The ports on the switches doesn't seems to display correct network throughput
Hi Team,
I installed Observium recently. It seems to be working just fine.
However, I do see discrepancies in the throughput result.
When I run a network performance tool (Nuttcp) on one of the server and monitor the corresponding port on the switch through observium. I observe that the the server is generating around 9.8 Gbps data. If I check the ports directly on the switch I see same thing (~9.8 Gbps) but when I see in through observium on the corresponding port , I see frequent spikes to 12 - 15 Gbps. I am not sure why I am seeing these high value because the NICs on servers are having a 10 G NICs and I never see anything higher than 10 Gbps on server or on the port on the switch. But the real time graphs on Observium shows the spikes of around 12 - 15 Gbps.
I would be really glad if you help me on this. I really like the tool but the discrepancy in the throughput result is kind of driving me crazy.
Thanks!
Regards Renuka
________________________________ This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
Hi Team,
I'll be grateful, if you can help me on this.
Thanks!
Regards Renuka ________________________________ From: Arya, Renuka [BSD] - MED Sent: Friday, September 05, 2014 6:33 PM To: Observium Network Observation System Subject: The ports on the switches doesn't seems to display correct network throughput
Hi Team,
I installed Observium recently. It seems to be working just fine.
However, I do see discrepancies in the throughput result.
When I run a network performance tool (Nuttcp) on one of the server and monitor the corresponding port on the switch through observium. I observe that the the server is generating around 9.8 Gbps data. If I check the ports directly on the switch I see same thing (~9.8 Gbps) but when I see in through observium on the corresponding port , I see frequent spikes to 12 - 15 Gbps. I am not sure why I am seeing these high value because the NICs on servers are having a 10 G NICs and I never see anything higher than 10 Gbps on server or on the port on the switch. But the real time graphs on Observium shows the spikes of around 12 - 15 Gbps.
I would be really glad if you help me on this. I really like the tool but the discrepancy in the throughput result is kind of driving me crazy.
Thanks!
Regards Renuka
________________________________ This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
Are you in the newest version of Observium? Did you install Observium on Ubuntu or Centos? Is it the paid version of Observium?
-- Sincerely, Joshua Hopper, A+ CE Network Administrator [cid:image001.jpg@01CFCC18.D16C2280] 420 3rd Ave NW Hickory NC 28601 Office: 828-449-1839x2160 | Cell: 828-855-7565
From: observium [mailto:observium-bounces@observium.org] On Behalf Of Arya, Renuka [BSD] - MED Sent: Tuesday, September 09, 2014 10:25 AM To: Observium Network Observation System Subject: Re: [Observium] The ports on the switches doesn't seems to display correct network throughput
Hi Team,
I'll be grateful, if you can help me on this.
Thanks!
Regards Renuka ________________________________ From: Arya, Renuka [BSD] - MED Sent: Friday, September 05, 2014 6:33 PM To: Observium Network Observation System Subject: The ports on the switches doesn't seems to display correct network throughput
Hi Team,
I installed Observium recently. It seems to be working just fine.
However, I do see discrepancies in the throughput result.
When I run a network performance tool (Nuttcp) on one of the server and monitor the corresponding port on the switch through observium. I observe that the the server is generating around 9.8 Gbps data. If I check the ports directly on the switch I see same thing (~9.8 Gbps) but when I see in through observium on the corresponding port , I see frequent spikes to 12 - 15 Gbps. I am not sure why I am seeing these high value because the NICs on servers are having a 10 G NICs and I never see anything higher than 10 Gbps on server or on the port on the switch. But the real time graphs on Observium shows the spikes of around 12 - 15 Gbps.
I would be really glad if you help me on this. I really like the tool but the discrepancy in the throughput result is kind of driving me crazy.
Thanks!
Regards Renuka
________________________________ This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
Protect Plus Air Holdings, LLC and Affiliates, Protect Plus, LLC and Affiliates, Imagine One Resources, LLC and Affiliates, Protect Plus Surfaces, LLC and Affiliates Confidentiality Notice: This e-mail is intended only for the addressee named above. It contains information that is privileged, confidential or otherwise protected from use and disclosure. If you are not the intended recipient, you are hereby notified that any review, disclosure, copying, or dissemination of this transmission, or taking of any action in reliance on its contents, or other use is strictly prohibited. If you have received this transmission in error, please reply to the sender listed above immediately and permanently delete this message from your inbox. Thank you for your cooperation.
