Website, CSF, and lots of attacks.

[Spork Schivago]

Re: Website, CSF, and lots of attacks.

The data is scattered about the platter/s if it can seek it faster the IOPS increases.

Imagine a conveyor belt in an Amazon warehouse with two order pickers. One is a little faster on his feet than the other, the parcels still move along the conveyor at the same speed but there are more parcels with the faster picker.

Gotcha, so I was right, then. The maximum transfer rate wouldn't change. Just with faster seek rate, you'd get the data faster because it'd be able to start sending it faster then maybe another drive with a slower seek rate.

There's access time as well. So, does the speed of the drive actually matter? Or in the end, is it just access time and seek time that really matter? Are there any 10,000 RPM drives that don't perform as well as a 7,200 RPM drive with faster seek and access times?

[mariushm]

Re: Website, CSF, and lots of attacks.

Regardless of rpm, any SSD kills classical hard drives, except those on SAS 12 gbps maybe (and even those are killed by some more high end SSD drives, like the pci-e based ones, r NVMe drives with pci-e x2 or x4 bandwidth).

The 15k drives are basically a compromise between capacity (600 GB) and access times, but the price of SSD drives and flash memory is going down fast that they're not worth it anymore.

Have a look at specs for 15k drives: http://www.seagate.com/gb/en/interna...15k-hdd/#specs

530 gigabit per square inch density, and they peak at 246 MB/s and up to 2400 MB/s bursts (reading from cache, stuff like that)

There's drives with higher data density out there and drives which can sustain transfer speeds close enough to those of 15k drives, so the only benefit they have is lower access times
For example here's the Seagate 10TB Helium drive : http://www.seagate.com/gb/en/enterpr...dd-10tb/#specs

They have sustained rate of 237/249 MB/s, close enough to the 15k drives ..

Makes no sense to pay close to 400$ for 300 GB 15k 12gbit drive when i could pay 750$ for 10 TB drive.

Or you could pay $250 for something like Plextor M8Pe Series 512GB which reads at 2300 MB/s and writes at up to 1300 MB/s

[stj]

Re: Website, CSF, and lots of attacks.

the other thing is data retention - i'v seen some pretty fucked up numbers on the entrophy of the cells in some MLC flash drives.

hard-drives, hold the data for years.

[keeney123]

Re: Website, CSF, and lots of attacks.

the other thing is data retention - i'v seen some pretty fucked up numbers on the entrophy of the cells in some MLC flash drives.

hard-drives, hold the data for years.

Which make hard drives good for storage or a clone drive when the SSD fails.

[Spork Schivago]

Re: Website, CSF, and lots of attacks.

Regardless of rpm, any SSD kills classical hard drives, except those on SAS 12 gbps maybe (and even those are killed by some more high end SSD drives, like the pci-e based ones, r NVMe drives with pci-e x2 or x4 bandwidth).

The 15k drives are basically a compromise between capacity (600 GB) and access times, but the price of SSD drives and flash memory is going down fast that they're not worth it anymore.

Have a look at specs for 15k drives: http://www.seagate.com/gb/en/interna...15k-hdd/#specs

530 gigabit per square inch density, and they peak at 246 MB/s and up to 2400 MB/s bursts (reading from cache, stuff like that)

There's drives with higher data density out there and drives which can sustain transfer speeds close enough to those of 15k drives, so the only benefit they have is lower access times
For example here's the Seagate 10TB Helium drive : http://www.seagate.com/gb/en/enterpr...dd-10tb/#specs

They have sustained rate of 237/249 MB/s, close enough to the 15k drives ..

Makes no sense to pay close to 400$ for 300 GB 15k 12gbit drive when i could pay 750$ for 10 TB drive.

Or you could pay $250 for something like Plextor M8Pe Series 512GB which reads at 2300 MB/s and writes at up to 1300 MB/s

Right, the question wasn't so much about what drives are better or faster. I'm just trying to wrap my head around the the RPMs. How that equates to faster transfer rates. I would think access time and seek time for the older style hard drives would be what mattered and the speed of the drive (in RPMs) wouldn't really matter at all. Just because it's 10,000 RPM, if it takes longer to find the data and access the data then a 7,200 RPM drive, wouldn't the 7,200 RPM drive be faster, even if it spins a bit slower?

