Community
- Command Central
- MVP Program
- Monthly Mission
- Blogs
- Groups
- Events
- Media Vault
Products
- Observability
- Network Management
- Application Management
- IT Security
- IT Service Management
- System Management
- Database Management
Content Exchange
- SolarWinds Platform
- Server & Application Monitor
- Database Performance Analyzer
- Server Configuration Monitor
- Network Performance Monitor
- Network Configuration Manager
- SQL Sentry
- Web Help Desk
Free Tools & Trials
Store

Avg Disk Queue length

I have been noticing gaps again in our graphs. We have been fairly active lately adding and removing devices, not a huge amount, about 6 a day are being removed and 6 new devices being added. So we haven't been increasing the amount of elements in the database.

The total number of elements now is 8369 on our single polling engine

I have checked the database and looked at the polling engine which has a completion percentage of 99.07%

Our collection statistics are as follows:

Polling interval for NODES, INTERFACES & VOLUMES are all at 180 sec

Statistic collection for

nodes = 15min

interfaces = 10 min

vol = 15 min

I had checked the database entries for some of the interfaces who's graphs had gaps at a 15 min interval in the web GUI. The interface traffic detail confirms the polls were missed. Also it seems that it keeps happening mostly to a same interfaces. I thought the polling cycle was random so to prevent this?

Also on the SQL database server I have noticed that the AVG Disk Queue is almost constantly at 100. With the highest number of I/O reads & writes coming from the SQL Process. (not surprising really) I'm fairly certain that a major contributing factor to this is Syslog messagees.

NOTE:

I increased the statistics collection interval on the interfaces to 15min for a about 4 hours yeaterday to see if there would be any improvement. The polling completion percent actually dropped to 98.746% so I have returned it back to the 10 min mark but the % has stayed in the 98% range.

Looking at the interface traffic detail a few days previous (when the Polling completeion % was at 99.07) and seeing a number interfaces that were barely polling once an HOUR!! leads me to doubt the acurracy of the polling completion % calculation or it's source.

I know I have viewed the missed polls in the database and I know that the avg dsk queue length will not affect the polling engine but is it possible that the avg disk queue length is creating a bottle neck that isn't allowing polling stats to be added to the database causing both

gaps in graphs

the gaps in the interface traffic detail query

Any suggestions

Find more posts tagged with

Accepted answers

All comments

Yann

...but is it possible that the avg disk queue length is creating a bottle neck that isn't allowing polling stats to be added to the database...

I think you found yourself the issue. We recommend RAID1+0 disk arrays for its performance in writing/reading data.

Which type of Disk Arrays do you have on the SQL Server?

Is the database server dedicated to Orion and its module?

Which modules are you running at the moment? APM, NetFlow, Wireless, VoIP?

Geek Tips to improve Database Performance

I Attached a SQL Performance Troubleshooting guide (cannot remember where I found it) if your DBA is in holidays.

Thanks,

Yann

TroubleshootingSQL.zip

Ciag

Hi Yann

I'm am looking into what arrays are set up on the database server. I'm not sure what array is being used as it was built before the application was handed to me. I do know that the server is not a virtual server and is dedicated to the Orion NPM database. Currently we do not have any bolt-on modules running on top of Orion NPM.

We have 1 server for the Orion suite of application and another server for the database, both are dedicated servers.

We do however have a considerable amount of syslog going to it (over 2000 devices unfiltered syslog) which I'm sure is a major contributing factor in the bottle-neck. We are trying to implement measures to reduce this, however it will be quite a while before we are able to do this.

Do you know a wuick way to check which array are running?

Cheers

Ciaran

Yann

Do you know a wuick way to check which array are running?

What is the model/brand of your database server?

Ciag

Hi Yann,

Can I send you a screenshot to your email account of what I'm seeing when I look at the disks. It looks like there is no RAID set up but I'm not sure as I don't deal with that stuff.

Cheers