Value System Build

[SushiEater]

malch wrote:

SushiEater wrote:

Trying? Hundreds maybe thousands of attempts, per minute.

Are you saying Amazon is hacked?

No, I'm saying Amazon is subject to a constant stream of attacks; most if it automated scanning, probing for vulnerabilities as well as brute force attempts to crack weak passwords.

Run any tiny web site and you can see the bots probing constantly.

I am running several websites and no one hacked me yet. We also have one hosting on Amazon servers and I think we had DOS once in 6 months.

 

[malch]

SushiEater wrote:

These are two totally separate issues.

I don't like parroting like you do but just for you:

The point of SSD and RAID 0 is to make computer faster.

I can live without SSD by simulating its performance as I already explained how.

You're "simulating" one aspect of SSD performance. Launching a program that is already cached in RAM. So what? That works just the same with RAID 0, single HDD's, optical disks, or a binary array of cheese and pickle sandwiches.

You're drawing conclusions about SSD's based on a silly benchmark that doesn't even involve the SSD.

If your point is... generous RAM and file system caching can deliver fast loading programs a good deal of the time... I would absolutely agree with you. It can and nobody is disputing that.

For for many of us, SSD's are not just about fast program loads. They're about sub millisecond random access times to storage. And you can't "simulate" that with RAID 0.

And yet it still has 1 SATA 3 port filled with SSD and 8GB of memory.

I just said it has 16GB. Let me check again... yep, 16GB still present and correct.

You mean you are using SD card for memory? Or you have 16GB single DDR3 single stick?

Four 4GB DDR3 sticks @ 800MHz. Windows Performance Index of the RAM is a respectable 7.8.

 

[Jim Cockfield]

SushiEater wrote:

Yet, most spam I get is from Gmail people. I don't even remember getting one from Yahoo.

You must be living in a bubble, since compromises of Yahoo mail users have been in the news a lot over the past year.

Here's one article on the subject:

http://www.hotforsecurity.com/blog/yahoo-accounts-hijacked-via-xss-type-attack-5172.html

Here's another:

http://www.hotforsecurity.com/blog/yahoo-accounts-hijacked-via-xss-type-attack-5172.html

Those vulnerabilities were present for a number of months, even after they had supposedly fixed them the first time (there are earlier articles about the same kind of thing).

Criminals took advantage of the ability to hijack Yahoo e-mail accounts by the use a vulnerability to steal a cookie that tells yahoo you're already logged in to an account.

So, no cracking of passwords was needed. All they needed to do was get a yahoo user to click on a link in their e-mail and they could then get full access to that user's Yahoo account by stealing the cookie associated with login credentials.

Then, they'd send the same kind of mail to compromise other accounts. For example, you'd see an e-mail from someone you know (with an already compromised account, thanks to it being hijacked), and click on the link in it so that they'd compromise your e-mail account, too (and the problem continued to escalate for months, with more and more compromised Yahoo Mail accounts).

Sure, other sites have experienced problems with vulnerabilities, too.

But, Yahoo Mail has had more than other mail sites from what I've seen in the press.

As for Gmail, I suspect that someone had malware on their system stealing login credentials if you saw the same thing from Gmail users, as Google has been a lot better than Yahoo at keeping security from vulnerabilities tighter.

Or, another site the users were logging into was hacked and their e-mail credentials were stolen. That happens a lot anymore. But, Yahoo's reputation is pretty bad in that area.

Also, Gmail has one of the best spam filtering systems available (if not the best).

--
JimC
------

 

[malch]

SushiEater wrote:

I am running several websites and no one hacked me yet. We also have one hosting on Amazon servers and I think we had DOS once in 6 months.

If you examine the web server and firewall logs, I guarantee that there have been many attempts.

Sure most are unsuccessful. These people are playing a numbers game rather like spammers. They send out billions of emails (or probes) in the hope that one or two will be successful.

 

[malch]

SushiEater wrote:

Nobody has suggested you use SSD's for long term storage.

If you read carefully others responses, they do.

I don't think so. They've suggested using SSD's for processing the data -- not for long term storage.

