Gregory Primosch
- Name:
- Gregory Primosch
- Status Level:
-
New Enzee (52 points)
- Member Since:
- Jan 13, 2011
- Groups:
- Enzee Universe
Recent Activity
CPU/IO usage in Twinfin 12
"I can't tell if HOST_FABRIC maxes out at 1.0 since the largest value is .25 in our database. I think our system is a P36 (276 nodes). I can" in Users • 7 replies |
5 days ago |
|
CPU/IO usage in Twinfin 12
"Shawn, all of the measures in _v_system_util have a max of 1.0 (100%) except for the FABRIC measures, which have a max of 2.0. How should we" in Users • 7 replies |
5 days ago |
|
Looping in SQL
"Shawn's approach is sound for a reasonable number of levels. However, if your depth gets much higher than 20, it would start to break down." in TwinFin Talk • 10 replies |
1 week ago |
|
Processing large set of records
"Perhaps my comments from the other posting may be helpful to you: http://www.enzeecommunity.com/message/9563#9563 The chunking approac" in Enzee Universe • 15 replies |
1 week ago |
|
Strange trend in performance
"Itachi wrote: Also if the rows are in several billions would it make sense to call sql over only million slices by and repeating it i" in Enzee Universe • 16 replies |
1 week ago |
|
Looping in SQL
"One of the problems with these kinds of queries is that it is difficult to partition your dataset so it runs distributed on all nodes of the" in TwinFin Talk • 10 replies |
1 week ago |
|
Looping in SQL
"Looks like you are trying to do tree traversal. This is often done in SQL using WITH clause recursive queries (in Oracle it is CONNECT BY)." in TwinFin Talk • 10 replies |
1 week ago |
|
Netezza best practices
"I am pretty sure an MV index helps most for single record lookups. If you need to scan large numbers of records, then performance is going t" in Users • 6 replies |
1 week ago |
|
Open file inside UDF
"Nick's point about forcing the UDF to run on the host was regarding the difference between passing literals and passing row values. If the U" in Users • 13 replies |
1 week ago |
|
Netezza best practices
"Nick is right that you should be naturally organized around date, and so with zone maps, there would be no advantage to further partitiong b" in Users • 6 replies |
1 week ago |
|
Netezza best practices
"Unioning tables will always be slower to query than a single well organized table. Collegues of mine have tested this approach and it has al" in Users • 6 replies |
1 week ago |
|
Open file inside UDF
"Thanks for the clarification Shawn. Is it possible to use the SPUpad with C++ or only with nzLua? UDX has no ability to run SQL agains" in Users • 13 replies |
1 week ago |
|
Atomic operation
"Another option to table swapping is database swapping. You have two databases with identical schemas. You truncate and load database A, then" in TwinFin Talk • 7 replies |
1 week ago |
|
Open file inside UDF
"It is possible to read a file from a process running on the SPUs. Put in the following directory on the host: /nz/export/ "The /n" in Users • 13 replies |
1 week ago |