These are questions and answers captured from the Webex chat during the z/OSMF Guild Session 6 on April 20, 2022. Questions were answered by Marna Walle, Gary Puchkoff, Hiren Shah, Travis Biro & Rolando Perez. If you have further questions, you can respond to this forum or contact any of us directly.
1. Where can I find the materials & links that were shared during Guild Session 6?
2. Where are the links to the SHARE Dallas 2022 Sessions?
3. Is there a formal IBM policy statement regarding how the collected customer data is managed and shared by IBM?
Yes, the formal statement is: The information IBM collects will only be used for the development and improvement of, as well as quality control activities related to, IBM's products and services.
4. I would like to know more about how to use this collector of data .. we have ZERT enabled on all LPARs?
The JSON file can be saved once z/OSMF has collected it. You can download it and process the JSON file any way you wish. ZERT is for network compliance and is a separate function within z/OSMF.
5. Was it covered which host sends the data to IBM? Does it require certain outbound internet capabilities?
The z/OSMF server uses the z/OSMF Notification API, which uses Java mail, which needs an SMTP host connection outbound. If you do not have that, you can still download the generated file and attach it to an email to send it. You just send it to zosmig@us.ibm.com which is the email address z/OSMF would have used.
6. What benefit do we really get from using the Feedback function instead of just opening a case?
Cases are for defects. The IBM Service personnel do not share any usage information that they could see from the problem determination materials provided, therefore no Telemetry data can be gathered from IBM Service. In addition, this Feedback information may not even be ascertainable from problem determination information used in a Case.
7. Would there be any value in having the client initiate the sending? Often it is true that a client browser has web access that the z/OS host doesn't have and might have a lower barrier to getting that data across the net to IBM.
That is true, but if the z/OS host does not have access, individually attaching it to email would suffice too. We will consider how we can also use your browser to simplify the sending for those whose systems don't have internet connectivity.
8. Could you use telemetry data to train AI?An interesting question, but for now, that is not the intention.
9. Is there an API for say PCOMM and telemetry that would allow for gathering workstation telemetry?
There is no intention for gathering non-z/OS data at this time.
10. Select responses received to discussion question - What would you use telemetry data for?
11. What is the benefit to me or my company to send the data?
It could benefit you, as it would help you understand better throughout your enterprise what functions were in use on each system. And even discover new functions that you haven't heard of which might help you. Being able to compare systems which you thought were using the same functions, and discovered were not. Or even functions you thought you stopped using, but still were.The benefit is that you would be helping IBM dedicate resources into the most used functions, for the most benefit. Possibly, if you would like to see the consolidated data from the marketplace, that could be shared if it would be helpful to you
12. Does IBM ensure no Personal Information is being sent?IBM has no intention of returning personal Information back to IBM. And you can see that yourself before you submit it if you want to verify that.
13. Wouldn't we be worried that if we shared what we used most that we would be charged more for that functionality in the future? IBM could obviously use this info to maximize profits as well as improve the environment.
There is absolutely no way for us to track back to the customer or machine what was shared. This information won't be share with marketing people (see the IBM policy wording above in the answer to question no. 3, which we have to comply to). If you are unsure about this, you can view it before sending it.
A one-way hash is taken, and there is no way for IBM to reverse the hash to know anything about the sender or machine, or even to compare it to entitlement information.
14. It would be nice to get some recommendation from IBM based on he data analysis.
Telemetry data isn't really best practice information like a health check would be, but if we consolidated market data, you might be able to see how different you were from others.
15. To make this market report more beneficial, should data include the industry, so a bank compares to banks, not oil rigs?
Right now, we have no way of knowing if the collected Telemetry data is test or production, or what geo or industry it is from. We've chatted about letting customers provide more of this type of data, if they wanted. But of course, that would be optional.
16. How would you know how reliable the data you get is?
Because it is built back up after each IPL, so that it is current. There is no storing of data between IPLs so we know it is current. Also, components would control what information is stored. They would be able to keep track of it, for instance if a function was removed from its configuration, they could remove it from the Telemetry data.
17. So basically we want to see infrastructure configurations not actual application/client data.
Exactly. Only system configuration data, and not application or client data.
18. Can you provide performance recommendations about how z/OSMF (IZU* tasks) are performing in our shop versus expectations and suggest improvements?
z/OSMF Performance is one of the topics that will be covered in the next z/OSMF Guild Session No 8.
19. To facilitate the sharing of the infrastructure data, IBM might want to explore designing a z/OS agent to serve the data collector. Customers could establish configuration controls to comply with their continuous monitoring practices.
Telemetry data is intended to represent a system configuration and not be a monitor for compliance. It is intended that customers could control what information is collected, as it is an opt-in model. Clients should be able to control access to particular data at a function level.
20. Select responses received to discussion question - Incentives, would you be more likely to share if:
- Feedback is a good incentive
- Numbers that show our usage and value to the organization. To help keep them from turning the box off. Show them solid numbers that show we do important work.
- So for me PFA or pro-active monitoring would be great help
- A Marketplace report would be helpful in persuading management to exploit or not use a function
- Whether to benefit from say a CICS doing TLS handshakes versus ATTLS/TCPIP.
------------------------------
Rolando Perez
------------------------------