[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: [PWE3] RE: Alignment of TotalPerf tables in PW MIBs
We can define the day Interval table and current day to align with
15min. Both (day tables) can be optional and the number of recommended
intervals can be 1, 7, 31 depending on agent resources and capabilities.
The 1st interval will old the previous day, if implemented.
-----Original Message-----
From: pwe3-bounces at ietf.org [mailto:pwe3-bounces at ietf.org] On Behalf Of
David Zelig
Sent: Wednesday, October 26, 2005 17:57
To: Ron Cohen; Orly Nicklass
Cc: Thomas D. Nadeau; Doron Tzur; pwe3 at ietf.org
Subject: RE: [PWE3] RE: Alignment of TotalPerf tables in PW MIBs
Ron and Orly,
I agree that the total tables are either redundant or problematic for
the implementation, however do not think that all applications would
require the day intervals, so it should be optional.
In addition, there is a need for previous day table since just after the
end of the day interval, you need to be able to read the previous day
performance.
David
-----Original Message-----
From: Ron Cohen [mailto:ronc at resolutenetworks.com]
Sent: Wednesday, October 26, 2005 2:14 PM
To: Orly Nicklass; Thomas D. Nadeau
Cc: pwe3 at ietf.org; Doron Tzur; David Zelig
Subject: RE: [PWE3] RE: Alignment of TotalPerf tables in PW MIBs
Orly,
I think your proposal to align to RFC3276 is a good idea.
Basically there are two main points to learn from RFC3276:
- Add clarification about persistency of the current, interval and total
performance tables (see below)
- Change Total to include daily intervals (This may be an overkill, but
I vote for it anyhow...)
Details:
RFC3276 defines three tables for performance monitoring (definition
copied at end).
- A current status table (persistent)
- A 15 minute interval table (non-persistent) Required to support 1
interval, optional to support one day (96 intervals)
- A daily table (non-persistent). Required to support 1 day. Optional to
support 1 month.
If this is agreed by everyone we can sketch a proposal for required
changes
Feedback?
Ron
hdsl2ShdslEndpointCurrTable OBJECT-TYPE
SYNTAX SEQUENCE OF Hdsl2ShdslEndpointCurrEntry
MAX-ACCESS not-accessible
STATUS current
DESCRIPTION
"This table contains current status and performance information
for segment endpoints in HDSL2/SHDSL Lines. As with other
tables in this MIB indexed by ifIndex, entries in this table
MUST be maintained in a persistent manner."
::= { hdsl2ShdslMibObjects 5 }
hdsl2Shdsl15MinIntervalTable OBJECT-TYPE
SYNTAX SEQUENCE OF Hdsl2Shdsl15MinIntervalEntry
MAX-ACCESS not-accessible
STATUS current
DESCRIPTION
"This table provides one row for each HDSL2/SHDSL endpoint
performance data collection interval. This table contains
live data from equipment. As such, it is NOT persistent."
::= { hdsl2ShdslMibObjects 6 }
hdsl2Shdsl1DayIntervalTable OBJECT-TYPE
SYNTAX SEQUENCE OF Hdsl2Shdsl1DayIntervalEntry
MAX-ACCESS not-accessible
STATUS current
DESCRIPTION
"This table provides one row for each HDSL2/SHDSL endpoint
performance data collection interval. This table contains
live data from equipment. As such, it is NOT persistent."
::= { hdsl2ShdslMibObjects 7 }
-----Original Message-----
From: Orly Nicklass [mailto:orly_n at rad.com]
Sent: Wednesday, October 26, 2005 10:08 AM
To: Ron Cohen; Thomas D. Nadeau
Cc: pwe3 at ietf.org; Doron Tzur; Davidz at corrigent.com
Subject: RE: [PWE3] RE: Alignment of TotalPerf tables in PW MIBs
Pls see in lines
-----Original Message-----
From: pwe3-bounces at ietf.org [mailto:pwe3-bounces at ietf.org] On Behalf Of
Ron Cohen
Sent: Tuesday, October 25, 2005 23:54
To: Orly Nicklass; Thomas D. Nadeau
Cc: pwe3 at ietf.org; Doron Tzur; Davidz at corrigent.com
Subject: [PWE3] RE: Alignment of TotalPerf tables in PW MIBs
Orly,
I agree that the Total table does not add any new information that can
not be compiled from the interval table if 24 hours worth intervals are
kept. However, maintaining 24 hour history is only an option (the
default being 32 intervals: 8 hours).
I do not agree that there is a difference between an T1 interface and a
T1 pw. They should be treated the same way. The T1 can be a physical
interface but can also be a channel carried in a T3 interface.
The current definition of total performance in pw mibs is problematic as
it states that the the counters count from the pw start time and you'll
have to handle the roll-out of each counter. In addition its not clear
what 'management application reset' means.
