DX NetOps Manager

CA Performance Management Office Hours Transcript - July 2015 

Jul 09, 2015 01:10 PM

Chat transcript from CA Performance Management Office Hours: A Live Online Chat [July 2015] which took place on July 9, 2015.

 

Melissa Potvin to Everyone: Hello Everyone! and thank you for joining out office hours today for Performance Management.

Melissa Potvin to Everyone: you can go ahead and start asking your questions!

mark early to Everyone: When is the next version being released and what is in it?

Aditya to Everyone: what is the use of Performance Management tool? And how it is different from CA UIM?

Paul Aherne to Everyone: What are the recommendations for running DR Repository cluster on vmware as hosts rather than bare metal I know don't use LVM

Catalin Farcasanu to Everyone: Where can we find an example on how MVEL should be used in custom expression for extracting information from STRING OIDs?

Michael Poller to Everyone: @ Paul Aherne: (Hi Paul!). Check out this URL: https://community.dev.hp.com/t5/Vertica-Wiki/Configuring-Virtual-Machines-for-HP-Vertica-7-1-x/ta-p/226932

Bernard Clairmont to Everyone: @Aditya - CAPM is really designed to poll large networks and non-snmp based devices - UIM is positiioned more for smaller networks and primarily looking at servers and applications leveraging probes.

John Murdough to Everyone: @Mark - The next release will be 2.6 a late summer, early fall release.  Some of the specifics include support for thresholding on aggregate CPU and Memory, Baseline support in on demand report tables, out of box self monitoring, and group admin enhancements.  Please consider attending our 2.6 sprint 1 end of sprint demo next week

Michael Poller to Everyone: @ Paul Aherne: In general they don't actually recommend using VM hosts but if its a must (and I know its often the only option) then that is a public facing document you can use to ensure they meet the necessary specifications.

Michael Poller to Everyone: @ Paul Aherne: For reference this is a similar doc link reference but specific to physical hardware used for a Vertica cluster:

Michael Poller to Everyone: https://community.dev.hp.com/t5/Vertica-Wiki/Recommendations-for-Sizing-HP-Vertica-Nodes-and-Clusters/ta-p/228148

Catalin Farcasanu to Everyone: @Mark: Will the 2.6 Sprint be available in the Polaris users group?

Melissa Potvin to Everyone: Hi Everyone! if you just joined us, thank you!  You can go ahead and ask questions in the chat window. Our team is standing by..

John Murdough to Everyone: @Mark yes schedule for EOS demo should be published soon

Fernando Moreira to Everyone: I saw in doc that I need to use a physical machina for vertica in cluster

Paul Aherne to Everyone: @Michael Many Thanks

John Murdough to Everyone: @Catalin - We're checking to see if we have some good MVEL examples in our self cert power users guide, but this link may be helpful too - http://stackoverflow.com/questions/29254434/mvel-expression-with-numeric-keys

Michael Poller to Everyone: @Paul: Anytime

John Murdough to Everyone: @Fernando - We strongly recommend that you use physical devices for your DR nodes.  The software will run on VMs; however, performance is less predictable

Josh Bovee to Everyone: So if PM is targeted for larger customers and UIM for smaller, when requests for things like integrating NFA console or CapMan in to PM PC are posted in the forums we get the response that those types of things aren't in the roadmap for PM due to future plans for UIM. Should those of us in larger installations be concerned?

Dave Moy to Everyone: You talked about self-monitoring in 2.6 - can you add details? 

Matthew Stormann to Everyone: @Fernando: Vetica recommends running on physical machines for best performance in large scale environments, however, it is possible to install/run on VMs.

Catalin Farcasanu to Everyone: @John: I already tried some versions, but using the rest client, my device certification xml is not accepted. Tried also the example provided here. It's not accepted either. https://communities.ca.com/message/241803158#241803158

Fernando Moreira to Everyone: @Matthew - My concerned is a large environment, I have 3 datacenter distributed

John Murdough to Everyone: @Dave - This includes some out of the box dashboards for Data Aggregator and OpenAPI system metrics, allowing at a glance view for some key health metrics of your system.  Much of this can be built today as a manual exercise, but in 2.6 we package the dashboards as part of the product

Catalin Farcasanu to Everyone: @John: same here. https://communities.ca.com/message/111876697#111876697. Is there a specific formatting required for this information when it is pushed as XML (rather than modifying it using the wizard). Tried to modify one in the wizard and then to export its definition, but I was unable to do it this way either.

