| Title: | DECmcc user notes file. Does not replace IPMT. |
| Notice: | Use IPMT for problems. Newsletter location in note 6187 |
| Moderator: | TAEC::BEROUD |
| Created: | Mon Aug 21 1989 |
| Last Modified: | Wed Jun 04 1997 |
| Last Successful Update: | Fri Jun 06 1997 |
| Number of topics: | 6497 |
| Total number of notes: | 27359 |
In MCC V1.3 the ILV encoding of FullEntities has changed.
I guess there are good reasons for this . Could you explain it ?
The only problem is that the new
encoding is about 3 times bigger that the previous one, which can lead to
ennoying side effects in existing MCC applications :
- much more memory needed (impacts on performance ?)
- all default ILV buffer sizes have to be changed
- ....
=======================================================================
Herafter is the demonstration : on FCL PM (with log), the command
"show domain domaina member domainb all attributes" is performed
1. on MCC V1.2
2. on MCC T130
1- MCC V1.2 . dump of outp returned :
[ 1 ] (
[ 1 ] (
[ 19 ] (
[ 1 ] 0a
[ 2 ] 00
[ 3 ] 00
)
[ 20 ] (
[ 1 ] 37 -- 7
[ 3 ] 00
[ 4 ] (
[ 0 ] (
[ 1 ] 01
[ 2 ] 08
[ 3 ] 01
[ 4 ] 05
[ 5 ] 00 00 00 00 00 00 00 00 00 00
00 00 00 00 0b 00 01 07 64 6f 6d 61 69 6e 62 00 00
)
)
)
)
)
2.(MCC T130) dump of outp returned
[ 1 ] (
[ 1 ] (
[ 19 ] (
[ 1 ] 0a
[ 2 ] 00
[ 3 ] 00
)
[ 20 ] (
[ 1 ] 37 -- 7
[ 3 ] 00
[ 4 ] (
[ 65535 ] 08 00 00 00 01 00 01 00 34 00 0e
02 30 00 00 00 1b 00 00 01
01 00 00 00 05 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
00 00 0b 00 01 07 64 6f 6d 61 69 6e 62 00 00 00 00 00 00 00
00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00
)
)
)
| T.R | Title | User | Personal Name | Date | Lines |
|---|---|---|---|---|---|
| 4235.1 | For performance reasons | TOOK::GUERTIN | MCC Managing everything for everyone everywhere | Fri Dec 11 1992 08:37 | 10 |
Performance analysis has shown that when many notifications are being
processed, the ILV encoding and decoding of AESs (FullEntities) are a
bottleneck. The ILV routines broke appart the AESs one level at a time
and encoded each level as a set of individual data elements. We
changed the code to do a single "marshall" of the AES and the
bottleneck was removed. I do not know why so much extraneous
information (looks like null padding) is encoded. I can check on that.
The bottom line is that performance should improve.
-Matt.
| |||||