Hi Josh,
1) I installed Observium on friday, hope it is the latest : http://www.observium.org/observium-community-latest.tar.gz
2) Ubuntu Version : It is installed on a virtual machine.
Distributor ID: Ubuntu Description: Ubuntu 12.04.2 LTS Release: 12.04 Codename: precise
3) It is open source NOT a paid version.
Thanks!
Regards Renuka
________________________________ From: observium [observium-bounces@observium.org] on behalf of Josh Hopper [joshhopper@protectplus.com] Sent: Tuesday, September 09, 2014 9:28 AM To: Observium Network Observation System Subject: Re: [Observium] The ports on the switches doesn't seems to display correct network throughput
Are you in the newest version of Observium? Did you install Observium on Ubuntu or Centos? Is it the paid version of Observium?
-- Sincerely, Joshua Hopper, A+ CE Network Administrator [http://www.protectplus.com/images/ppluslogo.jpg] 420 3rd Ave NW Hickory NC 28601 Office: 828-449-1839x2160 | Cell: 828-855-7565
From: observium [mailto:observium-bounces@observium.org] On Behalf Of Arya, Renuka [BSD] - MED Sent: Tuesday, September 09, 2014 10:25 AM To: Observium Network Observation System Subject: Re: [Observium] The ports on the switches doesn't seems to display correct network throughput
Hi Team,
I'll be grateful, if you can help me on this.
Thanks!
Regards Renuka ________________________________ From: Arya, Renuka [BSD] - MED Sent: Friday, September 05, 2014 6:33 PM To: Observium Network Observation System Subject: The ports on the switches doesn't seems to display correct network throughput
Hi Team,
I installed Observium recently. It seems to be working just fine.
However, I do see discrepancies in the throughput result.
When I run a network performance tool (Nuttcp) on one of the server and monitor the corresponding port on the switch through observium. I observe that the the server is generating around 9.8 Gbps data. If I check the ports directly on the switch I see same thing (~9.8 Gbps) but when I see in through observium on the corresponding port , I see frequent spikes to 12 - 15 Gbps. I am not sure why I am seeing these high value because the NICs on servers are having a 10 G NICs and I never see anything higher than 10 Gbps on server or on the port on the switch. But the real time graphs on Observium shows the spikes of around 12 - 15 Gbps.
I would be really glad if you help me on this. I really like the tool but the discrepancy in the throughput result is kind of driving me crazy.
Thanks!
Regards Renuka
________________________________ This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
Protect Plus Air Holdings, LLC and Affiliates, Protect Plus, LLC and Affiliates, Imagine One Resources, LLC and Affiliates, Protect Plus Surfaces, LLC and Affiliates Confidentiality Notice: This e-mail is intended only for the addressee named above. It contains information that is privileged, confidential or otherwise protected from use and disclosure. If you are not the intended recipient, you are hereby notified that any review, disclosure, copying, or dissemination of this transmission, or taking of any action in reliance on its contents, or other use is strictly prohibited. If you have received this transmission in error, please reply to the sender listed above immediately and permanently delete this message from your inbox. Thank you for your cooperation.
________________________________ This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
This is probably because your devices aren't capable of updating their counters quickly enough.
You should always include screenshots to explain what you're talking about, as it's not clear (and probably the reason you've not gotten many replies.)
adam.
On 2014-09-06 00:33, Arya, Renuka [BSD] - MED wrote:
Hi Team,
I installed Observium recently. It seems to be working just fine.
However, I do see discrepancies in the throughput result.
When I run a network performance tool (Nuttcp) on one of the server and monitor the corresponding port on the switch through observium. I observe that the the server is generating around 9.8 Gbps data. If I check the ports directly on the switch I see same thing (~9.8 Gbps) but when I see in through observium on the corresponding port , I see frequent spikes to 12 - 15 Gbps. I am not sure why I am seeing these high value because the NICs on servers are having a 10 G NICs and I never see anything higher than 10 Gbps on server or on the port on the switch. But the real time graphs on Observium shows the spikes of around 12 - 15 Gbps.
