|
Solid state drives (SSDs) have remade the storage industry in the last decade. They're costly, fast, and power efficient, but just how reliable are they? A new study of almost 1.4 million SSDs offers some surprising answers. Because spinning discs is so last century
|
|
|
|
|
Article wrote: from 0.53 percent for one 15TB drive versus 11.3 percent for its 15.3TB stable mate
 15TB ssd?
I just found another article speaking about a Samsung SSD of 15TB... around 13000$ 
M.D.V. 
If something has a solution... Why do we have to worry about?. If it has no solution... For what reason do we have to worry about?
Help me to understand what I'm saying, and I'll explain it better to you
Rating helpful answers is nice, but saying thanks can be even nicer.
|
|
|
|
|
I can't say as I notice much difference between SSDs and HDDs, except during boot (which I don't care about, because a: I almost never shut computers down, and b: if I have to, it's a good excuse to go and get coffee).
I do, however, max out memory, when buying a machine, which does make a big difference -- I only have machines with tons of memory and the biggest HDDs I can get my hands on, plus one laptop and one desktop with annoyingly small SSDs ("annoyingly small" incorporates anything under 2TB).
Most of the slowness of computers, these days, is to do with networks, not processors, chipsets, or discs -- a chain is only as strong as its weakest link, and all that. I've tested large compiles to be within microseconds of each other (which is only to be expected, if there's no disc-swapping involved). I have a feeling that a lot of the perceived "improvement" is subjective.
I don't even notice increased lag or latency in servers that don't have SSD storage -- again, the networks themselves are too slow for clients to notice a difference in read-access speeds (of course, system SSDs are good in this case, where boot times are somewhat more important).
So it's a lot of fuss about nothing, IMO. When they start producing large SSDs cheaper than large HDDs, I'll start considering them seriously.
I wanna be a eunuchs developer! Pass me a bread knife!
|
|
|
|
|
I think that you are missing the point. Computing performance is a matter of intelligently managing caching at different levels. As most people cannot afford 64GB (or more) of RAM, the next best step is to speed up the next cache layer, i.e. the hard disk. Given that SSDs are orders of magnitude faster than HDDs, but are not orders of magnitude more expensive, the choice is a no-brainer for most of us.
Freedom is the freedom to say that two plus two make four. If that is granted, all else follows.
-- 6079 Smith W.
|
|
|
|
|
Horses for courses. I'm happy with my way, other people are just as entitled to be happy with theirs.
I wanna be a eunuchs developer! Pass me a bread knife!
|
|
|
|
|
You are certainly right about the importance of caching. When you talk about 64GB of RAM, you should relate that to the "working set" concept: There are certain kinds of software that manages to address all of 64 GByte of data frequently - but not very many users need to simulate oil fields, or calculate their own weather forcast, or things like that. The major part of those 64GB are "never" referenced, once initialization is complete. Memory mapped executable files are great: The application code might fill gigabytes on disk, but if you make use of only 10% of the functionality, 90% may never be paged into RAM. Lots of users - even people with a computer related education - grossly overestimate the real RAM requirements. (Those who really need 64GB are not guessing/estimating, they know what they need - and can afford it. 64 GByte costs around 350-400 Euro nowadays, right?)
There is also a strong tendendy to overestimate the need for disk speed. Much large-volume data nowadays is real-time media. Once you have the capacity to stream HD video, even if you double your line speed, you won't be watching that two hour feature in one hour... Around here, fiber internet connection is standandard, 100 MBps is the minimum rate, enough to give each family member his own HD video stream. Yet people ask me why I won't upgrade to 500 Mbps, or 1 Gbps, for faster transfer. What for? Downloading a full length movie could be completed in a minute, rather than ten minutes, but if I download it, it is because I will watch it later; it makes no difference if the transfer completes a few minute later.
When I started video editing on my PC, the editor had to benchmark my harddisk to verify that it was fast enough. Since then, disk speeds have increased by a magnitude and a half; I never later had to benchmark any disk. Modern disks do extensive caching, and newer file systems allocating extents (rather than small pages) makes disk accesses and look-ahead caching far more efficient. Your video editing will not be much faster if you replace the old magnetic disk with an SSD. Very little of your time is spent waiting for the disk! We are no longer in the DOS age of PIO disk access where every byte had to go via the CPU registers - DMA came to PCs well above 20 years ago.
Even for CPU power, the real need for raw power is often overestimated. E.g. even today, I have collagues who point out as one great advantage of interpreted languages is that "you don't have to wait for the compilation". I run a complete rebuild, and typically, 6-8 compiled modules are emitted per second. Using either a make system explicitly or and IDE doing incremental builds, you rarely wait more than a single-digit number of seconds. Of course you could dream of a complete rebuild on every keystroke, but that is not a hard need.
Professionals working at supercomputer centers obviously do actual performance measurements. For the less professional ones, there percentage of "feel", "think" and "have a distinct impression" that the performance has improved, is quite high. There is very little blind testing of these feelings, very little measurement. I have a friend insisting that his monthly vacuum cleaning of the mainboard makes his PC significantly faster for a couple of weeks, until it is again slowed down from the dust  . In the old days of the floating point unit being separate (like 80287), there were lots of people on the internet insisting that Widows startup were sooo much faster after installation of an 87 (but MS asserted that there was not a single FP instruction in Windows code).
There are, of course, bottlenecks. But choose the bottlenecks to break with care! One example: I have been using the SketchUp 3D drawing system for many years. The first years I let SU steer my PC upgrade schedule. Then, after one SU update, I could forget about upgrades: Graph transformations had been moved out from the CPU to the graphics card. In a single sweep, the perceived performance improvment was larger than my previous five CPU upgrades. The graphics card had been hanging around for a long time, the question was about the software utilizing it!