Probably not if you're moving those files to different folders on the same device. You need to think about the manner in which file systems really work.

It takes several seconds to copy so yes files actually being copied.

I suggest you look into that. Within a given device, it should be possible to move files without copying. It's a lot faster and, in the case of SSD's, it will reduce wear.

Sound like there's some good scope for streamlining your workflow.

C'mon I don't need precise tools. Just not important. The point is that a lot of writing and deleting can destroy SSD faster than HD. That thing I did with panos was just an exercise to prove you wrong. I still got 7 years to go on the drive so who cares.

You're basing your conclusions on the output of a program that I'd call guessware. It's less about precision and more about fundamental credibility. Anyone can say your SSD will last 968.6328 months. Just because there's 7 digits of precisions does mean it's right or even in a sensible ballpark.

 

[SushiEater]

Every single day someone is getting hacked but if it does not effect me I would not care.

I don't read newspapers or technical bulletins like you guys do just like I don't read who got murdered today or who got married or who is in the new scandal.

I do read about digital cameras and this site for camera news.

 

[SushiEater]

malch wrote:

SushiEater wrote:

Nobody has suggested you use SSD's for long term storage.

If you read carefully others responses, they do.

I don't think so. They've suggested using SSD's for processing the data -- not for long term storage.

After I clearly said that processing the data and "long term" storage is the same thing for me because there is so much data is being written and erased on my computer.

Probably not if you're moving those files to different folders on the same device. You need to think about the manner in which file systems really work.

It takes several seconds to copy so yes files actually being copied.

I suggest you look into that. Within a given device, it should be possible to move files without copying. It's a lot faster and, in the case of SSD's, it will reduce wear.

I have already looked in to it.

Sound like there's some good scope for streamlining your workflow.

C'mon I don't need precise tools. Just not important. The point is that a lot of writing and deleting can destroy SSD faster than HD. That thing I did with panos was just an exercise to prove you wrong. I still got 7 years to go on the drive so who cares.

You're basing your conclusions on the output of a program that I'd call guessware. It's less about precision and more about fundamental credibility. Anyone can say your SSD will last 968.6328 months. Just because there's 7 digits of precisions does mean it's right or even in a sensible ballpark.

Sometimes a good guess is all I need. I am not a stickler for precision then not needed. It wouldn't make any difference if it is off by a year because I will replace that SSD way before it actually dies. And if you remember I said I clone SSD that means that the second SSD is not used as much so I will do cloning one last time and put it in when time comes. Yes, it will be 5-6 years old technology but as I explained before it does not matter how fast new SSDs are because SSD have the lowest WOW factor. IOW SSD is not what saving me time besides I am still putting my computer to sleep after 10 minutes of inactivity.

 

[SushiEater]

malch wrote:

SushiEater wrote:

These are two totally separate issues.

I don't like parroting like you do but just for you:

The point of SSD and RAID 0 is to make computer faster.

I can live without SSD by simulating its performance as I already explained how.

You're "simulating" one aspect of SSD performance. Launching a program that is already cached in RAM. So what? That works just the same with RAID 0, single HDD's, optical disks, or a binary array of cheese and pickle sandwiches.

You're drawing conclusions about SSD's based on a silly benchmark that doesn't even involve the SSD.

No, my conclusion is based on my workflow. Which is also the workflow of millions.

It is not a benchmark. Just simple way of computing. If you want to shut down your computer every time you are not using it be my guest.

If your point is... generous RAM and file system caching can deliver fast loading programs a good deal of the time... I would absolutely agree with you. It can and nobody is disputing that.

For for many of us, SSD's are not just about fast program loads. They're about sub millisecond random access times to storage. And you can't "simulate" that with RAID 0.

First of not many but few of you. I don't even know what exactly you do either.

Editing on SSD is not any faster than on RAID 0, at least in my case and I already proved it to myself.

And throwing "sub millisecond random access times" is not going to explain to anyone unless you are going to present the real life example that most people use.

And yet it still has 1 SATA 3 port filled with SSD and 8GB of memory.