Fernando Moreira to Everyone: Have we HA on CAPC?

Fernando Moreira to Everyone: I mean, in portal

John Murdough to Everyone: @Catalin - perhaps you can send me your email address, and we have someone follow up directly, and help debug the issue

Dave Moy to Everyone: Will there be information in the Dashboards that currently is only available through the DCDebug?

Michael Poller to Everyone: @Fernando: Can you clarify what you're referencing by use of the term "portal"? HA/DR for CAPC is not possible via OOTB functionality at the moment but it is on the roadmap for future releases. If more details are desired it might be best to engage Product Management.

John Murdough to Everyone: @Dave - not specifically the dashboards typically trend performance data about internal metrics, not specifically debug data.  There are some metrics related to DC processing, and bus handling of processes data

Bernard Clairmont to Everyone: @Dave bit more info: the DCDebug info won't be expaded in self monitoring dashboards but you'll notice in the 2.4.1  you can now debug down to the component ID number.

Michael Poller to Everyone: @Dave: Curious, what would you want to see in Dashboards related to debug or log errors?

Catalin Farcasanu to Everyone: @John: it would be great. Tried already to catch someone on the communities, with no luck. Is xxxxxxx@***.com your email address?

John Murdough to Everyone: @Catalin yes that's the correct email

Daniel Holmes to Everyone: Hi @Josh improved infraction of NFA with CAPM is aligned to our strategy - based on customer feedback we are going to focus on Spectrum integration first the 2H of this year and then focus on NFA as our #2 strategic data source / integration - this is critical to support CSPs and very large enterprises until such time that we update UIM to scale to the mega environments - we have not had a lot of requests for CapMan integration to CAPM beyond feeding performance data to CapMan for advanced analysis which we support today

Bob M to Everyone: @Michael - more info about SNMP failures, like eHealth had.  Now we have to sift through DC Debug logs.

Fernando Moreira to Everyone: @Michael - Supposing that I've 3 sites on my environment, What I need to do HA/DR for my Portal? I know that I can do this with Vertica, but I can't do this on Data Aggregator, CAPC.. Is this on the roadmap too?

Paul Aherne to Everyone: Is there any requests to add a separate poller page with the results fails etc to the CAPM product SNMP fails:-)

Josh Bovee to Everyone: Thank you @Dan.  That's good to hear.

John Murdough to Everyone: @Paul, Bob - We'll certainly capture this as a request, there are some metrics on poll failures available via OpenAPI.  I'll see if we can include that in next weeks demo too

Bob M to Everyone: @John - great, thanks

Paul Aherne to Everyone: @ John Tks

Fernando Moreira to Everyone:                Nobody knows about my question?

mark early to Everyone: When will Performance Center be supported on Windows 10 and the new Edge broswer?

Michael Poller to Everyone: @Fernando: We should have info from PM soon about that. You can do that for the DR's as Vertica (non-CA product) supports it as any good DB application would. But right now there is no HA/DR for the DA, DC or CAPC systems possible.

Fernando Moreira to Everyone: @Michael - What is your suggestion in this case?

Dave Moy to Everyone: We noticed from the Vertica sizing tool that we may need to double our RAM.   Our VM people will expect to see this RAM get utilized or we may lose it. If we are monitoring the memory utilization, will we see the RAM get used for certain processes/tasks?  and would you have any documentation?

Sean Moriarty to Everyone: @Mark - CAPC supports IE 10 and IE 11 in 2.4 but need to investigate the new edge browser piece to be certain as to answer your question.

Melissa Potvin to Everyone: Nice steady stream of questions so far. We are at the 1/2 hour mark. Please do keep the questions coming!

Michael Poller to Everyone: @Fernando: The best option being that there is no feature available for this, would be to submit Idea posts to the Community requesting this be added as a feature. There may already be Idea postings there under consideration as you're not the first to ask about those features.

