Help deciding RAID5 v RAID1 - archival purposes

[nanofrog]

the fact that intel and intel alone will be the ones marketing and selling the product gives me reason enough to believe that it will be quite expensive to begin with. it may even be that way for a very long time unfortunately.

That's part of it. The parts would have to be made on a massive scale to have lower production costs (cheap enough to be considered in the first place).

The other side of the equation, is what real world benefits does it provide?
That will be the second question, assuming the cost is low enough to use it in the first place. If it's low, it will show up rather quickly to give an edge over the competition (basically a spec war).

10G on the MP is long overdue, but as you say very expensive to be utilised by the hardware.

I don't see it that way (overdue). Not for MP's anyway, as they seem to be more aimed at individuals and SOHO/SMB's, not the enterprise world. They don't have a presence there. Not that they wouldn't like to have it, and a few could use it. But the cost of the switches are nutz for such users. One switch is more expensive than a single system. For a few large graphics design houses, it would be justafiable financially, but that's a very small niche within a niche from what I can tell.

tried to understand what you mean here but there are so many negatives its not possible at this time of day! are you saying 1G devices are considered to be expensive? our house is running 1G ethernet and it wasnt all that expensive, around $200 including the cables and an enterprise-ish grade switch. but then again that doesnt include the computers that is capable of running at that speed

I'm used to seeing 1G switches still in the $1200+USD range (some of the more expensive have 10G support or FCoE support). Certainly not home units, but I've seen users complain about those too.

And for most purposes, it's only being used to share an ISP connection anyway, which is can't even saturate 100M.

 

[DoFoT9]

That's part of it. The parts would have to be made on a massive scale to have lower production costs (cheap enough to be considered in the first place).

and then of course they would want to make their money back + some more, so the cheap production costs arent so cheap once you get to the consumer side. i dont know much about business but a technology such as this cant be cheap to plan, develop, test and then sell.

The other side of the equation, is what real world benefits does it provide?

well i guess there are positives and negatives here, of course. im not sure if LP+USB is a plus or not

That will be the second question, assuming the cost is low enough to use it in the first place. If it's low, it will show up rather quickly to give an edge over the competition (basically a spec war).

yup another war LP+USB would be nice i guess. firewire would really be showing its age if that happens.

I don't see it that way (overdue). Not for MP's anyway, as they seem to be more aimed at individuals and SOHO/SMB's, not the enterprise world. They don't have a presence there. Not that they wouldn't like to have it, and a few could use it. But the cost of the switches are nutz for such users. One switch is more expensive than a single system. For a few large graphics design houses, it would be justafiable financially, but that's a very small niche within a niche from what I can tell.

true, not many people would utilise it in a MP. especially not in an internet situation, intranets are a different story.

I'm used to seeing 1G switches still in the $1200+USD range (some of the more expensive have 10G support or FCoE support). Certainly not home units, but I've seen users complain about those too.

well my switch is only an 8port cisco thing, nothing that special - especially when compared to the enterprise things that can be purchased.

And for most purposes, it's only being used to share an ISP connection anyway, which is can't even saturate 100M.

100M? i dont even think there is such a thing in australia especially not below the $1,000 (per month)mark.

 

[nanofrog]

and then of course they would want to make their money back + some more, so the cheap production costs arent so cheap once you get to the consumer side. i dont know much about business but a technology such as this cant be cheap to plan, develop, test and then sell.

If Intel gets the production costs right, and board and device makers purchase the parts, Intel will make money. It's what happens after that which poses the greater uncertainty. They might have a hard time selling it, as there's not much that can truly use it. If it's cheap enough to just replace an existing interface bus chip with, it would be out there, but not really used.

But my guess is, the cost will be too high for that. Accountants watch fractions of pennies afterall (OK, cents if it makes more sense to you ).

100M? i dont even think there is such a thing in australia especially not below the $1,000 (per month)mark.

I was talking about the ISP speeds are so low, that at best (i.e. 80Mb/s speeds in Japan), a 100Mb/s spec Ethernet port will work, and still not be saturated.