I would be really glad if you help me on this. I really like the tool but the discrepancy in the throughput result is kind of driving me crazy.
Thanks!
Regards Renuka
This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you. _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Thanks Adam and really sorry for nit elaborating before.
To elaborate better, I have two servers (Each having a 10G NIC) connected to one of my switch (Arista 7050S-64R, running 4.13.6F ) ports on ports eth4 and eth2.
I run Nuttcp (network performance tool) from on of the server (eth4) to another server (eth2) .
Please find the snapshot: 1) Nuttcp_OnSender : This shows nuttcp -T1000 -i5 192.168.60.3 , which means the sender is sending 1000 TCP packets to receiver (192.168.60.3) with an interval of 5 secs. You can see the average throughput ~ 9.8 Gbps.
2) Throughput_onSwitchPort : This shows - show int e 4 counters rates . We are manually monitoring traffic on the switch on that particular port (Ethernet 2). It is ~ 9.9 Gbps, inline with what we see on the server.
3) RealTimeThroughput_Observium: This monitors the real time throughput on eth 4 on Observium. There we can see the frequent spikes of 12.5 - 12.8 Gbps
*All the metrics are taken for an interval of 5 seconds.
I do not understand, why we see spikes as high as 12.8 Gbps when the server never seems (or can never) push data more than 10 Gbps as it is limited by a 10 Gbps NIC. If I decrease the polling interval further say 2, 1 or 0.25 seconds, I see spikes upto 15.8 Gbps for the same network traffic (Nuttcp on server running at eth port 4 on the switch). See the last snapshot : Observium_0.25ms.
Looks like the number are definitely not getting reported correctly.
Hope I have made it a little clearer now!
It would be great if you an give me some inputs. Not pretty sure what can be done on the switch. Other thing is, Observium is running on a Virtual Machine but don't think it should be an issue.
Thanks!
Regards Renuka
________________________________________ From: observium [observium-bounces@observium.org] on behalf of Adam Armstrong [adama@memetic.org] Sent: Tuesday, September 09, 2014 1:42 PM To: Observium Network Observation System Subject: Re: [Observium] The ports on the switches doesn't seems to display correct network throughput
This is probably because your devices aren't capable of updating their counters quickly enough.
You should always include screenshots to explain what you're talking about, as it's not clear (and probably the reason you've not gotten many replies.)
adam.
On 2014-09-06 00:33, Arya, Renuka [BSD] - MED wrote:
Hi Team,
I installed Observium recently. It seems to be working just fine.
However, I do see discrepancies in the throughput result.
When I run a network performance tool (Nuttcp) on one of the server and monitor the corresponding port on the switch through observium. I observe that the the server is generating around 9.8 Gbps data. If I check the ports directly on the switch I see same thing (~9.8 Gbps) but when I see in through observium on the corresponding port , I see frequent spikes to 12 - 15 Gbps. I am not sure why I am seeing these high value because the NICs on servers are having a 10 G NICs and I never see anything higher than 10 Gbps on server or on the port on the switch. But the real time graphs on Observium shows the spikes of around 12 - 15 Gbps.
I would be really glad if you help me on this. I really like the tool but the discrepancy in the throughput result is kind of driving me crazy.
Thanks!
Regards Renuka
This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you. _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
_______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
________________________________ This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
The numbers are definitely being reported correctly, its simply that the devices aren't updating the counters quickly enough, showing spikes.
This is fairly common when you have a high polling rate and periodically updated counters.
I'd say only 50% of devices are capable of accurately providing counters with resolution greater than 10 seconds or so.
Adam.
Sent with AquaMail for Android http://www.aqua-mail.com
On 9 September 2014 23:41:45 "Arya, Renuka [BSD] - MED" rarya@medicine.bsd.uchicago.edu wrote:
Thanks Adam and really sorry for nit elaborating before.
To elaborate better, I have two servers (Each having a 10G NIC) connected to one of my switch (Arista 7050S-64R, running 4.13.6F ) ports on ports eth4 and eth2.
I run Nuttcp (network performance tool) from on of the server (eth4) to another server (eth2) .
