|
Home > Archive > alt.certification.cisco > March 2003 > High Network Utilization
You are viewing an archived Text-only version of the thread.
To view this thread in it's original format and/or if you want to reply to
this thread please [click here]
| Author |
High Network Utilization
|
|
|
| Can anyone help me with a strange situation that I am running into with my
network. Every hour we do a backup of some databases. It seems when those
backups (they last about 3 minutes) kick off all of the switch port max out.
It seems that all ports go to 95% and stay there until the backup is
complete. Has anyone seen this before?
thx, L
| |
| Hansang Bae 2003-03-24, 8:23 pm |
| In article <f1Gfa.4857$rw.3133@nwrddc01.gnilink.net>, lgovedi1
@tampabay.rr.com says...
> Can anyone help me with a strange situation that I am running into with my
> network. Every hour we do a backup of some databases. It seems when those
> backups (they last about 3 minutes) kick off all of the switch port max out.
> It seems that all ports go to 95% and stay there until the backup is
> complete. Has anyone seen this before?
Is your backup using multicasts/broadcasts? If multicasts, it may be
erroneously forwarding the packet and if it's a broadcast, you should
shoot the vendor.
--
hsb
"Somehow I imagined this experience would be more rewarding" Calvin
*************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
******************************
******************************
********
Due to the volume of email that I receive, I may not not be able to
reply to emails sent to my account. Please post a followup instead.
******************************
******************************
********
| |
| Bernie 2003-03-24, 11:24 pm |
| On Tue, 25 Mar 2003 02:13:06 GMT, Hansang Bae <uonr@alp.ee.pbz> wrote:
>In article <f1Gfa.4857$rw.3133@nwrddc01.gnilink.net>, lgovedi1
>@tampabay.rr.com says...
>> Can anyone help me with a strange situation that I am running into with my
>> network. Every hour we do a backup of some databases. It seems when those
>> backups (they last about 3 minutes) kick off all of the switch port max out.
>> It seems that all ports go to 95% and stay there until the backup is
>> complete. Has anyone seen this before?
>
>
>Is your backup using multicasts/broadcasts? If multicasts, it may be
>erroneously forwarding the packet
Or properly forwarding the packet. There are a number of load
balancing type products out there (like MS NLBS and Stonebeat) that
use Ethernet multicasts (note, this is not from the range reserved for
IP multicast) on the network. The proper behavior for this is to
treat it as broadcast (i.e. flood) unless the switch has the
capability of manually mapping a group address to certain ports.
Also, if it is IP multicast, without IGMP snooping enabled, you have
to flood. The only other option would be to drop packets when
snooping is disabled, but I doubt any vendor makes this the default
behavior. Don't know about Cisco either. I do know about the
specific L2 switches I am familiar with, and the behavior without
snooping is to flood.
>and if it's a broadcast, you should
>shoot the vendor.
--Bernie
| |
|
| I really appreciate your input on this. So far this is what I have done:
conf t
ip igmp snoop vlan 1
this seems to have corrected the broadcast/flood issue. Can anyone explain
what I have just done? I am still unclear as to what effect this has on our
network from a layer perspective. I really appreciate your guidence on this,
it has seemed to work so far...
Larry
"Hansang Bae" <uonr@alp.ee.pbz> wrote in message
news:MPG.18e985f5a800ae459898c6@news-server.nyc.rr.com...
> In article <f1Gfa.4857$rw.3133@nwrddc01.gnilink.net>, lgovedi1
> @tampabay.rr.com says...
> > Can anyone help me with a strange situation that I am running into with
my
> > network. Every hour we do a backup of some databases. It seems when
those
> > backups (they last about 3 minutes) kick off all of the switch port max
out.
> > It seems that all ports go to 95% and stay there until the backup is
> > complete. Has anyone seen this before?
>
>
> Is your backup using multicasts/broadcasts? If multicasts, it may be
> erroneously forwarding the packet and if it's a broadcast, you should
> shoot the vendor.
>
> --
>
> hsb
>
> "Somehow I imagined this experience would be more rewarding" Calvin
> *************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
> ******************************
******************************
********
> Due to the volume of email that I receive, I may not not be able to
> reply to emails sent to my account. Please post a followup instead.
> ******************************
******************************
********
| |
| Hansang Bae 2003-03-25, 7:24 pm |
| In article <Nx_fa.525$783.301@nwrddc01.gnilink.net>, lgovedi1
@tampabay.rr.com says...
> I really appreciate your input on this. So far this is what I have done:
>
> conf t
> ip igmp snoop vlan 1
>
> this seems to have corrected the broadcast/flood issue. Can anyone explain
> what I have just done? I am still unclear as to what effect this has on our
> network from a layer perspective. I really appreciate your guidence on this,
> it has seemed to work so far...