[Spork Schivago]

Re: Website, CSF, and lots of attacks.

the other thing is data retention - i'v seen some pretty fucked up numbers on the entrophy of the cells in some MLC flash drives.

hard-drives, hold the data for years.

Is this still a problem? I too saw the same issues but the way data is spread across the cells now and the better flash types, is data retention still an issue?

I mean, ultimately, you need redundancy (backups, RAID, both, etc) if the data's important.

I had a 2GB unit I bought, brand new, never opened, but it was maybe 10 - 15 years old. It was in high storage lockup at Staples. High storage is for high ticket items. Item's likely to get stolen. Only managers have keys and it was just sitting on a shelf for years. Never touched. I bought it. It was like 2$. Originally, it was like 600$.

I plugged it into a PC. It was IDE. The PC started, initialized the disc. I threw a Linux live disc in and was formatting the drive. About half way through the format, click click click. Drive died. Moving parts will always fail, sooner or later. I'm guessing these parts failed from age, but I guess someone could have dropped it during shipping or something.


I found a thumb drive that was 256MB in size in my house. It was when they first came out. It had a little display on it that showed how full it was. I didn't think the thumb drive would work. It must have been at least 10 years old, if not older. Back then, 256MB thumb drives were huge. Probably costed me close to 100$ back in the day. The bar that showed how much storage was being used, it was still active and displaying. I was like how the heck is this possible? We don't have batteries that last that long! Anyway, I plugged it into the PC and sure enough, data was perfectly fine and it brought back some good memories. I had some old C source codes on it from when I was majoring in Computer Science back in 1999 ~ 2000.

[stj]

Re: Website, CSF, and lots of attacks.

the flash problem is getting worse,
they keep shrinking the cells to get higher capacity.
as you shrink the cells, you reduce the residual charge.

very old flash chips could retain for 10years+
relativly old - 5years+

the very latest huge capacity chips with internal error remapping etc are rated in months sometimes - you need an o.s. that will refresh the cells with a filescan ocasionally, or an SSD firmware that will enforce a refresh.

BUT that wont help if it goes in storage or is simply not used for whatever reason for a relativly long time.

[mariushm]

Re: Website, CSF, and lots of attacks.

Right, the question wasn't so much about what drives are better or faster. I'm just trying to wrap my head around the the RPMs. How that equates to faster transfer rates. I would think access time and seek time for the older style hard drives would be what mattered and the speed of the drive (in RPMs) wouldn't really matter at all. Just because it's 10,000 RPM, if it takes longer to find the data and access the data then a 7,200 RPM drive, wouldn't the 7,200 RPM drive be faster, even if it spins a bit slower?

From what I can remember, 15k hard drives were used by companies in RAID6 or similar configurations, along with "short stroking", a technique where a partition smaller than the total capacity of the drive is created in order to make the hard drive move the heads much less, so faster access times to particular data tracks.
This was great for databases, for lots of I/O per second, to access random rows in tables of the database very fast. You bought 75 GB drives and made 20-30 GB partitions and had a 100 GB drive in a RAID6 or something like that, enough for databases.

WD had 10k raptor drives in 3.5" format and then moved to 2.5" drives in 3.5" shell (heatsink) pretty much in order to simulate this short stroking technique, by using smaller platters you get smaller access times, besides the other benefit of less platter expansion, fewer recalibrations etc etc

The technique was useful for reading tracks fast once the heads positioned themselves over the track.

Advances in head and platter technology made high speeds less important. Think giant magnetoresistance (GMR) , perpendicular recording , shingled magnetic recording (SMR) all these contributed to increasing area density so basically the head reads a lot more bits at a slower rpm compared to older drives.

Then you also have movement to Helium which basically reduces the heat produced by friction with regular air and vibrations are reduced as well, so platters may not expand and contract as much, and the platters can be made thinner.
It's also rumored (i'm not sure it's true) that due to less expansion and contraction, the platters can be made with fewer servo tracks, fewer calibration tracks, so more tracks per inch can be squeezed on platters.