----[on] counters are valid when they maintain some continuity. If we
relate to values though the life of a connection, discontinuity can
occur at system reset (enforced reboot) or by NMS command. I suppose the
later make no sense to use, so only reboot is applicable. If we worry
about roll-out, we can use larger counter.
Therefore my suggestion is either:
- Remove the Total performance tables from all PW MIBs.
- Align them with RFC3895
----[on]I suppose the alignment you are looking for is a day interval in
case the agent maintain less then 96 intervals of 15min. Otherwise there
is no point for such at all. Saying that maybe we should consider the
actual use. Total of 24 hours shifted every 15min provides duplicate of
information. Maybe we should really use the day interval as define in
3276?
Ron
-----Original Message-----
From: Orly Nicklass [mailto:orly_n at rad.com]
Sent: Tuesday, October 25, 2005 7:12 PM
To: Thomas D. Nadeau; Ron Cohen
Cc: Davidz at corrigent.com; pwe3 at ietf.org; Doron Tzur
Subject: RE: Alignment of TotalPerf tables in PW MIBs
I think we should re-evaluate the semantic, cumulative sum of existing
data provides no added information.
3895 maintains those for backward compatibility. 3592 does not have it,
and 3276 moves towards days interval. This indicates some trend as I
mentioned before.
-----Original Message-----
From: Thomas D. Nadeau [mailto:tnadeau at cisco.com]
Sent: Tuesday, October 25, 2005 16:48
To: Ron Cohen
Cc: Davidz at corrigent.com; Orly Nicklass; pwe3 at ietf.org; Doron Tzur
Subject: Re: Alignment of TotalPerf tables in PW MIBs
This suggestion looks okay to me.
--Tom
> Hi,
>
> I suggest to align the definition of TotalPereformance tables across
> the various PW MIBs as well as align with already defined Total
> performance tables in published RFCs.
>
> In particular, I suggest to align the definition in PW-MIB, TDM-MIB
> and CEP-MIB to that of RFC-3895:
>
> "The (place name here) Total Table contains the cumulative sum
> of the various statistics for the 24 hour period
> preceding the current interval."
>
> And to align the names to always include Total and structured in the
> same way, i.e. change in TDM-MIB the name pwVcTDMPerfTable to
> pwTDMPerfTotalTable.
>
> The different definitions for total tables in the relevant MIBs can be
> found below.
>
> Best
> Ron
>
> PW-MIB:
>
> pwPerfTotalTable OBJECT-TYPE
> SYNTAX SEQUENCE OF PwPerfTotalEntry
> MAX-ACCESS not-accessible
> STATUS current
> DESCRIPTION
> "This table provides per-PW Performance information from
> PW start time or management application reset."
> ::= { pwObjects 5 }
>
> CEP-MIB:
>
> pwCepPerfTotalTable OBJECT-TYPE
> SYNTAX SEQUENCE OF PwCepPerfTotalEntry
> MAX-ACCESS not-accessible
> STATUS current
> DESCRIPTION
> "This table provides per CEP PW performance information
> much like the pwCepPerfIntervalTable above. However,
> these counts represent complete historical information.
>
> TDM-MIB:
>
> pwVcTDMPerfTable OBJECT-TYPE
> SYNTAX SEQUENCE OF PwVcTDMPerfEntry
> MAX-ACCESS not-accessible
> STATUS current
> DESCRIPTION
> "This table provides performance information per TDM PW
> similar to the pwVcTDMPerfIntervalTable above. However,
> these counters represent complete historical information. "
> ::= { pwVcTDMObjects 7 }
>
>
> RFC-3895: Definitions of Managed Objects for the DS1, E1, DS2, and
> E2 Interface Types
>
> dsx1TotalTable OBJECT-TYPE
> SYNTAX SEQUENCE OF Dsx1TotalEntry
> MAX-ACCESS not-accessible
> STATUS current
> DESCRIPTION
> "The DS1 Total Table contains the cumulative sum
> of the various statistics for the 24 hour period
> preceding the current interval."
> ::= { ds1 9 }
>
> RFC-2558: Definition of Managed Objects for the SONET/SDH interface
> type:
>
> No total table, only current and interval tables
>
>
>
> Ron Cohen
> CTO
> Resolute Networks
> Redux and Lycium Unite
>
_______________________________________________
pwe3 mailing list
pwe3 at ietf.org
https://www1.ietf.org/mailman/listinfo/pwe3
_______________________________________________
pwe3 mailing list
pwe3 at ietf.org
https://www1.ietf.org/mailman/listinfo/pwe3
_______________________________________________
pwe3 mailing list
pwe3 at ietf.org
https://www1.ietf.org/mailman/listinfo/pwe3