Hot answers tagged

97

One aspect of my job is designing and building large-scale storage systems (often known as "SANs", or "Storage Area Networks"). Typically, we use a tiered approach with SSD's and HDD's combined. That said, each one has specific benefits. SSD's almost always have a higher Cost-per-Byte. I can get 10k SAS 4kn HDD's with a cost-per-gigabyte of $0.068/GB USD. ...


61

This is true, and it was one of the key motivation to backing the switch from SLC (fast and durable flash cells, but small capacity) to MLC (slower and less durable flash cells, but bigger capacity). To give you some ballpark numbers (on old 34nm tech): SLC drive: 100K P/E cycles (program-erase cycles), 100 GB in size, 10 DWPD (Drive Writes Per Day) x 5y, ...


53

The single biggest thing that crosses my mind isn't SSD-specific: that the biggest danger with RAID is that all the devices in any given RAID are often purchased from the same manufacturer, at the same time, and therefore tend to get to the far end of the bathtub curve and start dying at about the same time. In that sense, buying from different vendors is ...


50

Don't attempt to "wipe" an SSD with tools designed for spinning magnetic hard drives. You won't actually destroy all the data, and you'll just reduce the lifetime of the SSD. Instead, use an erase tool specifically designed for SSDs, which can use the drive's internal flash erase (discard) to discard all of the blocks, including the ones you can't access. ...


36

Drive manufacturers specify the reliability of their products in terms of two related metrics: the annualized failure rate (AFR), which is the percentage of disk drives in a population that fail in a test scaled to a per year estimation; and the mean time to failure (MTTF). The AFR of a new product is typically estimated based on accelerated life and stress ...


27

SSD are faster because there's no network latency, but it is ephemeral and you can't detach it from an instance and attach it to another. As you can see, it is available to more powerful instances. EBS are more flexible, since you can attach and detach it from instances, but is a little bit slower, as more suitable for general purpose. Now, in Step 4, you ...


26

I was struggling with the same problem on my notebook, but as I reboot it pretty much on a daily basis, the accepted answer wasn't helpful. I have a Samsung mSATA SSD, which happens to have the SMART attribute #241 Total_LBAs_Written. According to the official documentation, To calculate the total size (in Bytes), multiply the raw value of this attribute ...


26

Look for an industrial or ruggedized SSD for this application. A good example of a proper product spec. http://www.pretec.com/products/ssd-series/item/sata-ssd-series/a5000-industrial-grade .Standard 2.5" SATA III SSD, compatible with SATA III/II/I interface .Capacity: 32GB ~ 256GB .Data transfer rate: Up to 490 MB/s .Built-in ECC (Error Correction Code) ...


26

Let's try to reply one question at a time: Is TRIM support necessary for modern (2015-2016 era) SSDs? Short answer: in most cases, no. Long answer: if you reserve sufficient spare space (~20%), even consumer-grade drive usually have quite good performance consistency values (but you need to avoid the drives which, instead, choke on sustained writes). ...


22

Note: This answer is specific to the server components described in the OP's comment. Compatibility is going to dictate everything here. Dell PERC array controllers are LSI devices. So anything that works on an LSI controller should be okay. Your ability to monitor the health of your RAID array is paramount. Since this is Dell, ensure you have the ...


20

Short answer: This is the results of network latency and a serial workload (as you imposed by using direct=1, sync=1 and iodepth=1). Long answer: using direct=1, sync=1 and iodepth=1 you created a serial workload, as new writes can not be queued before the previous write was committed and confirmed. In other word, writes submission rate strictly depend on ...


19

Unfortunately the MTBF isn't what most people think... It is not how long an individual drive will last. Manufacturers expect their drives to last as long as the warranty, after that it really isn't their problem. Older electromagnetic platter hard drives will seize up after 10 or so years. Integrated circuits last an extremely long time, but other ...


18

ATA Secure Erase is part of the ATA ANSI specification and when implemented correctly, wipes the entire contents of a drive at the hardware level instead of through software tools. Software tools over-write data on hard drives and SSDs, often through multiple passes; the problem with SSDs is that such software over-writing tools cannot access all the storage ...


18

HDD is still quite preferred Is it? I'm not sure it is to be honest. HDD's come in large sizes for a decent price right now, that's undeniable, and I think people trust them for longer data retention than SSDs too. Also when SSDs die they tend to die completely, all in one go, whereas HDDs tend to die in a more predictable way that maybe allows more time ...


