Jump to content

Welcome to Geeks to Go - Register now for FREE

Geeks To Go is a helpful hub, where thousands of volunteer geeks quickly serve friendly answers and support. Check out the forums and get free advice from the experts. Register now to gain access to all of our features, it's FREE and only takes one minute. Once registered and logged in, you will be able to create topics, post replies to existing threads, give reputation to your fellow members, get your own private messenger, post status updates, manage your profile and so much more.

Create Account How it Works
Photo

Program Benchmarking


  • Please log in to reply

#1
xained

xained

    New Member

  • Member
  • Pip
  • 2 posts
hi!

I've been assigned the task of putting numbers to pretty much everything that happens at our sites. We need this to present to management so that we have quantifiable numbers for them to crunch.

OK thats the intro. Here's the meat of it. I need to be able to time things, how long the pc takes to boot, how long programs take to open, how long the pc takes to shut down. Then there's the bandwidth and latency measuring for the internet, local network usage, server benchmarking and usage analysis. You get the idea I hope.

Now I've managed to find 2 programs that measure system startup, one is by Microsoft themselves called bootvis (very cool), the other program is by planetsoft (boottimer) and gives you a reboot cycle time. What I need is something that will log EVERY boot, from as close to the bios until the last starup/msconfig program has run. That way we can analyze when and if users machines are slowing down, when this happens etc. Otherwise it becomes a subjective experience and pcs are NEVER fast enough for the user! So if we've got numbers then we have an objective comparision.

The next part is a program that does the same but for programs, it measures the and logs the length of time it takes to finish loading programs. Once again users complain that programs take forever to load, and we need to know how long forever is and is it different from yesterday.

Network can wait for a bit, but if you'ver got any suggestions I won't say no.

Finally server benchmarking. Sigh. where to start, something to measure load balancing between the various servers, but also something that can take perfmon (we're using 2003) and actually make it useful. Its all very well having all the numbers perfmon can throw at you, but if you've got to then process 6months of numbers to show to management and make it meaningful, I don't want the job! I've found a linux program called cactus that measures memory usage, it seems like an awesome product if the only thing that I wanted to measure was memory usage, is there something thats more feature full and measures other functions, is there something thats windows based?

OK I hope you guys can help me with this.
Thanks!
X
  • 0

Advertisements


#2
Neil Jones

Neil Jones

    Member 5k

  • Member
  • PipPipPipPipPipPipPipPip
  • 8,476 posts
Numbers in this context really don't mean anything. There are many many factors that could affect the system's boot-up time, even more so if a service pack was applied the previous night or a big wave of updates were applied on shut-down. These may be only one-offs but they add to the boot-up time.

A lot depends on the specifications of the PCs in question. If they don't have enough memory in them to start with they'll never boot up at a decent speed. A lot also depends on the state of the Windows and the rest of the hardware - a PC with a dying drive may start up properly and in good time but it doesn't mean its working properly or even usable. The numbers would only apply at that time you used it, therefore it's entirely possible an update to something throws your figures out. Your best bet would probably be to assume that as long as it comes on in a reasonable time (PC spec pending, 90 seconds is about average) and it does stuff, then worry about it when it doesn't boot anymore rather than throwing resources at what I consider, with all due respect, a pointless exercise.

The same principle applies to the programs - different overheads on different PCs with different hardware specifications will show you different load-times and unless you're prepared to match all the specs, programs down to every detail, again your figures will be meaningless because they're only at a point in time and updates, etc can throw it all out of sync.

If truth be told, management would be far better doing other things than trying to micromanage the office computers. Perhaps the fault(s) with the computers lie with those who use them? There are people in companies all over the world who spend all day in front of a computer and still don't have a clue how to work them.
  • 0

#3
xained

xained

    New Member

  • Topic Starter
  • Member
  • Pip
  • 2 posts
Hi Neil,

Thanks for the reply. Unfortunately this isn't an "if I feel like doing it" scenario, now on my own I wouldn't worry too much about the whole thing, if it boots in under 3min then that’s fine, I am not however the end user.

One advantage of having log on time and program load times is that the anomalies become useful. If you can see that for the past month the boot time has been 60sec ave. then it jumps to 90sec and stays there then you know you've got something to look at, otherwise the user complains that the pc is starting up slower and you go "tough". That doesn't solve the problem or make the user happy and seen as I'm not inclined to have my life made miserable by end-users complaining the whole time I'd be inclined to solve the problem.

Your argument of machine spec affecting the boot time is relevant, but only if we were looking at averages of the whole site or if we were running spot tests and then comparing them to other machines, if we run this continuously we establish a baseline for each machine and then can monitor system degradation.

The program launch time is another story entirely and is need to either disprove or support the users statements that “its taking a long time to load”. What is “a long time”, did it take longer than it did yesterday, last week, last month?

All of which brings me back to my original question:
Do you, or anyone reading this for that matter have any suggestions?

Thanks
X
  • 0






Similar Topics

0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users

As Featured On:

Microsoft Yahoo BBC MSN PC Magazine Washington Post HP