I just said it has 16GB. Let me check again... yep, 16GB still present and correct.

You mean you are using SD card for memory? Or you have 16GB single DDR3 single stick?

Four 4GB DDR3 sticks @ 800MHz. Windows Performance Index of the RAM is a respectable 7.8.

On what motherboard?

 

[malch]

SushiEater wrote:

No, my conclusion is based on my workflow. Which is also the workflow of millions.

Most people don't write GB files. Most people write less than 5GB per day.

Editing on SSD is not any faster than on RAID 0, at least in my case and I already proved it to myself.

Having the cache on an SSD will typically help.

And throwing "sub millisecond random access times" is not going to explain to anyone unless you are going to present the real life example that most people use.

Lots of random access. Many database applications for example. Esp. if they're large and unlikely to be cached in RAM.

Also great for lots of multi-tasking. Take a look at what happens when two different applications are hammering the same disk; big slowdowns when drives start constantly seeking between two active areas of disk. Performance nosedives rather like a system that's out of RAM and thrashing the swap file. Try running (just as a simple example) an AV scan and a backup on the same disk at the same time. Seek, seek, seek, seek...

Think about what HDD's have to do:

And tell me you don't want to eliminate all of that mechanical nonsense

Four 4GB DDR3 sticks @ 800MHz. Windows Performance Index of the RAM is a respectable 7.8.

On what motherboard?

Dell XPS 8500. Yeah, pretty boring but damned cost effective when purchased refurb and on sale.

 

[SushiEater]

malch wrote:

SushiEater wrote:

No, my conclusion is based on my workflow. Which is also the workflow of millions.

Most people don't write GB files. Most people write less than 5GB per day.

Regardless of the size workflow remains the same.

Now if someone goes for a single drive at first and then decides to to do something that would require large files (some databases are very large too not just photos) they would be screwed. But if they start with RAID 0 to begin with they be okay no matter what they do. Always think ahead.

Editing on SSD is not any faster than on RAID 0, at least in my case and I already proved it to myself.

Having the cache on an SSD will typically help.

I have cache now on SSD. Why do you think I am writing 100GB per week?

And throwing "sub millisecond random access times" is not going to explain to anyone unless you are going to present the real life example that most people use.

Lots of random access. Many database applications for example. Esp. if they're large and unlikely to be cached in RAM.

How many home users do you know who use large databases like that?

Do you know how many entrees would have to in the database to be that large to require SSD for random access? C'mon be serious.

Also great for lots of multi-tasking. Take a look at what happens when two different applications are hammering the same disk; big slowdowns when drives start constantly seeking between two active areas of disk. Performance nosedives rather like a system that's out of RAM and thrashing the swap file. Try running (just as a simple example) an AV scan and a backup on the same disk at the same time. Seek, seek, seek, seek...

Maybe on the slow system like yours but not on mine. I ran Malwarebytes full scan and copy to external HD just an hour ago. No problem at all. Even on the slow system, how often do you do it?

Think about what HDD's have to do:

And tell me you don't want to eliminate all of that mechanical nonsense

Sure, if you give me 6TB SSD with longevity of 7 years for $200 I will go for it.

BTW, this video is nothing. I have seen heads going so fast is like a blur.

Four 4GB DDR3 sticks @ 800MHz. Windows Performance Index of the RAM is a respectable 7.8.

On what motherboard?

Dell XPS 8500. Yeah, pretty boring but damned cost effective when purchased refurb and on sale.

That must be the custom motherboard I have never seen before.

BTW, why do you think all HD manufacturers using SATA3 interface if their hard drives can't even get to the speed of SATA2?

 

[malch]

SushiEater wrote:

Most people don't write GB files. Most people write less than 5GB per day.

Regardless of the size workflow remains the same.

But SSD media wear is a complete non-issue.

Lots of random access. Many database applications for example. Esp. if they're large and unlikely to be cached in RAM.

How many home users do you know who use large databases like that?

Do you know how many entrees would have to in the database to be that large to require SSD for random access? C'mon be serious.

Quite a lot. e.g. Lightroom users with large collections. They can have many GB of previews.

