Mailing List Archive: SNMP very slow on MX10003

SNMP very slow on MX10003

May 25, 2020, 9:25 AM

Post #1 of 5 (1346 views)

Hello,

we meter approximately 500 logical subinterface counters (e.g. ae0.95,
ae0.96, etc) of our Juniper MX10003. While this works it is also too
slow as we fetch the data every 60 seconds.

snmpbulkwalks of IF-MIB::ifHCInOctets and IF-MIB::ifHCOutOctets take
anything from 1 to 180 seconds.

The time it takes increases noticeably as more subinterfaces are configured.

I know Juniper is not known for fast SNMP queries, but is there
anything you can do to increase performance? 500 interfaces should
really not be too much for a fairly new router imho.
We already have some interface filters in place, but apparently they
are not sufficient.

If you think that SNMP is too slow, what alternative can you recommend
for fetching interface counters?

Thank you for your help.

Kind Regards
Leon Kramer

> show version
Model: mx10003
Junos: 18.4R2-S3

> show configuration snmp
interface [ fxp0.0 ];
filter-interfaces {
interfaces {
"fe-[0-9]/[0-9]/[0-9]+\.[0-9]+";
"xe-[0-9]/[0-9]/[0-9]+\.[0-9]+";
"ge-[0-9]/[0-9]/[0-9]+\.[0-9]+";
"et-[0-9]/[0-9]/[0-9]+\.[0-9]+";
}
all-internal-interfaces;
}
filter-duplicates;
community xxx {
authorization read-only;
routing-instance mgmt_junos;
}
community xxx {
authorization read-only;
routing-instance mgmt_junos;
}
community xxx {
authorization read-only;
routing-instance mgmt_junos;
}
community xxx {
authorization read-only;
routing-instance mgmt_junos;
}
routing-instance-access;
_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp

Re: SNMP very slow on MX10003 [ In reply to ]

david.roy at orange

May 25, 2020, 9:32 AM

Post #2 of 5 (1346 views)

Permalink

Hello

Could you issue this command & share the output : show snmp stats-response-statistics

Otherwise, we use telemetry to collect, among other things, interface stats.

David

-----Message d'origine-----
De?: juniper-nsp [mailto:juniper-nsp-bounces@puck.nether.net] De la part de Leon Kramer
Envoy??: lundi 25 mai 2020 18:26
??: juniper-nsp@puck.nether.net
Objet?: [j-nsp] SNMP very slow on MX10003

Hello,

we meter approximately 500 logical subinterface counters (e.g. ae0.95,
ae0.96, etc) of our Juniper MX10003. While this works it is also too
slow as we fetch the data every 60 seconds.

snmpbulkwalks of IF-MIB::ifHCInOctets and IF-MIB::ifHCOutOctets take
anything from 1 to 180 seconds.

The time it takes increases noticeably as more subinterfaces are configured.

I know Juniper is not known for fast SNMP queries, but is there
anything you can do to increase performance? 500 interfaces should
really not be too much for a fairly new router imho.
We already have some interface filters in place, but apparently they
are not sufficient.

If you think that SNMP is too slow, what alternative can you recommend
for fetching interface counters?

Thank you for your help.

Kind Regards
Leon Kramer

> show version
Model: mx10003
Junos: 18.4R2-S3

> show configuration snmp
interface [ fxp0.0 ];
filter-interfaces {
interfaces {
"fe-[0-9]/[0-9]/[0-9]+\.[0-9]+";
"xe-[0-9]/[0-9]/[0-9]+\.[0-9]+";
"ge-[0-9]/[0-9]/[0-9]+\.[0-9]+";
"et-[0-9]/[0-9]/[0-9]+\.[0-9]+";
}
all-internal-interfaces;
}
filter-duplicates;
community xxx {
authorization read-only;
routing-instance mgmt_junos;
}
community xxx {
authorization read-only;
routing-instance mgmt_junos;
}
community xxx {
authorization read-only;
routing-instance mgmt_junos;
}
community xxx {
authorization read-only;
routing-instance mgmt_junos;
}
routing-instance-access;
_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp

_________________________________________________________________________________________________________________________

Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc
pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler
a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,
Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.