Catalin Farcasanu to Everyone: @Fernando: some time ago, someone in CA used to build Veritas VCS agents for having the all components (DA,DC,PC) running in VCS environment. I don't know if they're still maintained by someone, but for sure they exist. Maybe they are offered as CA Services solution.

Aditya to Everyone: can someone give examples of Non-SNMP devices that PM can able to monitor?

John Murdough to Everyone: @Dave - On RAM usage for your DR nodes, we make some recomendations to support increased catalog size.  You will see this in use of the systems, it is to a great extent influenced by the number of items your monitoring, and the diversity of those item types.   So we provide guidances based on the other aspects you've included in the sizing, but of course your steady state numbers will be more meaningful

Michael Poller to Everyone: @Fernando: Here is one for DC HA:

https://communities.ca.com/ideas/103306275

And here is one for Vertica HA:

https://communities.ca.com/ideas/102996099

Matthew Stormann to Everyone: @Fernando - We're currently working with our architects to define an approach to phase in HA/DR for PM components over time.  This will take some time to unfold.  We'll likely start with an HA solution for the DCs.

Fernando Moreira to Everyone: Good to know @Mathew

Michael Poller to Everyone: @Aditya: The Data Aggregator tied to CAPC is only capable of SNMP polling. BUT... you can use another product called CAMM (CA Mediation Manager) that provides collection of data from devices that don't support SNMP. It passed that data along via its components for viewing of that data in CAPC in Dashboards and Views.

John Murdough to Everyone: @Dave some more specific detail on the Vertica use of memory, you should see a 95% usage which Vertica grabs based on the configuration.  As I re-read your question I think you were asking specifically if your system admins would see the memory in use

Daniel Holmes to Everyone: @Aditya Ericsson, Alcatel, Huaweii element managers provide access to configuration and performance data for infrastructure they are managing that may not be accesible with SNMP or cusotmers don't want to double poll ... specific examples are Ericssson OSSRC, Alcatel 5620SAM, Huaweii M2000

Paul Aherne to Everyone: Does CAMM sit alongside a DC or does it need its own seperate DC

Dave Moy to Everyone: @John - yes - they want to see it being used. So if run reports or open dashboards, or perform some other function, they'll want to see it

Daniel Holmes to Everyone: @Paul you want to deploy CAMM with its own DC

Michael Poller to Everyone: @Paul: We recommend CAMM use its own independent DC that does not also SNMP poll for the DA, but it is possible to use CAMM on a DC also performing DA polling. Just be cautious about performance the more things that DC is doing as time goes on.

Paul Aherne to Everyone: @dave/mike so one dc per DA so 2 DC's need 2 DA's

Fernando Moreira to Everyone: @michael and @Mathew: I know that the integration between, spectrum and PM is for trap. Have we some different integration like webservices? Because the integration has delay to arrive on spectrum.

Catalin Farcasanu to Everyone: @Daniel: good to see you here today. Will the 2.6 still be available in the Polaris group?

Michael Poller to Everyone: @Paul: No just one DA is needed. But in a perfect world, CAMM would be on a DC that is not performing SNMP polling for the DA, while a second DC is stood up in the environment doing the SNMP polling for the DA.

John Murdough to Everyone: @Fernando - actually the event flow from CAPC to Spectrum is not trap based.  The OneClick server collects events through the CAPC Event Manager web services.  There is an interval configured on the OC side as to how frequently we collect new events

Paul Aherne to Everyone: @Michael Cheers

Sean Moriarty to Everyone: @Dave - We would have to test IE 10 edge browser. The current product high level did seem to work but not been even close to exhaustive done. In short no one has formally tested validated. We have been waiting for the port to the new version of ExtJS to be done that looks to be marked for 2.7 release time frame at this point. Honestly, this is the first requested…

Fernando Moreira to Everyone: @John: Thank you

Catalin Farcasanu to Everyone: @John: that's one approach. The other one is to use nhLiveAlarm alarms sent from EM to SpectroSERVER. At least this is what I tried so far

Michael Poller to Everyone: @Catalin: That is possible, but its really just a work around to the integration code that does that for you, without having to configure the notification rules and trap sends through the network, and the resulting config on Spectrum to ensure those traps raise the appropriate events tied to the traps received.