Also great for lots of multi-tasking. Take a look at what happens when two different applications are hammering the same disk; big slowdowns when drives start constantly seeking between two active areas of disk. Performance nosedives rather like a system that's out of RAM and thrashing the swap file. Try running (just as a simple example) an AV scan and a backup on the same disk at the same time. Seek, seek, seek, seek...

Maybe on the slow system like yours but not on mine. I ran Malwarebytes full scan and copy to external HD just an hour ago. No problem at all. Even on the slow system, how often do you do it?

Doesn't sound like you were hitting the same drive. If you were, there would be conflict.

And I have multiple apps hitting different folders with heavy physical i/o (not RAM cache) on the SSD every day. If I swapped my SSD back to a HDD, I'd hear continuous seeking. In fact, the silence of my first SSD was one of the Wow factors.

And tell me you don't want to eliminate all of that mechanical nonsense

Sure, if you give me 6TB SSD with longevity of 7 years for $200 I will go for it.

I'll be happy to do so as soon as I can buy one for $150.

BTW, this video is nothing. I have seen heads going so fast is like a blur.

30 years ago long seeks took something like 25ms. Most drives today are still over 5ms. Seek times haven't improved that much relative to other aspects of computer technology and nor are they likely too. This and rotational latency are why HDD's are doomed. They will be replaced by solid state technology. The only question is how quickly.

That must be the custom motherboard I have never seen before.

BTW, why do you think all HD manufacturers using SATA3 interface if their hard drives can't even get to the speed of SATA2?

SSD's and bragging rights. SATA3 doesn't reduce seek times or make the platters spin any faster! A 7200rpm disk isn't going transfer data any faster just because it's on a SATA3 interface.

Last I looked, few SAS drives will actually push 3Gbps.

 

[SushiEater]

malch wrote:

SushiEater wrote:

Most people don't write GB files. Most people write less than 5GB per day.

Regardless of the size workflow remains the same.

But SSD media wear is a complete non-issue.

Why are we going around and around? It maybe non-issue for you because the way you use it. But if SSD is used like a HD it will be a huge issue.

Lots of random access. Many database applications for example. Esp. if they're large and unlikely to be cached in RAM.

How many home users do you know who use large databases like that?

Do you know how many entrees would have to in the database to be that large to require SSD for random access? C'mon be serious.

Quite a lot. e.g. Lightroom users with large collections. They can have many GB of previews.

First of there are not that many LR users to begin with, many but not as many as Photoshop. Second, definitely not many with large collections. I even after few months archive my photos so they don't clutter my drives. And even if you have large collection you are not accessing all of the photos at once.

Also great for lots of multi-tasking. Take a look at what happens when two different applications are hammering the same disk; big slowdowns when drives start constantly seeking between two active areas of disk. Performance nosedives rather like a system that's out of RAM and thrashing the swap file. Try running (just as a simple example) an AV scan and a backup on the same disk at the same time. Seek, seek, seek, seek...

Maybe on the slow system like yours but not on mine. I ran Malwarebytes full scan and copy to external HD just an hour ago. No problem at all. Even on the slow system, how often do you do it?

Doesn't sound like you were hitting the same drive. If you were, there would be conflict.

Malwarebytes or any other software for that matter is not just reading the files. It opens the file, compares the content to the strings it has in its database, closes the file and so on. If you system is slow the search will be slow too. Again experimenting on my own for example, if I slow down my system to the minimum speed of 1200Mhz and try to convert NEF file in ACR the bar comes up " Reading NEF file" and it takes several times longer. If I am running at the full speed of 4500 MHz it takes 2 seconds and no bar. So yes, speed of the computer makes a big difference how your hard drive works.

And I have multiple apps hitting different folders with heavy physical i/o (not RAM cache) on the SSD every day. If I swapped my SSD back to a HDD, I'd hear continuous seeking. In fact, the silence of my first SSD was one of the Wow factors.

Silence? I can hardly hear my whole computer even though I have 2 WD Blacks in it. I had to slow down the fan on my video card to 20% because it was louder than anything else.