16

While the other answer here beings up some points, your specific issues are due to platform limitations and OS configuration: You're throughput-limited by the use of consumer SATA SSDs on an HP Smart Array P410 RAID controller. SATA disks run at 3.0Gbps (3G) on these controllers rather than 6.0Gbps (6G). So that's a barrier that impacts the read speeds of ...


16

If used with SSDs without powerloss-protected write cache, the RAID controller's NVCACHE is extemely important to obtain good performance. However, as you are using SSDs with powerloss-protected write caches, performance should not vary much between the various options. On the other hand, there are other factors to consider: with hardware RAID is often ...


15

Sure, you can do this. You can tell ESXi that a particular datastore is comprised of SSDs. Here's how:Enabling the SSD option on SSD based disks/LUNs that are not detected as SSD by default (2013188) You end up having to set a datastore claim rule similar to: esxcli storage nmp satp rule add --satp SATP_TYPE --device naa.ID --option "enable_ssd" This ...


15

When suddenly losing power, MLC/TLC/QLC SSDs have two failure modes: they lose the in-flight and in-DRAM-only writes; they can corrupt any data-at-rest stored in the lower page of the NAND cell being programmed. The first failure condition is obvious: without power protection, any data which are not on stable storage (ie: NAND itself) but on volatile cache ...


14

We don't proactively replace disks. Wait until they fail or report prefailure status. This is why you have (hardware) RAID, management agents, failure indication LEDs on the server and a monitoring solution. Keeping your gear under warranty and maintain your support contracts (within reason), having cold spares of components that fail often (disks, power ...


14

MB/s states how many Megabytes per second the drive can handle as throughput. IOPS states how many single operations per seconds can be handled. Sequential access means that for example one big file is read, random access means you're reading single parts of different files. If you look for a drive for database usage, you should look for: An enterprise, ...


13

TRIM works at the filesystem level so as you're giving your KVM domains a raw block device then you need to enable TRIM from within the domain; your host can't know the domains filesystem utilisation without examining it. To enable this you need to ensure there's a discard='unmap' attribute added to the disk definition in the XML for the domain, this is ...


13

Q1: Are any of these configurations at risk for data loss or corruption on power loss? A1: You shouldn't have any issues, unless you'll configure cache in write-back mode, and w/out NV RAM. Q2: Which configuration should I expect to have the best write performance? A2: One having biggest amount of cache obviously! ...and no parity RAID, but RAID10 ...


13

SSDs wear out when you use up their block erase cycles. Each block can only be erased so many times. Larger SSDs have more blocks, so that means more block erase cycles. All other things being equal, you can write twice as many TB to a 1TB SSD as you can to a 512GB SSD before it wears out. Frankly, I wouldn't buy a bigger SSD to get a longer life though. A ...


12

Oh dear, where to begin? There is so much involved, and you need a good understanding of everything. Just throwing a bunch of disks against a RAID controller won't yield the results you're looking for. This can't be easily answered. But at least, here is a list of stuff you have to look at: Does the Controller even have the throughput needed? (-> ...


11

In theory yes, more drives in a raid0 would lead to higher performance because the load is shared over more drives. However in practice you would be limited by the the bandwidth of the raid controller, the CPU and memory performance and similar. The performance increase would not be linear, that is 4 disks is not exactly twice as fast as 2 disks. In any ...


11

Yes, larger SSDs have higher endurance. There's a couple of factors involved here, and it's not as simple as it appears: Larger SSDs have more NAND inside them, and any half-decent SSD supports wear leveling so that all the writes are spread evenly over the NAND. As a result, regardless of how much data you put on the drive, the simple fact that there's ...


11

Yeah. Don't get super cheap SSD - anything outside the low end consumer market has capacitators and full protection against power loss. Amd really does not cost that much more.


11

The measured low performance are the results of various factors: after creation the array is entirely synched, causing the allocation of most (if not all) flash data pages on half the SSDs. This will put the SSDs in a low performance state until a secure erase / trim "frees" all/most/some pages. This explain the increased performance after an fstrim; the (...


11

Those numbers are obviously bogus. Your SSD doesn't have that many sectors at all, let alone offline uncorrectable ones. Check for firmware updates for your SSD that will fix the problem. Otherwise, ignore it. If you're really paranoid, you could replace the drives, but in that case you should use different drives, not the same brand from the same ...


11

On Dell's R210, R220 and T130 I've successfully used Sandisk's consumer grade SSDs for years. I've done this with PERC's H410 and H710 controllers. It will be a matter of controller and controller's firmware more than server brand/model. Also, you'll probably have issues to find disk trays. I ended up buying it used on a famous refurbished hardware shop. ...


Only top voted, non community-wiki answers of a minimum length are eligible