My API returns JSON content with some errors, such as 500 (Bad Request). Under IIS 6, this worked fine. In IIS 7, once I turn off detailed error messages it simply returns the text "Bad Request". How can I keep the contents of my response intact?
On my old server, I separated them on to different RAID arrays to improve performance. Using a SAN is new to me, so I'm not sure of the optimal configuration.
SAN: 48 drives. 10K 600GB SAS. RAID 50. 4 Gbps connection.
Updated info:
All 48 drives are part of the same RAID array. I would like to know if there is any benefit to having data and logs in different volumes. I know that if I split my SAN up there can be some benefit, depending on the circumstances. The Dell representative that sold us the SAN said the best performance can be obtained by keeping all 48 disks in one array, and use seperate volumes ("LUNS"?). Sorry if my question isn't clear.
My new database server (Windows Server 2008 R2) has a 4 Gbps connection to my Dell EqualLogic iSCSI SAN. I've been testing it by backing up multiple databases at the same time. There appears to be a bottleneck with the network adapter. It shows up as 4 1 Gbps connections, and the network utilization never exceeds 2 Gbps. Any recommendations where to look next?
SAN: 48 drives. 10K 600GB SAS. Raid 50.
Update:
Flow control is set to auto, but I think it is being used because MPIO is installed. There is a switch and I'm working on jumbo frames, but while researching MPIO I noticed something that might be affecting performance.
In the database backup test I was performing, 4 volumes were involved: one for data, one for the log, and 2 for the backup files (split by RedGate SQL Backup to improve performance). I was trying to determine if there was any performance benefit to separate volumes on a SAN. On my old server, I would use 4 different RAID arrays in this situation.
Each of my 1 Gbps network adapters has an IP assigned in the range 10.0.1.251-10.0.1.254. Each volume has 2 of these IP addresses assigned to it in the "Dell EqualLogic MPIO" tab in iSCSI Initiator Properties. 251 & 252 are used for the log & both backup volumes. 252 & 253 are used for the data volume. It seems like this could be a limiting factor.
I'm going to see if I can assign them different IP addresses.
Update 2:
Using the MPIO Remote Setup Wizard, I changed the "Max sessions per volume slice" from the default 2 to 4 and rebooted. Each volume now has 4 paths, one for each NIC. The utilization is now fairly evenly distributed between the 4 NICs, but the total is still less than 2 Gbps. Each one maxes out at about 45%. Once I get access to my switch I will try to get jumbo frames working and see if that helps.
Solution:
Jumbo Frames. They had to be set on the database server NICs and on the switch. The SAN handled it automatically. This required a reboot on the DB server.