And tell me you don't want to eliminate all of that mechanical nonsense

Sure, if you give me 6TB SSD with longevity of 7 years for $200 I will go for it.

I'll be happy to do so as soon as I can buy one for $150.

BTW, this video is nothing. I have seen heads going so fast is like a blur.

30 years ago long seeks took something like 25ms. Most drives today are still over 5ms. Seek times haven't improved that much relative to other aspects of computer technology and nor are they likely too. This and rotational latency are why HDD's are doomed. They will be replaced by solid state technology. The only question is how quickly.

I can guarantee that it will not happen in your lifetime. The whole silicon process would have to be changed for that probably to a biological one.

That must be the custom motherboard I have never seen before.

BTW, why do you think all HD manufacturers using SATA3 interface if their hard drives can't even get to the speed of SATA2?

SSD's and bragging rights. SATA3 doesn't reduce seek times or make the platters spin any faster! A 7200rpm disk isn't going transfer data any faster just because it's on a SATA3 interface.

Last I looked, few SAS drives will actually push 3Gbps.

You still did not answer my question. Why slow drives have fast interface?

 

[kelpdiver]

SushiEater wrote:

| BTW, why do you think all HD manufacturers using SATA3 interface if their hard drives can't even get to the speed of SATA2?

SSD's and bragging rights. SATA3 doesn't reduce seek times or make the platters spin any faster! A 7200rpm disk isn't going transfer data any faster just because it's on a SATA3 interface.

Last I looked, few SAS drives will actually push 3Gbps.

You still did not answer my question. Why slow drives have fast interface?

as he said, marketing largely. There is a some gain from the cache hits that would benefit, but there's only 32 or 64mb of memory on these, so that may be difficult to measure.

 

[kelpdiver]

SushiEater wrote:

Your 150MB/s hard drive could write perhaps 12TB in a day.

Ha? At 150MB/sec it would take about 7 seconds to write 1TB.

No, in 7 seconds you could write 1GB. In 7000 seconds you could write 1TB. (or about 2 hours). This seems to be a pattern for you - misstating rates by several orders of magnitude. And may be why you refuse to accept the SSDs have the write endurance you need.

AnandTech Forums: Technology, Hardware, Software, and Deals

Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

www.anandtech.com

At 100GB/day (or 10x the 70gb/week you say is your norm), using a TLC drive which is one of the shortest lived forms at just over 1000 P/E cycles, on the smallest 120GB drive, still has a life expectancy of 4 years. Your use case translates to 40 years endurance. Get the 1tb model and it would be well over a century.

You've made it clear that no facts (not even your own) will convince you. That's your loss. I'm killfiling you now to save us all time.

But there is more. I also write video files that I can't really quantify.

large video files are about the lowest wearing - nearly no write amplification with large files.

After that I create folders and relocate all photos in to those folders while deleting many files. That is at least doubling the writing.

No, relocation files from folder to folder only involves rewriting the directory tables, not the file contents themselves. Only if you partitioned your stripe would you suffer this problem. The moral there is obvious - don't partition. Every device (or in your case striped volume) = one partition.

 

[SushiEater]

kelpdiver wrote:

SushiEater wrote:

Your 150MB/s hard drive could write perhaps 12TB in a day.

Ha? At 150MB/sec it would take about 7 seconds to write 1TB.

No, in 7 seconds you could write 1GB. In 7000 seconds you could write 1TB. (or about 2 hours). This seems to be a pattern for you - misstating rates by several orders of magnitude. And may be why you refuse to accept the SSDs have the write endurance you need.

I was only kidding to see if you catch it.

http://www.anandtech.com/show/7173/...w-120gb-250gb-500gb-750gb-1tb-models-tested/3

At 100GB/day (or 10x the 70gb/week you say is your norm), using a TLC drive which is one of the shortest lived forms at just over 1000 P/E cycles, on the smallest 120GB drive, still has a life expectancy of 4 years. Your use case translates to 40 years endurance. Get the 1tb model and it would be well over a century.

No, 100GB/week is just involuntary writing because of the scratch disk. If I was doing actual writing it would be a lot more.

