In this webinar we had MVP and Director of Education at Red Gate Software, Brad McGehee talk about performance monitoring and explain what items the proactive DBA should have on their monitoring checklist. We also had Ben Rees demonstrate Red Gate’s new SQL Monitor 2.0. We saw SQL Monitor’s intuitive overviews, mobile monitoring, alerting, reporting and how to go back in time to identify the source of a problem.
You can see this product in action at http://www.thefutureofmonitoring.com/ where it shows the live data from the SQLServerCentral database servers.
The slide decks for this webinar are available for download at the bottom of this page. We also have included the question and answer chat from the webinar.
Question and Answers
Q: Will the slide deck be available for download?
A: Yes, we will make the recording and slide deck available later this week.
Q: how much historical info is kept for comparison purposes?
A: This is user-configurable via Purge settings – the user can choose how far back to keep historical data.
Q: Do you recommend using WITH( NOLOCK ) to avoid frequent deadlocks?
A: No, this tends to be a programmatic issue, accessing resources in the wrong order.
Q: There are many counters to monitor from SQL profiler to DMV's and perfmon. The problems I find is knowing when a counter is normal or abnormal. Is there any papers on the web or links you could provide that gives help on how to identify these abnormal values?
A: This is a hard one. There are lots of articles, and Brad covers some good items in his books. You can get those as ebooks for free from SQLServerCentral (http://www.sqlservercentral.com/Books/) or buy printed copies from Amazon. The main thing is that you need to monitor as a baseline so that you know what applies to your particular instance and then notice over time what deviations are. As far as what's "normal", you often need to start working through specific queries and examining query plans. Grant Fritchey's book on performance tuning for queries is a good reference. http://www.amazon.com/Server-Performance-Tuning-Distilled-Experts/dp/1430219025/ref=sr_1_2?ie=UTF8&s=books&qid=1289932283&sr=8-2
Q: I have installed and set up SQL Monitor for several of our servers. There are a number of features that don't seem to work correctly and I am not totally sold on the interface when compared to SQL response. Is there a site dedicated to logging these sorts of issues, other than the SQL Monitor forum? Thanks.
A: I think the best thing would be to contact support (firstname.lastname@example.org) with specific questions. I would be interested in hearing as well, and perhaps putting some documents/case studies together.
Q: where does most of the processing take place? the web server, client machine?
A: Nothing is installed on the machines that you are monitoring. A Base Monitor service is installed on a machine of your choosing – this carries out all of the polling of the monitored machines. The web service then access the Base Monitor to serve up the web pages for the application.
Q: Can this product be used on all SQL versions, 2000, 2005, 2008 etc?
A: The product can monitor SQL 2000, 2005, 2008 and 2008R2. The backend database used for storing data must be either SQL 2005, 2008 or 2008R2. Requirements page here shows the list: http://www.red-gate.com/products/SQL_Monitor/index.htm
Q: i can recommend the new DMV book from red-gate: http://www.red-gate.com/about/book_store/dynamic_management_views.htm
Q: Can it report on mountpoints as well as physical drives?
A: Not right now. Only physical drives are being monitored. It is on the enhancement list.
Q: can you tell us about Analysis and what it dies?
A: It allows you to view various counters and their values over time, or even compare them with other ranges, for example – “What was the value of counter X today vs. yesterday?”
Q: What version of OS and SQL is it compatible with
A: You can get the list here: http://www.red-gate.com/products/SQL_Monitor/requirements.htm See previous answer regarding SQL Server versions supported.
Q: Will the alerts email the Deadlock details... the victim and other query?
A: Yes, a link comes out in the alert with details showing this information.
Q: Does this monitor SSAS
A: only basic up/down.
Q: Do you show somthing similar to a deadlock graph?
A: We show a lot of detail for deadlocks on the deadlock Alert Details page.
Q: Does that list include all the permon SQL counters?
A: No, limited list – limited to the most useful counters – for now. That may change, based on feedback.
Q: can i monitor servers from different countries
A: Yes, as long as there is connectivity
Q: do we need to purchase a special plug-in for ERP packages (e.g. SAP) to identify login user-ids?
A: We would just report the SPID.
Q: can any of the administration function done within SQLServer administrator be done through SQLmonitor (such as user creation) or is this strictly monitoring software?
A: Strictly monitoring/analysis.
Q: when you create filters on alerts to show a subset of servers and a small set of metrics ... can preferences be saved on those filters
A: yes, you can create custom Alert Filters and save these for later use.
Q: can u perform intelligent analysis and automatically review the data on I/O, CPU Utilization, etc and give advise on what should be done to fix the problem?
A: limited amount of help in this area in v2.0, though we do provide documentation in the product explaining counters and metrics along with guideline values that you should be looking for.
Q: Do you have subscriptions for reports. i.e. ability to email out an email each week on the cpu for the last week etc.
A: You can write your own reports in SSRS. We don't do that automatically.
Q: can i send the analysis report to management via email?
A: we don't do reporting. You can send URLs of the relevant Analysis page, or screenshots.
Q: On the analysis page, can you display multiple metrics at the same time, ie, disk read and disk write?
A: enhancement that is logged. One metric at a time right now
Q: can you modify the threshold on the alerts? for ex: do not alert me if CPU is high unless the condition has been on for at least 2 minutes...
A: Yes, there is some setting for 3 levels – High, Medium and Low, across a configurable time period (i.e. only alert if CPU is above a given threshold for a certain time period).
Q: Can monitor show blocked and blocking processes with details like host name, text command, user?
A: There is an alert for blocks, and you can get details on the user, SPID, time, SQL.
Q: any t-sql tuning capability?
A: Not right now