This message and its attachments may contain confidential or privileged information that may be protected by law;
they should not be distributed, used or copied without authorisation.
If you have received this email in error, please notify the sender and delete this message and its attachments.
As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.
Thank you.

_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp

Re: SNMP very slow on MX10003 [ In reply to ]

lists at tobias-heister

May 25, 2020, 9:36 AM

Post #3 of 5 (1346 views)

Permalink

Hi,

On 25.05.2020 18:25, Leon Kramer wrote:
> If you think that SNMP is too slow, what alternative can you recommend
> for fetching interface counters?

Probably the canonical answer these days is streaming telemetry which can be done directly from the PFE for a lot of counters.

In a former life when SNMP got too slow we sometimes switches to screenscraping "show interfaces | display XML" via SSH/netconf. But this would only help if the data itself is presented fast enough. We had boxes where presenting thh data itself towards the cli/SNMP agent was the bottleneck so the method of pulling it made no difference.

--
Kind Regards
Tobias Heister
_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp

Re: SNMP very slow on MX10003 [ In reply to ]

leonkramer at gmail

May 25, 2020, 9:42 AM

Post #4 of 5 (1346 views)

Permalink

> show snmp stats-response-statistics

Average response time statistics:
Stats Stats Average
Type Responses Response
Time (ms)
ifd(non ae) 15311162 5.60
ifd(ae) 482105 34.68
ifl(non ae) 5930128 5.96
ifl(ae) 74751690 8.80
firewall 66199433 10.87

Bucket statistics:
Bucket Stats
Type(ms) Responses
0 - 10 117512092
11 - 50 36891316
51 - 100 3645978
101 - 200 1647536
201 - 500 1330240
501 - 1000 889495
1001 - 2000 707479
2001 - 5000 40989
More than 5001 9393

Bad responses:
Response Request Stats Key
Time Type
(ms) (UTC)
14048.00 2020-04-30 02:26:36 ifl(ae) 713
13649.32 2020-05-06 16:50:12 ifl(ae) 1319
13649.03 2020-05-06 16:50:12 ifl(ae) 1320
13648.82 2020-05-06 16:50:12 ifl(ae) 1321
13648.59 2020-05-06 16:50:12 ifl(ae) 1322
13648.38 2020-05-06 16:50:12 ifl(ae) 1323
13647.94 2020-05-06 16:50:12 ifl(ae) 1324
13647.72 2020-05-06 16:50:12 ifl(ae) 1325
12174.38 2020-05-19 07:08:22 ifl(ae) 1517
12145.67 2020-05-06 16:53:38 ifl(ae) 3743
12145.42 2020-05-06 16:53:38 ifl(ae) 3744
12145.08 2020-05-06 16:53:38 ifl(ae) 3745
12144.83 2020-05-06 16:53:38 ifl(ae) 3786
12144.52 2020-05-06 16:53:38 ifl(ae) 3787
12144.27 2020-05-06 16:53:38 ifl(ae) 3788
12144.05 2020-05-06 16:53:38 ifl(ae) 3789
12143.77 2020-05-06 16:53:38 ifl(ae) 3790
12143.49 2020-05-06 16:53:38 ifl(ae) 3791
12143.22 2020-05-06 16:53:38 ifl(ae) 3792
12143.00 2020-05-06 16:53:38 ifl(ae) 3793