I am not sure that hard disk speed is the most essential bottleneck to break for most of us, at least not for stationary PCs. For a laptop, power consumption, weight, no noise and shock tolerance may be essential, but not for a stationary PC. With SSD, startup is faster (both boot-up and application startup), but once running, you will rarley notice the difference. My main disk is an SSD, but I am just about to buy another two 8 TB disks dedicated to video storage. A plain magnetic disc can deliver several continous HD videostreams in parallell; I don't need more than that. SSD costs 3-4 times as much, and I would gain nothing in practice. (Besides, SSD units of 8TB is not common, so I would run out of SATA sockets).
|
|
|
|
|
I agree with a lot of what you say. i.e. I have 50 Mb connection and I upgraded because I added landline and the price was the same as before 16 Mb...
But here...Member 7989122 wrote: With SSD, startup is faster (both boot-up and application startup), but once running, you will rarley notice the difference.
I have to disagree... using the PC of my father-in-law was a patience exercise. I convinced him to change to an SSD and it is much faster even working with it.
M.D.V. 
If something has a solution... Why do we have to worry about?. If it has no solution... For what reason do we have to worry about?
Help me to understand what I'm saying, and I'll explain it better to you
Rating helpful answers is nice, but saying thanks can be even nicer.
|
|
|
|
|
Windows 10 looks set to get a redesigned Start menu in a future update, with Microsoft finally killing off Live Tiles for good. No, not *that* useless feature, the other one
No - the other, other one
Yeah, that one.
|
|
|
|
|
Why can I not stop thinking that they are going to mess it up and do it even worst with something "new and cool" to replace what they are deprecating?
M.D.V.
If something has a solution... Why do we have to worry about?. If it has no solution... For what reason do we have to worry about?
Help me to understand what I'm saying, and I'll explain it better to you
Rating helpful answers is nice, but saying thanks can be even nicer.
|
|
|
|
|
Getting back the W7 Start Menu ?
And may be Office classic menu too 
Patrice
“Everything should be made as simple as possible, but no simpler.” Albert Einstein
|
|
|
|
|
Even for phones, live tiles never made sense.
|
|
|
|
|
So dead tiles now, then.
A dull, lifeless interface with ten times as much "whitespace" (in dull pastel rectangles) as usable area, because horizontally-sorted boxes full of whitespace are sooo much more functional and easier to use than tight vertical lists.
The ridiculous and annoying amazingly great horizontal sorting of the applets in the Control Panel proved that.
Just don't hope that anything they do will be an improvement -- expecting a better outcome with the same people making the decisions is foolish.
I wanna be a eunuchs developer! Pass me a bread knife!
|
|
|
|
|
Yes, yes, yes... but is the icon going to change?
“That which can be asserted without evidence, can be dismissed without evidence.”
― Christopher Hitchens
|
|
|
|
|
GuyThiebaut wrote: Yes, yes, yes... but is the icon going to change?
All of 'em will be changed again!
Busy and productive times lie ahead for the great creative minds at microsoft!
I wanna be a eunuchs developer! Pass me a bread knife!
|
|
|
|
|
Firefox will start switching browser users to Cloudflare's encrypted-DNS service today and roll out the change across the United States in the coming weeks. Now, no one will know I'm visiting 76.74.234.210!
Sadly, US-only for the moment, but fingers crossed.
|
|
|
|
|
Kent Sharkey wrote: US-only for the moment
By default. You can still turn it on manually in other countries.
"These people looked deep within my soul and assigned me a number based on the order in which I joined."
- Homer
|
|
|
|
|
So that's why FF is going so damn slow since last update?
M.D.V.
If something has a solution... Why do we have to worry about?. If it has no solution... For what reason do we have to worry about?
Help me to understand what I'm saying, and I'll explain it better to you
Rating helpful answers is nice, but saying thanks can be even nicer.
|
|
|
|
|
I understand that in Canada, Trudeau will personally route all internet traffic.
|
|
|
|
|
And apologize for it.
TTFN - Kent
|
|
|
|
|
Last week, an email popped into my mailbox with a simple subject: "Jif vs. GIF." Its sender asked if I was interested in hearing about a peanut butter producer's interest in "setting the record straight on how to pronounce GIF." Everyone knows it's pronounced 'GIF'
|
|
|
|
|
"Graphics Interchange Format", so not "Jraphics Interchange Format". Simple.
Bastard Programmer from Hell 
If you can't read my code, try converting it here[^]
"If you just follow the bacon Eddy, wherever it leads you, then you won't have to think about politics." -- Some Bell.
|
|
|
|
|
It's pronounced, Ob-se-leet
|
|
|
|
|
No-one should need more than 256 colours!
I wanna be a eunuchs developer! Pass me a bread knife!
|
|
|
|
|
Tech companies are famous for having some of the world's most challenging and competitive interview processes. Almost as far apart as I'd like to be from the interviewer?
Or can I hang the interviewer from the cable?
No interviewers were harmed in the posting of this blurb.
|
|
|
|
|
You don't need to be a maths-professor to understand MSDN. Being good at maths (abstractions) was good advice in 1970 if you wanted to go into "computing".
Nowadays, patience and perseverence gets you were maths can't.
Bastard Programmer from Hell
If you can't read my code, try converting it here[^]
"If you just follow the bacon Eddy, wherever it leads you, then you won't have to think about politics." -- Some Bell.
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