John Murdough to Everyone: @Catalin - correct you could send traps to Spectrum as an alternate approach.  One of the important differences is that the web service approach is that it looks specifically for threshold hold violations for models that Spectrum has synchronized with CAPC

Catalin Farcasanu to Everyone: @Michael: Thanks for that 

Fernando Moreira to Everyone: Thanks for all

Melissa Potvin to Everyone: 15 minute warning... now is the time to ask any final questions..

Fernando Moreira to Everyone: Have we integration with nimsoft?

Dave Moy to Everyone: Yes - thanks to all for your responses!

Catalin Farcasanu to Everyone: Maybe is just another way to identify devices that are not in sync between PM and SPECTRUM. If you use the nhLiveAlarm, once you get a notification from a device that is monitored in PM but not in SPECTRUM, you'll see it separately

Josh Bovee to Everyone: So when will we get to do custom color palettes and branding on PM PC? 

Fernando Moreira to Everyone: Have we integration with nimsoft? lol

Matthew Stormann to Everyone: @Catalin - 2.6 sprint 1 wraps up next week.  Soon after, we'll post a note in the CA IM Pre-Release Group (aka: CA Polaris) that explains how to download the kit.

Michael Poller to Everyone: @Catalin: That is an interesting way to ID devices not sync'ed between the systems

Melissa Potvin to Everyone: @Fernando, we will get a clarification from product management to your nimsoft question. please stand by...

Fernando Moreira to Everyone: sure. Sorry

John Murdough to Everyone: @Josh - At this point we don't have specific support planned, we can get this question to PM for sure

Catalin Farcasanu to Everyone: @Michael: for Spectrum side, you just check the corresponding attribute, if it is defined, but if the device exists in PM and is not in SPECTRUM, you cannot tell, unless searching for it in SPECTRUM.

Paul Aherne to Everyone: Can you change from no tenancy to Multi Tenacy in PM . Say for instance that the initial instll was for no tenancy and then a request to monitor tenants or would this require a new install .Would a mutli tenancy setup be a better default?

Matthew Stormann to Everyone: @Fernando - The long term strategy/vision is to eventually converge the PM solution with the UIM platform. However, there are no short term plans to create an integration between UIM and PM.

Catalin Farcasanu to Everyone: @Matthew: Thanks for that! Looking forward for the kit.

Michael Poller to Everyone: @Josh: Probably not a bad idea to submit that as an Idea posting to the Community. I suspect many others would vote it up and I don't see one logged for CAPC at the moment.

Fernando Moreira to Everyone: @Matthew - My Dream is the all products one day, they will be integrated

John Murdough to Everyone: @Paul - there are a couple different aspects to multi-tenancy so I'll provide a general answer, but we can chat more.  So... you technically always have at least one tenant called the default tenant. You can at any time add more tenants to CAPC and tenant users/groups etc...  So for data access that's easy enough.  If you wanted to do multi-tenant and add monitoring configs that are tenant specific you'd need to add additional DCs etc... 

John Murdough to Everyone: @Paul - probably more than we can type in the small windows on this topic, but if you'd like we can set up some time to discuss in more detail about your specific needs

Catalin Farcasanu to Everyone: @Fernando: dream on 

Paul Aherne to Everyone: @john No problem

John Murdough to Everyone: @Paul - feel free to drop me an email

Fernando Moreira to Everyone: Thank you for everyone.

Catalin Farcasanu to Everyone: I hope to see you all on communities also, not just on the Office Hours here.

Paul Aherne to Everyone: @john your email?

John Murdough to Everyone: @Paul - xxxxxxxx@***.com

Melissa Potvin to Everyone: OK everyone!  Believe we answered all the questions and we have a few follow up items...  Thank you for joining Office Hours today and we hope you found the session today useful !

Paul Aherne to Everyone: @john cheers

Catalin Farcasanu to Everyone: Thank you all!

Paul Aherne to Everyone: @ all Tks

Fernando Moreira to Everyone: Cheers

Melissa Potvin to Everyone: Transcripts will post to the community by EOD. Have a great rest of your day. 

Statistics
0 Favorited
2 Views
0 Files
0 Shares
0 Downloads

Related Entries and Links

No Related Resource entered.