 

[DoFoT9]

If Intel gets the production costs right, and board and device makers purchase the parts, Intel will make money. It's what happens after that which poses the greater uncertainty. They might have a hard time selling it, as there's not much that can truly use it. If it's cheap enough to just replace an existing interface bus chip with, it would be out there, but not really used.

intel has big plans though, they want it to be used with HD video cameras and everyday photo cameras and all that - could you imagine how much it would be used then!? especially if sony, panasonic, jvc etc caught on. what else would use it... ummm.. peripherals i guess lol. HDDs, scanners, printers etc. thats probably a harder market to 'convince'. imagine if there were no needed drivers - now that would be cool.

But my guess is, the cost will be too high for that. Accountants watch fractions of pennies afterall (OK, cents if it makes more sense to you ).

ahh makes much more cen....i mean...sense

I was talking about the ISP speeds are so low, that at best (i.e. 80Mb/s speeds in Japan), a 100Mb/s spec Ethernet port will work, and still not be saturated.

only 80Mb/s in japan? i would have expected more. residential lines do not go above 20Mb/s here though so i guess i cant really talk much. this isnt including business line speeds of course, but they dont go very high either.

 

[nanofrog]

intel has big plans though, they want it to be used with HD video cameras and everyday photo cameras and all that - could you imagine how much it would be used then!? especially if sony, panasonic, jvc etc caught on. what else would use it... ummm.. peripherals i guess lol. HDDs, scanners, printers etc. thats probably a harder market to 'convince'. imagine if there were no needed drivers - now that would be cool.

Understandable, but can such devices saturate existing connection methods available?

Light Peak's far faster than such devices are capable of, unless those types of devices take a major leap in throughput methodologies (i.e. mulit channel controllers used to parallel the data channels for increased bandwidth, such as SSD's do with Flash chips). I could see it benefiting HD video camera's soon enough (given the massive data they have to store), but not printers and scanners any time soon.

only 80Mb/s in japan? i would have expected more. residential lines do not go above 20Mb/s here though so i guess i cant really talk much. this isnt including business line speeds of course, but they dont go very high either.

IIRC, there is a 120Mb/s service as well, but I don't know if it's still residential or part the commercial plan strata, as it would be in the US, and I presume in many other nations.

Much higher speeds are possible in places like the US, but it's not affordable for home users, if available at all (hint: optical lines directly to the structure, so it usually is only available in commercial property zones).

 

[DoFoT9]

Understandable, but can such devices saturate existing connection methods available?

well no. but combine but a few of these devices and im sure it could. *think multiple video streams from cameras* maybe it could be a new advancement for this typf of technology.

Light Peak's far faster than such devices are capable of, unless those types of devices take a major leap in throughput methodologies (i.e. mulit channel controllers used to parallel the data channels for increased bandwidth, such as SSD's do with Flash chips). I could see it benefiting HD video camera's soon enough (given the massive data they have to store), but not printers and scanners any time soon.

yes for cameras it would be very very nice! for printers/scanners god no. useless - unless of course they introduce something completely innovative. maybe printers that chug along all day (thousands of thousands of pages resulting in multitudes of bandwidth requirements) could benefit?

IIRC, there is a 120Mb/s service as well, but I don't know if it's still residential or part the commercial plan strata, as it would be in the US, and I presume in many other nations.

our government is implementing FTTH - god knows how long it will take though. 5 years i guess? it will be amazingly more expensive though if only.

Much higher speeds are possible in places like the US, but it's not affordable for home users, if available at all (hint: optical lines directly to the structure, so it usually is only available in commercial property zones).

so advanced compared to us. but in the scheme of things its like comparing processor speeds. i wonder how long we will always be struggling to keep up with everything for. (very passionate about it incase you didnt realise haha). grr

 

[nanofrog]

well no. but combine but a few of these devices and im sure it could. *think multiple video streams from cameras* maybe it could be a new advancement for this typf of technology.

The fact it can daisy chain is rather nice, and that would be it's real selling point. Fewer differing interfaces required, as multiple (if not all) of a user's peripheral devices can be connected to a single port. But like FW, there's going to be a cost in switching overhead. No way around that, as the optical signal has to be converted back to electrical. Particularly in the beginning, as the spec is 10Gb/s (1.25GB/s). That translates to 3.333 SATA ports at full bandwidth per (not that a current drive can sustain that, but it's not concievably that far off). Rather nice, but not the end-all of interface ports just yet.

for printers/scanners god no. useless - unless of course they introduce something completely innovative. maybe printers that chug along all day (thousands of thousands of pages resulting in multitudes of bandwidth requirements) could benefit?