Please find the snapshot:
- Nuttcp_OnSender : This shows nuttcp -T1000 -i5 192.168.60.3 , which
means the sender is sending 1000 TCP packets to receiver (192.168.60.3) with an interval of 5 secs. You can see the average throughput ~ 9.8 Gbps.
- Throughput_onSwitchPort : This shows - show int e 4 counters rates . We
are manually monitoring traffic on the switch on that particular port (Ethernet 2). It is ~ 9.9 Gbps, inline with what we see on the server.
- RealTimeThroughput_Observium: This monitors the real time throughput on
eth 4 on Observium. There we can see the frequent spikes of 12.5 - 12.8 Gbps
*All the metrics are taken for an interval of 5 seconds.
I do not understand, why we see spikes as high as 12.8 Gbps when the server never seems (or can never) push data more than 10 Gbps as it is limited by a 10 Gbps NIC. If I decrease the polling interval further say 2, 1 or 0.25 seconds, I see spikes upto 15.8 Gbps for the same network traffic (Nuttcp on server running at eth port 4 on the switch). See the last snapshot : Observium_0.25ms.
Looks like the number are definitely not getting reported correctly.
Hope I have made it a little clearer now!
It would be great if you an give me some inputs. Not pretty sure what can be done on the switch. Other thing is, Observium is running on a Virtual Machine but don't think it should be an issue.
Thanks!
Regards Renuka
From: observium [observium-bounces@observium.org] on behalf of Adam Armstrong [adama@memetic.org] Sent: Tuesday, September 09, 2014 1:42 PM To: Observium Network Observation System Subject: Re: [Observium] The ports on the switches doesn't seems to display correct network throughput
This is probably because your devices aren't capable of updating their counters quickly enough.
You should always include screenshots to explain what you're talking about, as it's not clear (and probably the reason you've not gotten many replies.)
adam.
On 2014-09-06 00:33, Arya, Renuka [BSD] - MED wrote:
Hi Team,
I installed Observium recently. It seems to be working just fine.
However, I do see discrepancies in the throughput result.
When I run a network performance tool (Nuttcp) on one of the server and monitor the corresponding port on the switch through observium. I observe that the the server is generating around 9.8 Gbps data. If I check the ports directly on the switch I see same thing (~9.8 Gbps) but when I see in through observium on the corresponding port , I see frequent spikes to 12 - 15 Gbps. I am not sure why I am seeing these high value because the NICs on servers are having a 10 G NICs and I never see anything higher than 10 Gbps on server or on the port on the switch. But the real time graphs on Observium shows the spikes of around 12 - 15 Gbps.
I would be really glad if you help me on this. I really like the tool but the discrepancy in the throughput result is kind of driving me crazy.
Thanks!
Regards Renuka
This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you. _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Thanks Adam!
I was just wondering, I do not see this when I monitor those ports directly on the switch, as you can see in the snapshots I sent earlier. That polling is done on a 5 secs load interval.
Is there is a way to debug it better? I can see the higher spike issue in the FAQ section, probably should be looking into that.Or should I be contacting the Arista?
I really liked the real time network throughput feature on Observium and probably the one I'll use the maximum for our high speed networking environment. I really wish it to work correctly for me :(
Thanks!
Regards Renuka
Sent from my iPhone
On Sep 9, 2014, at 18:39, "Adam Armstrong" adama@memetic.org wrote:
The numbers are definitely being reported correctly, its simply that the devices aren't updating the counters quickly enough, showing spikes.
This is fairly common when you have a high polling rate and periodically updated counters.
I'd say only 50% of devices are capable of accurately providing counters with resolution greater than 10 seconds or so.
Adam.
Sent with AquaMail for Android http://www.aqua-mail.com
On 9 September 2014 23:41:45 "Arya, Renuka [BSD] - MED" rarya@medicine.bsd.uchicago.edu wrote:
Thanks Adam and really sorry for nit elaborating before.
To elaborate better, I have two servers (Each having a 10G NIC) connected to one of my switch (Arista 7050S-64R, running 4.13.6F ) ports on ports eth4 and eth2.
I run Nuttcp (network performance tool) from on of the server (eth4) to another server (eth2) .
