Windows 2003 ramdisk

A RAMDisk acts as a virtual drive on your system. It allows you to create directories, copy files to and from it, etc…. However , the data is not written onto a hard disk, but remains purely stored into a particular part of the physical RAM memory.

This makes them relative lazy. A RAMDisk does not need to seek , and by this , it can read and write the same data to upon times faster than a hard disk! It supports the loading and saving of images for persistent storage options. You may use the program to mount image files as well. Actually, it offers two options here. You may select drive letters during setup as well, and may set them to read-only. Data can be saved to image files so that you can load it in the future again.

You may still download it from portals such as Major Geeks though. The program ticks all the right boxes. It does not limit the RAM that you may use to create RAM disks, and you may create as many as you have free drive letters for.

It supports the saving of images so that data persists across sessions, a lot of command line switches, file system selection for each RAM disk, and an option to launch programs when RAM disks load. StarWind requires you to create an account on the website before the download link of the company's RAM Disk software is sent to the email account you used to register the account.

The program may be used to create multiple RAM Disks that are not size limited. It offers basic functionality besides that however: you cannot assign drive letters to disks, and there is no option to save and load disk images as well. The following programs are not listed in the main section above for one reason or the other. They may still prove useful to you in some situations:. The overview and description of each program is certainly helpful in determining the best suitable program for your use cases.

It offers great functionality, and the downside that it is no longer maintained does not bother me as it works fine and without issues. My Firefox cache is set to go there. A free trail can be downloaded though. I have some junctions in place redirecting app folders to it, and a boot time batch script will pre-fill it with data exactly the way I want it. Never a problem. I use a ram disk to hold the sandbox created by Sandboxie.

Downloading and reloading a 64 bit version is probably better if you have more than 4GB, but this is only a FYI. Broken page here. The banner at the top of the page, with the menu buttons, is turned into text. Colour background of comments is gone. Edit 1 : just publishing this comment corrects the problem. This has happened me once before, and I had deleted my warning comment. Edit 2 : under Palemoon and Opera, refreshing the page takes care of the problem.

This has begun to happen since changes were announced to the blog interface. Under Palemoon and Opera no extensions , the faulty page opened also through https. Not sure how I got there. Probably through a search engine. Now the broken page is here again in Firefox, https. Just refreshing the page before publishing the comment does not correct the problem.

Under Palemoon and Opera, cleaning cache and everything since ever, then loading Ghacks brings it through http, and the layout is correct. Displayed address is still https.

The yellow warning says : connection not secure, parts of this page are not secure such as images. Edit 3 : once the above comment has been moderated and published, to a correct webpage, just refreshing the page breaks it again.

I may have fixed the issue. Can you please try again, do a refresh Ctrl-F5 , and let me know please? Thanks for reporting the issue. Letting it run would make gHacks opened via https. This is performed with an empty cache. Https at this time is fluid, renders pages as fast as http, and renders them complete.

Amazing service! Thank you! I was surprised to read this article. I thought that RAM disks had gone the way of the dodo, and had been rendered obsolete by modern operating systems as well as the enormous power of contemporary hardware.

Windows 95 era? Maybe there are some specific situations where some very knowledgeable users could still take advantage of such software. I would love to learn more about that. This is far more problematic than the few milliseconds you might save here and there. Generally, wholeheartedly agree. Most modern, mainstream software will make plentiful use of your ram is need be and its available which negates most RAM disk benefits. And data security outweighs any gains.

I have however a niche use case where there are significant benefits. I work as a structural engineer designing buildings.

This still allows for concurrent loading, but has the following advantages: Minimal use of transaction log space for the loading process Concurrent loading from multiple clients You won't have the index problems Read the bulk loading stuff in books online.

I'm not looking at the , but BOL, but I believe the process is still exactly the same. It also has guidelines on how much data you have to load for the index drop and recreation process to make sense.

That won't work for various reasons. In fact, that is the way we were doing things Now we have to move to a system where large amounts of data are constantly streaming into the database. And at times it would allocate that for me then fail after a certain period of time. I think the i-RAM might be your best solid and reilable bet. Well, if my test with the software RAM drive shows an improvement, then we will probably go with a hardware RAM drive. At any rate, I'm pretty sure a software RAM disk is a bad idea.

Froogle www. Is this SQL or ? Several people have informed you that SQL does a good job of managing memory. If you intend to try a memory-resident RAMDrive, increase the system memory by the same amount.

This will give a fair comparison. This is for the card and battery, but no memory. Frankly, I'm not totally convinced that if the penalty for writing the Index is so large that it shouldn't just be removed. The index sounds rather small if a MiB ramdisk would cover it. Is it really that critical? I'm not sure if you considered this already, and my experience is with Oracle so apologies if this is not relevant.

In Oracle you can create a table space for you indexes indices and put them on a separate disk volume. If indexes are a bottleneck you could easiy make that disk volume a RAID0. In fact you could stripe it over 3 disks, or even 4 disks to make it super fast.

You don't care so much that RAID0 provides no redundancy because you can quickly rebuild the indexes if one drive goes bad. Just something to think about. Or could it not have a device driver that allows it to be addressed as a block device aka PCI bus?

Hardware RAM disks show up as and are accessed like any other block device. The problem is that you either give up the RAM for good if it wires the pages in, or you're stuck being backed by slow disk if it pages out under pressure. Doh, my bad. It is SQL Server That really doesn't matter. If it ends up being about the same, then you could argue that I have not proven something either way I'm not convinced either - hence why we are going to test the hypothesis. The indexes are not huge, but they definitely have an impact on performance.

Right, and by testing our assumptions using a software RAM drive, we can test the hypothesis for free. I understand you guys that are adamant about SQL Server being agressive about caching, and knowing how to best manage available memory, but in this case I am testing a specific hypothesis: "Is the physical IO activity due to page splits actually causing the slowdown we think it is, and will moving the indexes to a much faster device improve performance".

You have ot be careful here. Using the wrong software product will not give you applicable results. If it does any paging, you lose. It seems to me you can confirm the writing part of this hypothesis just by turning the index off. It also seems it's a much fairer and accurate test anyway. This is a well-framed statement and your hypothesis is very testable as stated. These profiling options will allow you to turn on profiling during specific or general activities. And will measure which SQL operations, tables, and disk operations are proportionally using resources.

It graphs resource usage, breaks out specific SQL statements, and give you recommendations for performance improvements. Does SQL Server have anything like that to aid in your benchmarking analysis? Keep us up to date. This will be interesting. I would however presume they're running bit SQL Server. I'll go ahead and just give away the inevitable answer - it won't do jack squat. Even the hardware versions have been tested replacing the OS's pagefile on an iRam drive. I'm assuming you've done the following, but really need to make sure: Have you checked the server in question to see how much memory SQL Server is set to use?

How are your drives set up?