Wouldn't help, unless the imaging methodologies are improved in terms of speed. Currently, the documents not yet printed are stored in memory, which is typically 100/133MHz DDR SODIMM's. Not blazingly fast, as the imaging systems aren't fast enough to require higher bandwidths.

so advanced compared to us. but in the scheme of things its like comparing processor speeds. i wonder how long we will always be struggling to keep up with everything for. (very passionate about it incase you didnt realise haha). grr

Well, faster speeds are certainly available, as large corporations need it in their data centers. It's just prohibitive in terms of cost for an individual, just as it is in many countries (though the specifics vary).

If you've the cash, and in the right location, you could get optical to your building. But be prepared to make a mortgage sized or more monthy payment to your ISP.

 

[DoFoT9]

The fact it can daisy chain is rather nice, and that would be it's real selling point. Fewer differing interfaces required, as multiple (if not all) of a user's peripheral devices can be connected to a single port. But like FW, there's going to be a cost in switching overhead. No way around that, as the optical signal has to be converted back to electrical. Particularly in the beginning, as the spec is 10Gb/s (1.25GB/s). That translates to 3.333 SATA ports at full bandwidth per (not that a current drive can sustain that, but it's not concievably that far off). Rather nice, but not the end-all of interface ports just yet.

one would think that USB + LP would help the selling point substantially. half the job is done already. if backwards compatibility was as standard then it sure would look good.

If you've the cash, and in the right location, you could get optical to your building. But be prepared to make a mortgage sized or more monthy payment to your ISP.

we cant australia sucks for internet! 10% of the country can only get cable! not that i really like the concepts of cable. ADSL it is im afraid, expensive and slow.

 

[nanofrog]

one would think that USB + LP would help the selling point substantially. half the job is done already. if backwards compatibility was as standard then it sure would look good.

It might happen in a round-about-way, assuming there's a chip made to allow it. Intel's in negotiations with chip makers over that one.

Currently though, Light Peak is a pair of really fine optical cables, and copper jackets (which appears will end up being used as a means of powering connected devices).

we cant australia sucks for internet! 10% of the country can only get cable! not that i really like the concepts of cable. ADSL it is im afraid, expensive and slow.

That's residential though, and it's the same here for the most part. If you happen to live in the right location, and have lots of $$$, you might be able to get an optical signal to the structure for higher bandwidth. But the Cable and ADSL is off a backbone, and it's a faster pipe. That means optical (SONET for example). So it's available, but insanely expensive, and restricted to certain areas.

But there is optical service in Australia (though it's aimed at academics).

As per the residential end-user (or some SOHO users), I don't care for Cable either. I don't like the idea of a shared connection, as the throughput is variable according to network load. Yuck.

 

[DoFoT9]

It might happen in a round-about-way, assuming there's a chip made to allow it. Intel's in negotiations with chip makers over that one.

all very technical i imagine

Currently though, Light Peak is a pair of really fine optical cables, and copper jackets (which appears will end up being used as a means of powering connected devices).

i wonder how much power LP can push through. USB and FW cant really do anything really, a few watts - great

That's residential though, and it's the same here for the most part. If you happen to live in the right location, and have lots of $$$, you might be able to get an optical signal to the structure for higher bandwidth. But the Cable and ADSL is off a backbone, and it's a faster pipe. That means optical (SONET for example). So it's available, but insanely expensive, and restricted to certain areas.

But there is optical service in Australia (though it's aimed at academics).

we have something like that at uni, they offer the "eduroam" service that we have at uni (allows users from a number of universities to have internet access when they travel to a different uni). but yea most definetely for a refined target market unfortunately

As per the residential end-user (or some SOHO users), I don't care for Cable either. I don't like the idea of a shared connection, as the throughput is variable according to network load. Yuck.

i hear that the cable TV is included in that network load amount as well!?!?!?! that is just rediculous, especially when people start to stream HD videos with their devices, watch normal cable TV, download torrents etcetc. it would just slow down too much. silly concept.