Please find the snapshot:
Nuttcp_OnSender : This shows nuttcp -T1000 -i5 192.168.60.3 , which means the sender is sending 1000 TCP packets to receiver (192.168.60.3) with an interval of 5 secs. You can see the average throughput ~ 9.8 Gbps.
Throughput_onSwitchPort : This shows - show int e 4 counters rates . We are manually monitoring traffic on the switch on that particular port (Ethernet 2). It is ~ 9.9 Gbps, inline with what we see on the server.
RealTimeThroughput_Observium: This monitors the real time throughput on eth 4 on Observium. There we can see the frequent spikes of 12.5 - 12.8 Gbps
*All the metrics are taken for an interval of 5 seconds.
I do not understand, why we see spikes as high as 12.8 Gbps when the server never seems (or can never) push data more than 10 Gbps as it is limited by a 10 Gbps NIC. If I decrease the polling interval further say 2, 1 or 0.25 seconds, I see spikes upto 15.8 Gbps for the same network traffic (Nuttcp on server running at eth port 4 on the switch). See the last snapshot : Observium_0.25ms.
Looks like the number are definitely not getting reported correctly.
Hope I have made it a little clearer now!
It would be great if you an give me some inputs. Not pretty sure what can be done on the switch. Other thing is, Observium is running on a Virtual Machine but don't think it should be an issue.
Thanks!
Regards Renuka
From: observium [observium-bounces@observium.org] on behalf of Adam Armstrong [adama@memetic.org] Sent: Tuesday, September 09, 2014 1:42 PM To: Observium Network Observation System Subject: Re: [Observium] The ports on the switches doesn't seems to display correct network throughput
This is probably because your devices aren't capable of updating their counters quickly enough.
You should always include screenshots to explain what you're talking about, as it's not clear (and probably the reason you've not gotten many replies.)
adam.
On 2014-09-06 00:33, Arya, Renuka [BSD] - MED wrote:
Hi Team,
I installed Observium recently. It seems to be working just fine.
However, I do see discrepancies in the throughput result.
When I run a network performance tool (Nuttcp) on one of the server and monitor the corresponding port on the switch through observium. I observe that the the server is generating around 9.8 Gbps data. If I check the ports directly on the switch I see same thing (~9.8 Gbps) but when I see in through observium on the corresponding port , I see frequent spikes to 12 - 15 Gbps. I am not sure why I am seeing these high value because the NICs on servers are having a 10 G NICs and I never see anything higher than 10 Gbps on server or on the port on the switch. But the real time graphs on Observium shows the spikes of around 12 - 15 Gbps.
I would be really glad if you help me on this. I really like the tool but the discrepancy in the throughput result is kind of driving me crazy.
Thanks!
Regards Renuka
This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you. _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
________________________________ This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
It works directly on the switch because the switch is knows when the counters were last updated so can calculate a rate correctly.
There isn't really any debugging to be done. We know why this happens, it is expected behaviour and it can't be fixed.
Adam.
Sent with AquaMail for Android http://www.aqua-mail.com
On 10 September 2014 01:30:20 "Arya, Renuka [BSD] - MED" rarya@medicine.bsd.uchicago.edu wrote:
Thanks Adam!
I was just wondering, I do not see this when I monitor those ports directly on the switch, as you can see in the snapshots I sent earlier. That polling is done on a 5 secs load interval.
Is there is a way to debug it better? I can see the higher spike issue in the FAQ section, probably should be looking into that.Or should I be contacting the Arista?
I really liked the real time network throughput feature on Observium and probably the one I'll use the maximum for our high speed networking environment. I really wish it to work correctly for me :(
Thanks!
Regards Renuka
Sent from my iPhone
On Sep 9, 2014, at 18:39, "Adam Armstrong" adama@memetic.org wrote:
The numbers are definitely being reported correctly, its simply that the
devices aren't updating the counters quickly enough, showing spikes.
This is fairly common when you have a high polling rate and periodically
updated counters.
I'd say only 50% of devices are capable of accurately providing counters
with resolution greater than 10 seconds or so.
Adam.
Sent with AquaMail for Android http://www.aqua-mail.com
On 9 September 2014 23:41:45 "Arya, Renuka [BSD] - MED"
rarya@medicine.bsd.uchicago.edu wrote:
Thanks Adam and really sorry for nit elaborating before.
