Linux - ServerThis forum is for the discussion of Linux Software used in a server related context.
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
Hello all. I apologize, this is sort of mixed between hardware and software but it seemed more appropriate to me here. I'm building a server for very fast disk access. We have 8x32GB SSD SATA drives, and 4x300GB SSD SATA MFT drives. The 300GB SSD drives are the slow kind of flash that writes slowly, and strangely is limited to 10K writes per sector. Long term data integrity isn't a big deal because it is backed up continuously but fast access to data is desired.
Additionally the filesystem that contains this data deletes about 2.5 - 4 gigs of data per day, and adds about 2.5-4 gigs of data per day. My plan is to create a hybrid drive of sorts, where the smaller 32G drives, lined up in RAID0, create a fast "buffer" disk, and on some increment what is in the buffer is written in bulk to the slower writing 300GB SSDs. I had two thoughts on how to achieve this, but ultimately I think that LVM snapshots are the best way to achieve this, put the read only "snapshot" on the big SSD drive and the other "differencing" part of the snapshot on the faster raid0.
I'd much prefer a simpler solution where there is one block device to mount and all this is handled in the background. Can anyone comment on this idea, tell me if its crap, and if there is a better way to accomplish my goal?