As a quick follow-up, I'd like to clarify some of my comments below. First, it is my belief from doing extensive research into this problem that it is a Microsoft problem, not a problem with the functionality of the NetApp filer. Also, Network Appliance does provide a solution to this particular problem using their SnapDrive product. We participated in the beta testing of this product and we believe that it does in fact solve the error 59 problem with SQL Server. However, we have chosen not to implement SnapDrive at our site because of the overhead needed to run snapshots (50% of usable disk space) and in order to realize the increased performance of SnapDrive, we would have to do some extensive network re-engineering that at this time is not cost effective for us to do.

Sincerely sorry if I misled anyone regarding our experience.

Lella Bennett Systems Engineer NAREX Inc. 303.526.4000 ext 714 303.526.5130 fax 720.333.8175 cell www.narex.com

-----Original Message----- From: Lella Bennett Sent: Thursday, March 14, 2002 4:56 PM To: Maglinger, Paul; NetApps list server Subject: RE: SQL I/O error

We've seen it quite a bit, actually. In fact, this error is the reason we ended up removing our production databases from netapp filers. We've gone round and round with both NetApp and Microsoft on this issue and the bottom line is this: this error is due to the fact that the SQL traffic is going through the TCP/IP stack. Evidently, there is some latency that SQL server experiences while trying to communicate with the data file and the SQL server thinks that it's lost the connection and throws this error, marks the database suspect, etc... The other reason why this error may occur is that you've reached the maximum number of threads available. If this is the case, you will see an error in the NT event log referring to this. You can increase the number of available threads using a reg hack. If it's not caused by the threads issue, then I can't help ya. Like I said we experienced A LOT of pain due to this error and never could find a solution to it.

Lella Bennett Systems Engineer NAREX Inc. 303.526.4000 ext 714 303.526.5130 fax 720.333.8175 cell www.narex.com

-----Original Message----- From: Maglinger, Paul [mailto:PMAGLINGER@scvl.com] Sent: Thursday, March 14, 2002 3:55 PM To: 'NetApps list server' Subject: SQL I/O error

Anyone seen an SQL Server Error Log message saying :

I/O error 59(An unexpected network error occurred.) detected during read of BUF pointer = 0x14685900, page ptr = 0x64c78000, pageid = (0x1:0x378), dbid = 7, status = 0x801, file = e:\sqldevices\dynamics\dyn\gpsdyn1dat.mdf..

where the e drive is mapped to our 740 filer running 6.1.1R2P1. We were looking at sysstat and CPU was at 95-100%, cache age fell from 35 to 0.

Paul Maglinger, A+, CA, CET, MCSE Systems Administrator Shoe Carnival Inc. (812)867-4674 pmaglinger@scvl.com