It watches the multicast frames and finds out who really needs to listen
to it. Only those ports will receive the frames.
--
hsb
"Somehow I imagined this experience would be more rewarding" Calvin
*************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
******************************
******************************
********
Due to the volume of email that I receive, I may not not be able to
reply to emails sent to my account. Please post a followup instead.
******************************
******************************
********
| |
|
| Hansang,
As it turns out the tape library was hung and no backups ran yesterday
morning when I was checking. So the problem was still there. If you come up
with any other ideas I would appreciate your input... Are there any
networking web sites etc. that might be able to provide input?
Thx, Larry
"Hansang Bae" <uonr@alp.ee.pbz> wrote in message
news:MPG.18eac082d4cb1b649898d2@news-server.nyc.rr.com...
> In article <Nx_fa.525$783.301@nwrddc01.gnilink.net>, lgovedi1
> @tampabay.rr.com says...
> > I really appreciate your input on this. So far this is what I have done:
> >
> > conf t
> > ip igmp snoop vlan 1
> >
> > this seems to have corrected the broadcast/flood issue. Can anyone
explain
> > what I have just done? I am still unclear as to what effect this has on
our
> > network from a layer perspective. I really appreciate your guidence on
this,
> > it has seemed to work so far...
>
> It watches the multicast frames and finds out who really needs to listen
> to it. Only those ports will receive the frames.
>
>
> --
>
> hsb
>
> "Somehow I imagined this experience would be more rewarding" Calvin
> *************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
> ******************************
******************************
********
> Due to the volume of email that I receive, I may not not be able to
> reply to emails sent to my account. Please post a followup instead.
> ******************************
******************************
********
>
| |
| Hansang Bae 2003-03-26, 4:24 pm |
| In article <dhjga.34073$j8.961632@twister.tampabay.rr.com>, lgovedi1
@tampabay.rr.com says...
> As it turns out the tape library was hung and no backups ran yesterday
> morning when I was checking. So the problem was still there. If you come up
> with any other ideas I would appreciate your input... Are there any
> networking web sites etc. that might be able to provide input?
You can try the vendor's website. But can you just stick a sniffer on
any of those ports to see what it is that you see?
--
hsb
"Somehow I imagined this experience would be more rewarding" Calvin
*************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
******************************
******************************
********
Due to the volume of email that I receive, I may not not be able to
reply to emails sent to my account. Please post a followup instead.
******************************
******************************
********
| |
| Bernie 2003-03-26, 9:24 pm |
| On Wed, 26 Mar 2003 15:01:29 GMT, "LWG" <lgovedi1@tampabay.rr.com>
wrote:
>Hansang,
>
> As it turns out the tape library was hung and no backups ran yesterday
>morning when I was checking. So the problem was still there. If you come up
>with any other ideas I would appreciate your input... Are there any
>networking web sites etc. that might be able to provide input?
It's time to sniff. Specifically look at the destination MAC address
of the packets.
I forgot to mention that even if it is IP multicast, flooding can
still be the normal behavior even when snooping is enabled.
Specifically, some coders don't join their own groups, and it is the
proper behavior for a L2 switch to flood such traffic (i.e. traffic
not coming from a router port). The only spec that specifically
defines snooping is the IGMPv3 draft and this is the way it is
written. Some vendors allow you to disable flooding under these
circumstances.
>Thx, Larry
>"Hansang Bae" <uonr@alp.ee.pbz> wrote in message
>news:MPG.18eac082d4cb1b649898d2@news-server.nyc.rr.com...
>> In article <Nx_fa.525$783.301@nwrddc01.gnilink.net>, lgovedi1
>> @tampabay.rr.com says...
>> > I really appreciate your input on this. So far this is what I have done:
>> >
>> > conf t
>> > ip igmp snoop vlan 1
>> >
>> > this seems to have corrected the broadcast/flood issue. Can anyone
> explain
>> > what I have just done? I am still unclear as to what effect this has on
> our
>> > network from a layer perspective. I really appreciate your guidence on
> this,
>> > it has seemed to work so far...
>>
>> It watches the multicast frames and finds out who really needs to listen
>> to it. Only those ports will receive the frames.
>>
>>
>> --
>>
>> hsb
>>
>> "Somehow I imagined this experience would be more rewarding" Calvin
>> *************** USE ROT13 TO SEE MY EMAIL ADDRESS ****************
>> ******************************
******************************
********
>> Due to the volume of email that I receive, I may not not be able to
>> reply to emails sent to my account. Please post a followup instead.
>> ******************************
******************************
********
>>
>
--Bernie
|
|
|
|
|