Am Mo., 25. Mai 2020 um 18:32 Uhr schrieb <david.roy@orange.com>:
>
> Hello
>
> Could you issue this command & share the output : show snmp stats-response-statistics
>
> Otherwise, we use telemetry to collect, among other things, interface stats.
>
> David
>
>
> -----Message d'origine-----
> De : juniper-nsp [mailto:juniper-nsp-bounces@puck.nether.net] De la part de Leon Kramer
> Envoyé : lundi 25 mai 2020 18:26
> À : juniper-nsp@puck.nether.net
> Objet : [j-nsp] SNMP very slow on MX10003
>
> Hello,
>
> we meter approximately 500 logical subinterface counters (e.g. ae0.95,
> ae0.96, etc) of our Juniper MX10003. While this works it is also too
> slow as we fetch the data every 60 seconds.
>
> snmpbulkwalks of IF-MIB::ifHCInOctets and IF-MIB::ifHCOutOctets take
> anything from 1 to 180 seconds.
>
> The time it takes increases noticeably as more subinterfaces are configured.
>
> I know Juniper is not known for fast SNMP queries, but is there
> anything you can do to increase performance? 500 interfaces should
> really not be too much for a fairly new router imho.
> We already have some interface filters in place, but apparently they
> are not sufficient.
>
> If you think that SNMP is too slow, what alternative can you recommend
> for fetching interface counters?
>
> Thank you for your help.
>
>
> Kind Regards
> Leon Kramer
>
>
> > show version
> Model: mx10003
> Junos: 18.4R2-S3
>
>
> > show configuration snmp
> interface [ fxp0.0 ];
> filter-interfaces {
> interfaces {
> "fe-[0-9]/[0-9]/[0-9]+\.[0-9]+";
> "xe-[0-9]/[0-9]/[0-9]+\.[0-9]+";
> "ge-[0-9]/[0-9]/[0-9]+\.[0-9]+";
> "et-[0-9]/[0-9]/[0-9]+\.[0-9]+";
> }
> all-internal-interfaces;
> }
> filter-duplicates;
> community xxx {
> authorization read-only;
> routing-instance mgmt_junos;
> }
> community xxx {
> authorization read-only;
> routing-instance mgmt_junos;
> }
> community xxx {
> authorization read-only;
> routing-instance mgmt_junos;
> }
> community xxx {
> authorization read-only;
> routing-instance mgmt_junos;
> }
> routing-instance-access;
> _______________________________________________
> juniper-nsp mailing list juniper-nsp@puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp
>
> _________________________________________________________________________________________________________________________
>
> Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc
> pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler
> a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,
> Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.
>
> This message and its attachments may contain confidential or privileged information that may be protected by law;
> they should not be distributed, used or copied without authorisation.
> If you have received this email in error, please notify the sender and delete this message and its attachments.
> As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.
> Thank you.
>
_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp

Re: SNMP very slow on MX10003 [ In reply to ]

david.roy at orange

May 25, 2020, 10:27 AM

Post #5 of 5 (1343 views)

Permalink

You hit sometimes high response time ( > 14 secs) to collect stats of AE's units. Could you correlate this timestamp with significant changes on the router of routing events ?

-----Message d'origine-----
De : Leon Kramer [mailto:leonkramer@gmail.com]
Envoyé : lundi 25 mai 2020 18:42
À : ROY David DTSI/DERS; juniper-nsp@puck.nether.net
Objet : Re: [j-nsp] SNMP very slow on MX10003

> show snmp stats-response-statistics

Average response time statistics:
Stats Stats Average
Type Responses Response
Time (ms)
ifd(non ae) 15311162 5.60
ifd(ae) 482105 34.68
ifl(non ae) 5930128 5.96
ifl(ae) 74751690 8.80
firewall 66199433 10.87

Bucket statistics:
Bucket Stats
Type(ms) Responses
0 - 10 117512092
11 - 50 36891316
51 - 100 3645978
101 - 200 1647536
201 - 500 1330240
501 - 1000 889495
1001 - 2000 707479
2001 - 5000 40989
More than 5001 9393

Bad responses:
Response Request Stats Key
Time Type
(ms) (UTC)
14048.00 2020-04-30 02:26:36 ifl(ae) 713
13649.32 2020-05-06 16:50:12 ifl(ae) 1319
13649.03 2020-05-06 16:50:12 ifl(ae) 1320
13648.82 2020-05-06 16:50:12 ifl(ae) 1321
13648.59 2020-05-06 16:50:12 ifl(ae) 1322
13648.38 2020-05-06 16:50:12 ifl(ae) 1323
13647.94 2020-05-06 16:50:12 ifl(ae) 1324
13647.72 2020-05-06 16:50:12 ifl(ae) 1325
12174.38 2020-05-19 07:08:22 ifl(ae) 1517
12145.67 2020-05-06 16:53:38 ifl(ae) 3743
12145.42 2020-05-06 16:53:38 ifl(ae) 3744
12145.08 2020-05-06 16:53:38 ifl(ae) 3745
12144.83 2020-05-06 16:53:38 ifl(ae) 3786
12144.52 2020-05-06 16:53:38 ifl(ae) 3787
12144.27 2020-05-06 16:53:38 ifl(ae) 3788
12144.05 2020-05-06 16:53:38 ifl(ae) 3789
12143.77 2020-05-06 16:53:38 ifl(ae) 3790
12143.49 2020-05-06 16:53:38 ifl(ae) 3791
12143.22 2020-05-06 16:53:38 ifl(ae) 3792
12143.00 2020-05-06 16:53:38 ifl(ae) 3793