To elaborate better, I have two servers (Each having a 10G NIC)
connected to one of my switch (Arista 7050S-64R, running 4.13.6F ) ports on ports eth4 and eth2.
I run Nuttcp (network performance tool) from on of the server (eth4) to
another server (eth2) .
Please find the snapshot:
- Nuttcp_OnSender : This shows nuttcp -T1000 -i5 192.168.60.3 , which
means the sender is sending 1000 TCP packets to receiver (192.168.60.3) with an interval of 5 secs. You can see the average throughput ~ 9.8 Gbps.
- Throughput_onSwitchPort : This shows - show int e 4 counters rates .
We are manually monitoring traffic on the switch on that particular port (Ethernet 2). It is ~ 9.9 Gbps, inline with what we see on the server.
- RealTimeThroughput_Observium: This monitors the real time throughput
on eth 4 on Observium. There we can see the frequent spikes of 12.5 - 12.8 Gbps
*All the metrics are taken for an interval of 5 seconds.
I do not understand, why we see spikes as high as 12.8 Gbps when the
server never seems (or can never) push data more than 10 Gbps as it is limited by a 10 Gbps NIC. If I decrease the polling interval further say 2, 1 or 0.25 seconds, I see spikes upto 15.8 Gbps for the same network traffic (Nuttcp on server running at eth port 4 on the switch). See the last snapshot : Observium_0.25ms.
Looks like the number are definitely not getting reported correctly.
Hope I have made it a little clearer now!
It would be great if you an give me some inputs. Not pretty sure what
can be done on the switch. Other thing is, Observium is running on a Virtual Machine but don't think it should be an issue.
Thanks!
Regards Renuka
From: observium [observium-bounces@observium.org] on behalf of Adam
Armstrong [adama@memetic.org]
Sent: Tuesday, September 09, 2014 1:42 PM To: Observium Network Observation System Subject: Re: [Observium] The ports on the switches doesn't seems to
display correct network throughput
This is probably because your devices aren't capable of updating their counters quickly enough.
You should always include screenshots to explain what you're talking about, as it's not clear (and probably the reason you've not gotten many replies.)
adam.
On 2014-09-06 00:33, Arya, Renuka [BSD] - MED wrote:
Hi Team,
I installed Observium recently. It seems to be working just fine.
However, I do see discrepancies in the throughput result.
When I run a network performance tool (Nuttcp) on one of the server and monitor the corresponding port on the switch through observium. I observe that the the server is generating around 9.8 Gbps data. If I check the ports directly on the switch I see same thing (~9.8 Gbps) but when I see in through observium on the corresponding port , I see frequent spikes to 12 - 15 Gbps. I am not sure why I am seeing these high value because the NICs on servers are having a 10 G NICs and I never see anything higher than 10 Gbps on server or on the port on the switch. But the real time graphs on Observium shows the spikes of around 12 - 15 Gbps.
I would be really glad if you help me on this. I really like the tool but the discrepancy in the throughput result is kind of driving me crazy.
Thanks!
Regards Renuka
This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you. _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
This email is intended only for the use of the individual or entity to
which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you. _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
Observium is simply polling SNMP data. To see what Adam is talking about, redo your test and run something like this command during the test:
bash$ while :; do time +%T.%N; snmpget <INSERT SNMP OPTIONS HERE>; done
Q.E.D. -- Cameron Moore
-----Original Message----- From: observium [mailto:observium-bounces@observium.org] On Behalf Of Arya, Renuka [BSD] - MED Sent: Tuesday, September 09, 2014 7:29 PM To: Observium Network Observation System Subject: Re: [Observium] The ports on the switches doesn't seems to display correct network throughput
Thanks Adam!
I was just wondering, I do not see this when I monitor those ports directly on the switch, as you can see in the snapshots I sent earlier. That polling is done on a 5 secs load interval.
Is there is a way to debug it better? I can see the higher spike issue in the FAQ section, probably should be looking into that.Or should I be contacting the Arista?
I really liked the real time network throughput feature on Observium and probably the one I'll use the maximum for our high speed networking environment. I really wish it to work correctly for me :(
Thanks!
