HI, a quick question, I hope someone can help me.
We have a Navision 3.7 on native DB, DB size 75 Gb, located on Raid1 (mirror), on 2x 125G SAS 15kRPM HP HDD-s.
These are FAST. Copying to/from the RAID array runs at about 25-30 Mbytes/sec. Not bad. Of course, sequential reading is fast.
However, Navision DB reading (via the service) is NEVER more than 1-1.2 Mbytes/sec! I also tried to create "test" reports that try to collect data from parts of the DB which I think is quite "sequentially readable" (i mean it's not fragmented) but no, the read rate is NEVER, ever more than 1.2 Mbytes/sec, which is just odd.....
Is there any reason to this? Did I miss something in the config? (DBMS cache is 1G, commit cache is on, device write cache is off (Win2k3 does not allow me to change this).....
Thanks in advance....
0
Comments
2) defragment disk
Reading 1.2 Mbps could be network botlleneck (100MBps).
What are your readings if you read database in single-user mode (directly opened, so you have to make a hotcopy of database)?
2. how? :shock: Disk has 1 file on it (database.fdb)
network: no. all tests were done locally.
Single-user mode opening from client: 1.2 Mbyte/s also....
Sadly, the server can only handle 6 HDDs (we curently haver 2 for SYS, 2 for DB)....
We might need a SAN.....
Yes, that is right. But also remember that is a minimum suggestion. If you need to add disk to suppor that then also consider expansion.
Or maybe it's time for a new server. I would go for DAS over SAN with the native database. There are a number of servers out there that hold a fair number of drives. Another option is to add external DAS to the existing server.
I'm not familiar with HP making a 125 GB SAS drive (is that the formated capacity?). Of course they may make some products for other countries.
Don't use such large drives. Go with more, smaller drives.
OKay, my mistake, 150G
OK, I'll consider changing to (approx) 8-10 smaller drives.
Thanks for your help!
The Navision DB is NOT like a SQL DB.
SQL puts all the data together based on the clustered index. And if you don't have a clustered index, SQL puts the data where it happens to be fastest to write it.
The Navision works like that. The Navision-DB is a large heap. The records of 1 table are not necessarily together on the disk.
So even if you read a table, the disk is not doing a sequential read, but is doing random reads on the disk.
No PM,please use the forum. || May the <SOLVED>-attribute be in your title!