Am Mo., 25. Mai 2020 um 18:32 Uhr schrieb <david.roy@orange.com>:
>
> Hello
>
> Could you issue this command & share the output : show snmp stats-response-statistics
>
> Otherwise, we use telemetry to collect, among other things, interface stats.
>
> David
>
>
> -----Message d'origine-----
> De : juniper-nsp [mailto:juniper-nsp-bounces@puck.nether.net] De la part de Leon Kramer
> Envoyé : lundi 25 mai 2020 18:26
> À : juniper-nsp@puck.nether.net
> Objet : [j-nsp] SNMP very slow on MX10003
>
> Hello,
>
> we meter approximately 500 logical subinterface counters (e.g. ae0.95,
> ae0.96, etc) of our Juniper MX10003. While this works it is also too
> slow as we fetch the data every 60 seconds.
>
> snmpbulkwalks of IF-MIB::ifHCInOctets and IF-MIB::ifHCOutOctets take
> anything from 1 to 180 seconds.
>
> The time it takes increases noticeably as more subinterfaces are configured.
>
> I know Juniper is not known for fast SNMP queries, but is there
> anything you can do to increase performance? 500 interfaces should
> really not be too much for a fairly new router imho.
> We already have some interface filters in place, but apparently they
> are not sufficient.
>
> If you think that SNMP is too slow, what alternative can you recommend
> for fetching interface counters?
>
> Thank you for your help.
>
>
> Kind Regards
> Leon Kramer
>
>
> > show version
> Model: mx10003
> Junos: 18.4R2-S3
>
>
> > show configuration snmp
> interface [ fxp0.0 ];
> filter-interfaces {
> interfaces {
> "fe-[0-9]/[0-9]/[0-9]+\.[0-9]+";
> "xe-[0-9]/[0-9]/[0-9]+\.[0-9]+";
> "ge-[0-9]/[0-9]/[0-9]+\.[0-9]+";
> "et-[0-9]/[0-9]/[0-9]+\.[0-9]+";
> }
> all-internal-interfaces;
> }
> filter-duplicates;
> community xxx {
> authorization read-only;
> routing-instance mgmt_junos;
> }
> community xxx {
> authorization read-only;
> routing-instance mgmt_junos;
> }
> community xxx {
> authorization read-only;
> routing-instance mgmt_junos;
> }
> community xxx {
> authorization read-only;
> routing-instance mgmt_junos;
> }
> routing-instance-access;
> _______________________________________________
> juniper-nsp mailing list juniper-nsp@puck.nether.net
> https://puck.nether.net/mailman/listinfo/juniper-nsp
>
> _________________________________________________________________________________________________________________________
>
> Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc
> pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler
> a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,
> Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.
>
> This message and its attachments may contain confidential or privileged information that may be protected by law;
> they should not be distributed, used or copied without authorisation.
> If you have received this email in error, please notify the sender and delete this message and its attachments.
> As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.
> Thank you.
>

_________________________________________________________________________________________________________________________

Ce message et ses pieces jointes peuvent contenir des informations confidentielles ou privilegiees et ne doivent donc
pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu ce message par erreur, veuillez le signaler
a l'expediteur et le detruire ainsi que les pieces jointes. Les messages electroniques etant susceptibles d'alteration,
Orange decline toute responsabilite si ce message a ete altere, deforme ou falsifie. Merci.

This message and its attachments may contain confidential or privileged information that may be protected by law;
they should not be distributed, used or copied without authorisation.
If you have received this email in error, please notify the sender and delete this message and its attachments.
As emails may be altered, Orange is not liable for messages that have been modified, changed or falsified.
Thank you.

_______________________________________________
juniper-nsp mailing list juniper-nsp@puck.nether.net
https://puck.nether.net/mailman/listinfo/juniper-nsp