Regards Renuka
Sent from my iPhone
On Sep 9, 2014, at 18:39, "Adam Armstrong" adama@memetic.org wrote:
The numbers are definitely being reported correctly, its simply that the devices aren't updating the counters quickly enough, showing spikes.
This is fairly common when you have a high polling rate and periodically updated counters.
I'd say only 50% of devices are capable of accurately providing counters with resolution greater than 10 seconds or so.
Adam.
Sent with AquaMail for Android http://www.aqua-mail.com
On 9 September 2014 23:41:45 "Arya, Renuka [BSD] - MED" rarya@medicine.bsd.uchicago.edu wrote:
Thanks Adam and really sorry for nit elaborating before.
To elaborate better, I have two servers (Each having a 10G NIC) connected to one of my switch (Arista 7050S-64R, running 4.13.6F ) ports on ports eth4 and eth2.
I run Nuttcp (network performance tool) from on of the server (eth4) to another server (eth2) .
Please find the snapshot:
Nuttcp_OnSender : This shows nuttcp -T1000 -i5 192.168.60.3 , which means the sender is sending 1000 TCP packets to receiver (192.168.60.3) with an interval of 5 secs. You can see the average throughput ~ 9.8 Gbps.
Throughput_onSwitchPort : This shows - show int e 4 counters rates . We are manually monitoring traffic on the switch on that particular port (Ethernet 2). It is ~ 9.9 Gbps, inline with what we see on the server.
RealTimeThroughput_Observium: This monitors the real time
throughput on eth 4 on Observium. There we can see the frequent spikes of 12.5 - 12.8 Gbps
*All the metrics are taken for an interval of 5 seconds.
I do not understand, why we see spikes as high as 12.8 Gbps when the server never seems (or can never) push data more than 10 Gbps as it is limited by a 10 Gbps NIC. If I decrease the polling interval further say 2, 1 or 0.25 seconds, I see spikes upto 15.8 Gbps for the same network traffic (Nuttcp on server running at eth port 4 on the switch). See the last snapshot : Observium_0.25ms.
Looks like the number are definitely not getting reported correctly.
Hope I have made it a little clearer now!
It would be great if you an give me some inputs. Not pretty sure what can be done on the switch. Other thing is, Observium is running on a Virtual Machine but don't think it should be an issue.
Thanks!
Regards Renuka
From: observium [observium-bounces@observium.org] on behalf of Adam Armstrong [adama@memetic.org] Sent: Tuesday, September 09, 2014 1:42 PM To: Observium Network Observation System Subject: Re: [Observium] The ports on the switches doesn't seems to display correct network throughput
This is probably because your devices aren't capable of updating their counters quickly enough.
You should always include screenshots to explain what you're talking about, as it's not clear (and probably the reason you've not gotten many replies.)
adam.
On 2014-09-06 00:33, Arya, Renuka [BSD] - MED wrote:
Hi Team,
I installed Observium recently. It seems to be working just fine.
However, I do see discrepancies in the throughput result.
When I run a network performance tool (Nuttcp) on one of the server and monitor the corresponding port on the switch through observium. I observe that the the server is generating around 9.8 Gbps data. If I check the ports directly on the switch I see same thing (~9.8 Gbps) but when I see in through observium on the corresponding port , I see frequent spikes to 12 - 15 Gbps. I am not sure why I am seeing these high value because the NICs on servers are having a 10 G NICs and I never see anything higher than 10 Gbps on server or on the port on the switch. But the real time graphs on Observium shows the spikes of around 12 - 15 Gbps.
I would be really glad if you help me on this. I really like the tool but the discrepancy in the throughput result is kind of driving me crazy.
Thanks!
Regards Renuka
This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you. _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you.
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
________________________________ This email is intended only for the use of the individual or entity to which it is addressed and may contain information that is privileged and confidential. If the reader of this email message is not the intended recipient, you are hereby notified that any dissemination, distribution, or copying of this communication is prohibited. If you have received this email in error, please notify the sender and destroy/delete all copies of the transmittal.
Thank you. _______________________________________________ observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium
participants (4)
-
Adam Armstrong -
Arya, Renuka [BSD] - MED -
Josh Hopper -
Moore, Cameron