See text, comments and pictures in this article on Anandtech if you're curious to learn more about Helium usage in drives: http://www.anandtech.com/Show/Index/...ultrastar-he10

[stj]

Re: Website, CSF, and lots of attacks.

Helium permiates most materials slowly - inc rubber.
i bet those drives dont contain helium for very long!

[Spork Schivago]

Re: Website, CSF, and lots of attacks.

the flash problem is getting worse,
they keep shrinking the cells to get higher capacity.
as you shrink the cells, you reduce the residual charge.

very old flash chips could retain for 10years+
relativly old - 5years+

the very latest huge capacity chips with internal error remapping etc are rated in months sometimes - you need an o.s. that will refresh the cells with a filescan ocasionally, or an SSD firmware that will enforce a refresh.

BUT that wont help if it goes in storage or is simply not used for whatever reason for a relativly long time.

Wow, I had no idea. I thought the technology was getting better. What about those M.2 drives? They're solid state, right? Same issue?

[Spork Schivago]

Re: Website, CSF, and lots of attacks.

From what I can remember, 15k hard drives were used by companies in RAID6 or similar configurations, along with "short stroking", a technique where a partition smaller than the total capacity of the drive is created in order to make the hard drive move the heads much less, so faster access times to particular data tracks.
This was great for databases, for lots of I/O per second, to access random rows in tables of the database very fast. You bought 75 GB drives and made 20-30 GB partitions and had a 100 GB drive in a RAID6 or something like that, enough for databases.

WD had 10k raptor drives in 3.5" format and then moved to 2.5" drives in 3.5" shell (heatsink) pretty much in order to simulate this short stroking technique, by using smaller platters you get smaller access times, besides the other benefit of less platter expansion, fewer recalibrations etc etc

The technique was useful for reading tracks fast once the heads positioned themselves over the track.

Advances in head and platter technology made high speeds less important. Think giant magnetoresistance (GMR) , perpendicular recording , shingled magnetic recording (SMR) all these contributed to increasing area density so basically the head reads a lot more bits at a slower rpm compared to older drives.

Then you also have movement to Helium which basically reduces the heat produced by friction with regular air and vibrations are reduced as well, so platters may not expand and contract as much, and the platters can be made thinner.
It's also rumored (i'm not sure it's true) that due to less expansion and contraction, the platters can be made with fewer servo tracks, fewer calibration tracks, so more tracks per inch can be squeezed on platters.

See text, comments and pictures in this article on Anandtech if you're curious to learn more about Helium usage in drives: http://www.anandtech.com/Show/Index/...ultrastar-he10

Thanks Mariushm. That essentially answers my question then. Just because the drive spins faster doesn't necessarily make it a faster drive. Thanks! I wasn't expecting just a detailed response. You guys seem to really know your hard drives!

[Spork Schivago]

Re: Website, CSF, and lots of attacks.

Helium permiates most materials slowly - inc rubber.
i bet those drives dont contain helium for very long!

You mean they can't keep the helium in because it just slowly leaks out through the material? If they're using helium to replace the air, wouldn't they make sure to use a material it couldn't permeate? Kinda like those double or triple pane windows. They have some gas inside and it's not supposed to leak out. We have one in the master bedroom where it leaked out.

These must be cheap windows. We can't find any make or model information anywhere, on any of them. There's some stickers but just warnings and stuff. If you carefully look at the metalish looking strip between the panes, you can read 2 MI. I don't think they make the windows though. I tried contacting them and never got a response. We need new screens but these aren't normal screens. They're weird. I guess eventually, we'll replace the windows with those low-e's. The window things way off topic though. Sorry {: - )

[stj]

Re: Website, CSF, and lots of attacks.

maybe this explains your Telnet probes.
http://www.theregister.co.uk/2016/10/03/iot_botnet/

[Spork Schivago]

Re: Website, CSF, and lots of attacks.

maybe this explains your Telnet probes.
http://www.theregister.co.uk/2016/10/03/iot_botnet/

Ahh, that would make sense, wouldn't it?