are mirrored (if the customer is not using RAID-S or RAID-5 or higher). For Remote Copy activity, we will need to know which volumes are involved and the type of Remote Copy (essentially, synchronous or not). For HDS controllers, if GraphTrack data is available and can be processed through our Osx software, configuration information such as the number of Host Directors and Device Controllers will be automatically available.
• Other names for Host Director: Channel Adapter Processor (CAP), Client-Host Interface Processor (CHIP)
• Other names for Device Controller: Disk Adapter Processor (DAP), Array Control Processor (ACP)
- Lead Time includes the explicit representation of Business Continuance Volumes (BCVs or Dynamic Mirrors) for EMC controllers. We do charge additional fees for configurations with BCVs, relative to how much work is involved. This depends on whether the controllers are in a Shared DASD environment, and on the number of BCVs to consider.
Deliverables
We will deliver to the customer via e-mail or FTP from our Web site the following.
- For base model study
• calibrated model of the current configuration
• analysis of bottlenecks in the current configuration including an estimate of sustainable I/O throughput.
- For "what if" scenario study
• base model study deliverables as above
• report describing results of customer's scenarios
• Lead Time models corresponding to these scenarios
- For controller consolidation study
• base models of current controllers
• report describing expected results of the consolidation
• Lead Time model(s) corresponding to the consolidation scenario
In all cases, you get a 2-month license to use the Lead Time I/O modeling software. This allows you to do additional work, if desired, with the models produced. While working with Xtract, you are entitled to free technical support on Xtract issues directly related to the job. You are also entitled to three free technical support calls regarding the results of the study and Lead Time use.
II. STUDY WITHOUT MONITOR DATA
|
Types of study
In some cases, one is only interested in quick sizing of the expected performance of a given type of workload on a specific storage controller. It is possible to perform such a study without first building a detailed base model. In this case, the customer needs to supply as much of the workload information through the input spreadsheet as he or she desires. Adequate information regarding the physical configuration is required. In this case, we will perform:
• speculative "what if"
• consolidation scenario studies, including
• moving from ESCON to FICON
• general configuration sizing, etc
Deliverables
We will deliver to the customer via e-mail or FTP from our Web site the following.
- For predictive "what if" study
• report describing results of customer's scenarios
• Lead Time models corresponding to these scenarios
- For controller consolidation study
• report describing expected results of the consolidation
• Lead Time model(s) corresponding to the consolidation scenario
In all cases, you get a 2-month license to use the Lead Time I/O modeling software. This allows you to do additional work, if desired, with the models produced. You are entitled to three free technical support calls regarding the results of the study and Lead Time use.
|