Can someone please help in following queries :-


Q1: How to calculate the maximum IOPS supported by a given disk , what utility i can use & how ?

Q2: how to calculate that what type of disk to use in amazon cloud if user mentioned its IOPS requirement.eg: if a user mentions its application is expecting to have 5000 read operations per second so what type of disk i choose for him & how ?

I don't think so, the short answer is that you want a better formulated pile of questions, which means studying the area.

The following things can affect speed: Motherboard & chipset; bus type & speed; disk type, cache, loading, filesystem, mechanical/ssd, interface type & interface load; kernel; Size of operation - writing 1 file of 10 megs or 1000 files of 10k, etc.

1. ...and the VERY FIRST HIT in Google for "how to calculate the max iops of a disk" is: http://www.ryanfrantz.com/posts/calculating-disk-iops/
2. As said by business_kid, there are lots of variables.

You've been here four years now, and should be able to do basic research of your own....as stated in the "Question Guidelines" link. Further, it seems like you're asking US on how to do your job for YOUR CLIENT...who is paying YOU. If you're not able to do the job (or even show effort in researching things), then asking us to do your work for you, FOR FREE, is fairly rude.

Too bad TB0ne chose to be snarky instead of answering the question. The link posted is dead and this forum is now a top search result, but with no answer.

Too bad you chose to ignore the answer the OP was given, because the search-term given to them STILL provide the answer; one of over 177,000, and this isn't the top search result for what the OP was given. And if you performed that search, and read what business_kid also posted, you'd see that there IS no set answer, since it all depends on too many variables to write out here.

Why did you bother to spend your time registering here just to post this on a closed, two-year-old thread?? Instead of being snarky, you should have looked at the OP's posting history, to see why they got the answer they did. Not to mention the fact they were HIRED by someone to answer that question, and the extent of the work they did was to post here, asking us to do it for them.

Well the thought process went something like this:
1. I have a question
2. I search said question
3. I find this forum
4. I get excited I'm about to get learnt by some linux guru on how to best solve my problem
5. I realize the linux guru was just being a turd.
6. I click on the link provided, its dead.
7. I wallow in the irony of the situation. (A top search result for a problem contains someone telling people to search for their own answers, citing a top search result which is now dead)
8. I decide to be a warrior and register to put the linux guru in his place.


This is why sites like stack encourage people to answer the question directly instead of relying on outside links.
Preservation of knowledge and further enhancing the utility of the forum.

I think your thought process was more like this:

• I have a question
• I search for the question
• One of the links on the page says something that doesn't spoon-feed me an answer
• Rather than reading the answer and doing a FURTHER search to find what I was actually looking for (because I can't think about the fact that information/hardware/software CHANGES from one year to the next) I choose to whine about it

If you find it too hard to read or understand the many answers you'd have seen if you DID search for the terms the OP was originally given, that's not anyones problem but yours. Again, as the OP was told, there are TOO MANY VARIABLES to answer the question in a meaningful way. Any answers given now would differ quite a bit than those given years ago, so anything in this thread would have been outdated and not answered you any more than the (somehow INSURMOUNTABLE PROBLEM OF) the dead link would. How you calculate IOPS on a SAN volume over FC is different than a SATA, different than SAS, etc. Controller? Memory? CPU? Users?? ALL need to be taken into account.

Again, the 'top search result' wasn't this site for the search term given. Again: had you actually read the answers given and applied that search term you would have found your answer already, instead of whining about it.

And you're not a 'warrior'...you just appear to be a common troll. Grow up.

All members please remember the guidelines from the LQ Site Rules:

• Do not post any messages that are obscene, vulgar, sexually-orientated, hateful, threatening, hostile or insulting.
• Personal attacks on others will not be tolerated.
• Challenge others' points of view and opinions, but do so respectfully and thoughtfully ... without insult and personal attack. Differing opinions is one of the things that make this site great.
• When posting in an existing thread, ensure that what you're posting is on-topic and relevant to the thread. If the content of your post will interfere with the current discussion, you should start a new thread.

Calling each other names and conducting an in-thread argument should stop immediately.

I'd like to add that I very much agree that answering questions directly on site is preferable to a post with just external links for the very reason mentioned in this thread. In this case, the update URL is http://www.ryanfrantz.com/posts/calc...disk-iops.html but in many cases good content goes away permanently. LQ is very much meant to be an archive.

--jeremy

1 members found this post helpful.

As, typically, do I, Jeremy. However in this case, given the developments of storage media, controllers, etc., etc. (all the variables that go into calculating IOPS), there's no real meaningful answer. Updated links are available with a brief search that let the user apply their own specs to calculate things.

The OP was even told years ago that without all of the variables (and they were mentioned), the question couldn't be answered, and that still holds true. Not to mention when you start factoring in RAID or LVM, that skews what you'll be able to do, taking IOPS from the disk and putting some of that on the OS, taking THAT back to CPU, Memory, users on the system, system load, etc., etc.

This entire thread has no answer, past a general guideline about how to calculate things. And even then...it's an arbitrary number, which will fluctuate when ANY of the variables change, and (for the most part) is meaningless. What good is a hugely-fast disk, if your CPU bottlnecks? Or you run out of memory and swap?