Thats why consumer grade ssd work often very bad with storage spaces. The better the discs are - so the better the average latency is - the better works the scaling. If the data-chunk (slab) is striped (more than one column) it does just mean, that reads and writes are probably faster because more discs are involved for each read/write. Well it does never become more efficient in Mirror-Setups as every data-chunk (slab) is stored in two different locations without calculated parity in "Two-Mirror-Setup" and three times the Space in "Three-Way-Mirror". But I think you did missunderstood a bit how mirroring works in SP. Note: Please follow the steps in our documentation to enable e-mail notifications if you want to receive the related email notification for this thread. If the Answer is helpful, please click " Accept Answer" and upvote it. For example, 4-disk 2-way in 2-column mode is essentially a striped mirror, but without being able to withstand a 2-disk failure. So, almost every other drive in the pool has something to contribute to the repair!īut the difference is that Storage Spaces can tolerate 2 disks failed if you got 3-way replication configured or double parity! If you have a two way mirror storage space with more than two disks, it can only tolerate one disk failed. But, these surviving copies are evenly distributed across almost every other drive in the pool! One lost slab might have its other copy on Drive 15 another lost slab might have its other copy on Drive 03 another lost slab might have its other copy on Drive 07 and so on. And we do need to read from each slab's surviving copy in order to repair. We do lose all the slabs stored on that drive. Likewise, when drives fail, all volumes are affected for the very short time it takes to repair them.Ĭonsider the scenario where some drive fails. For example, using two-way mirroring with three servers does not leave one server empty. Storage Spaces does not keep whole copies of volumes – rather, it divides them into tiny 'slabs' which are distributed evenly across all drives in all servers. Thanks you for posting such interesting questions. The more discs are involved, the longer is the time for the ACK.
For Storage Spaces latency as low and stable as possible is crucial. What I don't like about most actual SSD's is the fact that IOPS raise but latency and specially average latency is going worse.
Two way mirror Offline#
Does Storage Spaces act the same in *two-way-mirror" setups when usingen more than two disks? So does it use automatically metadata copies on separate discs than the 256MB chunks? For example a one column two-way-mirror with three or four drives.
Two way mirror upgrade#
It makes sense to me that metadata is looked separately.Īs I have to Upgrade one of my first installations, i ask me some questions: It is really hard to find information how Storage Spcades is acting in reality but some time ago I've found a good explanation for the Three-Way-Mirrors and why it needs five discs as a requirement.
(spoiler: I'm somewhat unhappy, that I need two DC licenses instead of STD for small clients for SSP direct).Īs Space is increasing and quality of SSD's is +- falling with the raise of layers I'm thinking about Three-Way-Mirrors but also thinking about two-way-mirror other ways. First just for VDI, since some years for everything. Always in Two-Way mirror setups with good/best quality and low latency SSD's to have low maintenance cost and good low queue depth performance and easy setup. I've been using Storage Spaces since almost first days.
0 kommentar(er)
