Service Operations Insight

Hello Team,

As we had discussed, the team has started evaluating the item around Spectrum connector taking 2-3 hours to start up.

The team believes that this is due to initialization of models and hence we wanted to reach out and confirm how many models does the Spectrum connector process in your environment..

We believe it is 100K, but don't want to assume for such critical piece.

Can you folks please respond with your number?

Thanks in advance.

Regards,Kiran DIwakar

Kiran,

  Our connectors generally report on landscapes with 2000 - 8000 devices.  The Spectrum (v9.2.3) databases are anywhere from 200,000 models to 450,000 models.  I'm not sure how many CI's that ends up translating to in SOI, but our connectors do take quite a while to get through the initializing phase.  I also have a couple of dev environments that I can utilize if you want any debug, traces or if you need me to test out anything.

One thing that I have also noticed, and yes, it could be due to my bias towards Linux, but our Linux connectors to Spectrum tend to sync up much quicker than our Windows connector does.

Yes Bill, we all know you are biased towards Linux, many of us are

This is very helpful. Thanks for volunteering to help (as always).

Others, please chime in..

Regards,Kiran Diwakar

Hello Kiran,

i believe that the number of alerts and additional attributes are relevant as well. Especially when the “Reload cleared Alerts” on the SOI Manager is set to “yes”.

Regarding the results. Removing a Spectrum Connector is a much bigger problem for us. It took 1 to 4 hours and if you try to start it while SOI is up and online, there is a chance that SOI stops processing other alerts until the removal has been finished.

We have opened an Idea for that: https://communities.ca.com/ideas/235729415

Is the recommendation that you should not have more than 200k CI´s on a connector system still valid? This will be a problem with Spectrum 10 and much more objects per landscape.

Environment 1: VM´s with w2k8; 2 CPU´s; 12GB RAM

• “Reload cleared Alerts” setting is set to “No”.
• 4 additional attributes are defined.
• DNS Lookup disabled

System 1:

1: 193.886 CI´s 18:00 minutes

System 2:

2:5 CI´s  00:10 minutes :-)

3: 19.121 Cis 01:57 minutes

4: 23.923 Cis  02:08 minutes

System 3:

5: 53.197 Cis 07:10 minutes

6: 7.568 Cis 00:55 minutes

7: 38.109 Cis  04:13 minutes

System 4:

8: 154.226 31:12 minutes

9:84.962 Cis 12:25 minutes

System 5:

10: 128.905 08:20 minutes

11: 19.095 Cis 01:40 minutes

Environment 2: VM´s with w2k12; 2 CPU´s; 48GB RAM

• “Reload cleared Alerts” setting is set to yes.
• 4 additional attributes are defined.
• DNS Lookup disabled

System 1:

Connector E: 25.210  CI´s + 2012 Alerts in Catalyst Startup 88:40 minutes

(The Spectro Server is located at the other end the world. Startup is bandwidth related)

System 2:

Connector A: 158.245 CI´s + 11626 Alerts in Catalyst Startup: 33:44 minutes

Connector B: 1331 CI´s + 41.999 Alerts in Catalyst Startup: 01:20 minutes

System 3:

Connector C: 97.265 CI´s + 16.704 Alerts in Catalyst Startup: 03:45

Connector D: 23.811 CI´s + 31.075 Alerts in Catalyst Startup 09:04 minutes

System 4:

Connector F: 70.255 Cis + 135.979 Alerts in Catalyst Startup 33:08 minutes

Connector G: 97.156 Ci´s + 13823 Alerts in Catalyst Startup  20:14 minutes

System 5:

Connector H: 95.276 CI´s + 57.990 Alerts in Catalyst Startup 11:02 minutes

Regards

André

Hello Andre,

Thanks so much, this is very detailed and very helpful.
Our teams will use this for our research and analysis.

Thanks so much!

Regards,

Kiran Diwakar