Monitoring on a Budget - Part 2

Question

Post reply

Monitoring on a Budget - Part 2

Bert Carles

SSCrazy

Points: 2634
More actions
January 26, 2008 at 8:51 pm

#181260

Comments posted to this topic are about the item Monitoring on a Budget - Part 2
Take care,
Bert
"Speculations? I know nothing about speculations. I'm resting on certainties. I know that my Redeemer lives, and because He lives, I shall live also." - Michael Faraday

Viewing 8 posts - 1 through 7 (of 7 total)

You must be logged in to reply to this topic. Login to reply

giorgio-459258 SSC Enthusiast Points: 155 More actions · Answer 1

Hi,

I had a look at your article and it looks good so well done.

I have only a few comments and questions also 🙂

You don't need to go to point 2 again and do all again. You can go to "Data" then "PivotTable and PivotChart report" and then select "Another PivotTable report or PivotChart report" Instead of "External Source" and then select "PivotChart report (with PivotTable report)". Then select the previous PivotTable then next...

My questions now 🙂

This article is quite nice BUT what does the numbers mean??! To me and to most people they don't make much sense! How do I know if my SQL Servers are performing well or not? What are the threshholds? Benchmarks? etc... It is very nice to see all the graphs but what do they actually mean?

Can you please tell us what the various columns and numbers mean?

Thanks

Bert Carles SSCrazy Points: 2634 More actions · Answer 2

I don't think you'd want to use this facility for point in time monitoring. However, you could use the data to establish benchmarks. And, as we do, use it to track increased/decreased system utilization over time. Armed with this data, you can reasonbly estimate when systems will become at/or near capacity and when you'll need to buy new hardware.

Take care,

Bert

"Speculations? I know nothing about speculations. I'm resting on certainties. I know that my Redeemer lives, and because He lives, I shall live also." - Michael Faraday

giorgio-459258 SSC Enthusiast Points: 155 More actions · Answer 3

How can you know that the systems are at maximum capacity? These number don't mean anything if you don't have something to compare with! How do you know CPU is at maximum capacity? What are the values for reference? What do the numbers exactly mean?

As far as I can see I don't know if my CPU is performing well or not because there't nothing that tells me that the CPU should not be having more than 13040200020 as total for 1 day for example!

Bert Carles SSCrazy Points: 2634 More actions · Answer 4

That's what establishing baselines is all about. This tool can help you do that, but no tool does it for you.

Take care,

Bert

"Speculations? I know nothing about speculations. I'm resting on certainties. I know that my Redeemer lives, and because He lives, I shall live also." - Michael Faraday

giorgio-459258 SSC Enthusiast Points: 155 More actions · Answer 5

Baselines? What are you talking about?

Sorry but I am confused with your answers!

Bert Carles SSCrazy Points: 2634 More actions · Answer 6

Baseline is the same as a benchmark.

Take care,

Bert

"Speculations? I know nothing about speculations. I'm resting on certainties. I know that my Redeemer lives, and because He lives, I shall live also." - Michael Faraday

humbleDBA Hall of Fame Points: 3540 More actions · Answer 7

Hi,

Came across this discussion from a link to a later article...so sorry for the latness of reply, but felt I had to correct the misconception that a Baseline and Benchmark are one an the same, because they're not.

http://en.wikipedia.org/wiki/Benchmark_(computing)

http://en.wikipedia.org/wiki/Baseline_(configuration_management)

A benchmark is used to assess the performance of a system using a set of tests that are standardised, so as to give comparable results. Often different system architectures are tested, such as Oracle, SQL Server, DB2, etc., in the Database TPC tests (www.tpc.org). Often here, you'll find that those who participate are looking to achieve the highest performance and set the benchmark. When I benchmark a set of systems presented by different vendors as the best system/setup I often use a tool something like SQLIO, which I will run in a pre-defined and standard way. I then compare the results to see which gave the best performance characteristics for the type of workload I'm looking for the system to do.

With baselines, you are looking to set a starting point from which to compare future performance of a particular system, for example. This can be seen in the DMVStats tool for performance monitoring of SQL Server 2005. In this tool you can create baselines for different time intervals on the same system and then compare the results. Often in baselining you create a 'before change' baseline, carry out some change to the system, and then create an 'after change' baseline. From these you can compare and see what effect, if any, the change you made had.

So, IMHO, baselines and benchmarks are not the same thing.

HTH