![]() | |
![]() |
| | Thread Tools | Display Modes |
#1
| |||
| |||
|
#2
| |||
| |||
|
#3
| |||
| |||
|
|
what kindof storage -- Molap Realtime, Molap Scheduled; etc? |
#4
| |||
| |||
|
|
Molap. After pulling new data into the base and processing all new add, changes, and deletes, we do a full process on the cube. Running from BIDS (interactive) or our 2 step batch file - base warehouse in one, AS in the second, during the day results in the 2 hour or less time. The batch processes are kicked off through a remote command from an AS400. Batch process number 1 (all SQL activities) is consistant night an day. Number 2 is dimension and cube builds. We have run several tests - from BIDS, manually launching the batch files, as well as launching through remote command. Same 2 hour or less results during the day. We have the cube data on local disc, and go to the SAN for the base warehouse data. If we compare overall SAN activity during the day vs. overnight, it is higher during the day than at any point during the night. So we don't see this as a constraint. Looking at the timestamps on all the files used to build the cube, we have 1 dimension that we see a time gap of several hours or more on. We are going to take this dimension out tonite and see if it finishes quicker (or rather finishes like a run during the middle of the day). One of our LAN admins has been looking at all kinds of perfmon data, but does not see anything unusual. Plenty of disc (local and SAN), memory seems fine, etc. All we see is activity flat lining for a period of time when the cube is almost complete. This flat line equates to the extended run times we are experiencing. -- Greg E "aaron.kempf (AT) gmail (DOT) com" wrote: what kindof storage -- Molap Realtime, Molap Scheduled; etc? |
#5
| |||
| |||
|
#6
| |||
| |||
|
|
Molap. After pulling new data into the base and processing all new add, changes, and deletes, we do a full process on the cube. Running from BIDS (interactive) or our 2 step batch file - base warehouse in one, AS in the second, during the day results in the 2 hour or less time. The batch processes are kicked off through a remote command from an AS400. Batch process number 1 (all SQL activities) is consistant night an day. Number 2 is dimension and cube builds. We have run several tests - from BIDS, manually launching the batch files, as well as launching through remote command. Same 2 hour or less results during the day. We have the cube data on local disc, and go to the SAN for the base warehouse data. If we compare overall SAN activity during the day vs. overnight, it is higher during the day than at any point during the night. So we don't see this as a constraint. Looking at the timestamps on all the files used to build the cube, we have 1 dimension that we see a time gap of several hours or more on. We are going to take this dimension out tonite and see if it finishes quicker (or rather finishes like a run during the middle of the day). One of our LAN admins has been looking at all kinds of perfmon data, but does not see anything unusual. Plenty of disc (local and SAN), memory seems fine, etc. All we see is activity flat lining for a period of time when the cube is almost complete. This flat line equates to the extended run times we are experiencing. -- Greg E "aaron.kempf (AT) gmail (DOT) com" wrote: what kindof storage -- Molap Realtime, Molap Scheduled; etc? |
#7
| |||
| |||
|
|
Has that dimension experienced many changes lately or an increase in size? "Greg E" <gte (AT) newsgroups (DOT) nospam> wrote in message news:7B9AF12D-06F0-44DC-9E6F-2125E7FE33C9 (AT) microsoft (DOT) com... Molap. After pulling new data into the base and processing all new add, changes, and deletes, we do a full process on the cube. Running from BIDS (interactive) or our 2 step batch file - base warehouse in one, AS in the second, during the day results in the 2 hour or less time. The batch processes are kicked off through a remote command from an AS400. Batch process number 1 (all SQL activities) is consistant night an day. Number 2 is dimension and cube builds. We have run several tests - from BIDS, manually launching the batch files, as well as launching through remote command. Same 2 hour or less results during the day. We have the cube data on local disc, and go to the SAN for the base warehouse data. If we compare overall SAN activity during the day vs. overnight, it is higher during the day than at any point during the night. So we don't see this as a constraint. Looking at the timestamps on all the files used to build the cube, we have 1 dimension that we see a time gap of several hours or more on. We are going to take this dimension out tonite and see if it finishes quicker (or rather finishes like a run during the middle of the day). One of our LAN admins has been looking at all kinds of perfmon data, but does not see anything unusual. Plenty of disc (local and SAN), memory seems fine, etc. All we see is activity flat lining for a period of time when the cube is almost complete. This flat line equates to the extended run times we are experiencing. -- Greg E "aaron.kempf (AT) gmail (DOT) com" wrote: what kindof storage -- Molap Realtime, Molap Scheduled; etc? |
#8
| |||
| |||
|
|
I presume you schedule this process. have you try to change the schedule time? like a time in the day, when you can monitor more closely what's appends (without doing an interactive process) maybe during the night you have a lock on some tables due to other schedules on the server or on the SAN. if at the same time you do the process the SAN replicate or backup some data (or do any IO intensive task) this could cause issues. "Greg E" <gte (AT) newsgroups (DOT) nospam> wrote in message news:7B9AF12D-06F0-44DC-9E6F-2125E7FE33C9 (AT) microsoft (DOT) com... Molap. After pulling new data into the base and processing all new add, changes, and deletes, we do a full process on the cube. Running from BIDS (interactive) or our 2 step batch file - base warehouse in one, AS in the second, during the day results in the 2 hour or less time. The batch processes are kicked off through a remote command from an AS400. Batch process number 1 (all SQL activities) is consistant night an day. Number 2 is dimension and cube builds. We have run several tests - from BIDS, manually launching the batch files, as well as launching through remote command. Same 2 hour or less results during the day. We have the cube data on local disc, and go to the SAN for the base warehouse data. If we compare overall SAN activity during the day vs. overnight, it is higher during the day than at any point during the night. So we don't see this as a constraint. Looking at the timestamps on all the files used to build the cube, we have 1 dimension that we see a time gap of several hours or more on. We are going to take this dimension out tonite and see if it finishes quicker (or rather finishes like a run during the middle of the day). One of our LAN admins has been looking at all kinds of perfmon data, but does not see anything unusual. Plenty of disc (local and SAN), memory seems fine, etc. All we see is activity flat lining for a period of time when the cube is almost complete. This flat line equates to the extended run times we are experiencing. -- Greg E "aaron.kempf (AT) gmail (DOT) com" wrote: what kindof storage -- Molap Realtime, Molap Scheduled; etc? |
#9
| |||
| |||
|
![]() |
| Thread Tools | |
| Display Modes | |
| |