You've made it clear that no facts (not even your own) will convince you. That's your loss. I'm killfiling you now to save us all time.

No such definition in the dictionary.

But there is more. I also write video files that I can't really quantify.

large video files are about the lowest wearing - nearly no write amplification with large files.

Not if you store and delete them after processing. Try doing it every week and see what happens.

All I know on my 120GB drive in 2 weeks of doing panos I lost one year. I don't know what kind of writing was happening. All I know is the final result.

 

[SushiEater]

kelpdiver wrote:

SushiEater wrote:

| BTW, why do you think all HD manufacturers using SATA3 interface if their hard drives can't even get to the speed of SATA2?

SSD's and bragging rights. SATA3 doesn't reduce seek times or make the platters spin any faster! A 7200rpm disk isn't going transfer data any faster just because it's on a SATA3 interface.

Last I looked, few SAS drives will actually push 3Gbps.

You still did not answer my question. Why slow drives have fast interface?

as he said, marketing largely. There is a some gain from the cache hits that would benefit, but there's only 32 or 64mb of memory on these, so that may be difficult to measure.

I did not ask about cache. I asked about SATA3 interface. At 150MB/sec drives don't need SATA3 and yet they have it.

 

[Tom_N]

SushiEater wrote:

kelpdiver wrote:

SushiEater wrote:

You still did not answer my question. Why slow drives have fast interface?

as he said, marketing largely. There is a some gain from the cache hits that would benefit, but there's only 32 or 64mb of memory on these, so that may be difficult to measure.

I did not ask about cache. I asked about SATA3 interface. At 150MB/sec drives don't need SATA3 and yet they have it.

His reply was relevant.

If the computer asks the hard drive for some data, and a copy resides in the drive's RAM cache, the drive can theoretically play that data back at full SATA3 speed. Of course, as he notes, HDD cache isn't big enough to keep a fast interface occupied this way for long.

Another poster mentioned marketing: "SATA3 HDD" sounds better than "SATA1 HDD" to customers who don't know any better.

 

[SushiEater]

Tom_N wrote:

SushiEater wrote:

kelpdiver wrote:

SushiEater wrote:

You still did not answer my question. Why slow drives have fast interface?

as he said, marketing largely. There is a some gain from the cache hits that would benefit, but there's only 32 or 64mb of memory on these, so that may be difficult to measure.

I did not ask about cache. I asked about SATA3 interface. At 150MB/sec drives don't need SATA3 and yet they have it.

His reply was relevant.

If the computer asks the hard drive for some data, and a copy resides in the drive's RAM cache, the drive can theoretically play that data back at full SATA3 speed. Of course, as he notes, HDD cache isn't big enough to keep a fast interface occupied this way for long.

Another poster mentioned marketing: "SATA3 HDD" sounds better than "SATA1 HDD" to customers who don't know any better.

Better answer: Multiple drives in RAID0 work better on SATA3.

 

[theswede #100093]

I was only kidding to see if you catch it.

Right.

All I know on my 120GB drive in 2 weeks of doing panos I lost one year. I don't know what kind of writing was happening. All I know is the final result.

MTBF predicts this. You've been lucky with HDD's and had bad luck with an SSD. Statistics support that happening. It says nothing at all about either HDD longevity or SSD longevity; it takes a lot more datapoints than that to draw conclusions.

So you're right. That is all you know. And it tells you nothing about how a properly managed SSD will last compared to a properly managed HDD.

Jesper

 

[SushiEater]

You keep skewing this argument towards SSD benefit providing hearsay reports that SSD can last for a very long time completely ignoring the premiss of size/cost/reliability/convinience.

Even if I assume (but I have not and most likely will not) that SSD lasts for a very long time enduring many large writes it still would be outrageously stupid to get 3 or more terabytes (as I find out I probably need more now) of SSD which I would still have to put in RAID 0 for convenience at the cost of at least $2400 for 4TB while I can buy 2 WD 2TB Blacks in RAID 0 for $300 plus getting 5 year warranty vs 3 years for SSD.

Sorry but RAID 0 wins all the time in my book.