 

[DoFoT9]

It might happen in a round-about-way, assuming there's a chip made to allow it. Intel's in negotiations with chip makers over that one.

all very technical i imagine

Currently though, Light Peak is a pair of really fine optical cables, and copper jackets (which appears will end up being used as a means of powering connected devices).

i wonder how much power LP can push through. USB and FW cant really do anything really, a few watts - great

That's residential though, and it's the same here for the most part. If you happen to live in the right location, and have lots of $$$, you might be able to get an optical signal to the structure for higher bandwidth. But the Cable and ADSL is off a backbone, and it's a faster pipe. That means optical (SONET for example). So it's available, but insanely expensive, and restricted to certain areas.

But there is optical service in Australia (though it's aimed at academics).

we have something like that at uni, they offer the "eduroam" service that we have at uni (allows users from a number of universities to have internet access when they travel to a different uni). but yea most definetely for a refined target market unfortunately

As per the residential end-user (or some SOHO users), I don't care for Cable either. I don't like the idea of a shared connection, as the throughput is variable according to network load. Yuck.

i hear that the cable TV is included in that network load amount as well!?!?!?! that is just rediculous, especially when people start to stream HD videos with their devices, watch normal cable TV, download torrents etcetc. it would just slow down too much. silly concept.

 

[nanofrog]

i hear that the cable TV is included in that network load amount as well!?!?!?! that is just rediculous, especially when people start to stream HD videos with their devices, watch normal cable TV, download torrents etcetc. it would just slow down too much. silly concept.

The cable content is on what the cable company ran wire wise. It's not also dropping on the backbone (i.e. going upstream) they used to obtain the internet access that they use to share amongst it's subscribers (cable content is downstream only). So their own infrastructure should actually run with more bandwidth than thier ISP trunks they have.

Think of a T junction with plumbing.
Input 1 = Internet access
Input 2 = Cable content from satellite sources
Output = 1 + 2 (the combined output requires more bandwidth)

That said, to stream movies,... more bandwidth will be needed. What you get as PPV on cable (satellite too), is compressed data. That's how they've made it work so far, as there's not enough to send 1080p signals in uncompressed form (if they even offer 1080p). And if it's going to come from another source (content NOT provided by the cable company), it's only going to have access to the ISP band you've signed up for at best (assuming the traffic isn't high enough to cause your band to dip).

As for me, my HDTV signals are satellite, and it's 1080i content. No 1080p service yet. And it comes in as MPEG-4 compressed data. The box then has to uncompress it for display. DVR is nice though.

 

[DoFoT9]

well, its about 2 months after i started this thread lol and ive decided to decide on a few components (if that makes sense).

ive deciced to go with an external dual bay caddy - in RAID1 form. it looks really solid and meets my requirements. i need the enclosure to have 2xfw800 ports so that i can continue to daisy chain its going to be about $150Aus delivered.

i am then looking at 2x2TB Hitachi DeskstarTM 7K2000 drives, for $216Aus each (cheapest consumer ones out there atm).

i just have a concern related to drive failures - im planning on using the enclosures "built in" RAID features. my concerns lie with compatibility functions between the computer and the enclosure. e.g. if i want to take the hdd out of the enclosure and put it into a computer - would that be possible? im hoping so in a RAID1 situation.

anyway if all goes to plan i will hopefully be getting this within the next month or so (fingers crossed).

 

[nanofrog]

ive deciced to go with an external dual bay caddy - in RAID1 form. it looks really solid and meets my requirements. i need the enclosure to have 2xfw800 ports so that i can continue to daisy chain its going to be about $150Aus delivered.

i am then looking at 2x2TB Hitachi DeskstarTM 7K2000 drives, for $216Aus each (cheapest consumer ones out there atm).

In it's most basic sense, it will work. Keep in mind, daisy chaining them will give you access to more drives, but the overall throughput won't exceed what FW800 can do, so additional members won't improve things for you.

You'd be better off IMO, of going with an eSATA card, and enclosures with eSATA (you can do a RAID 1 via Disk Utility). It should also be cheaper for the card + enclosure/s than what you'd spend on the linked solution (say ~$100USD or so). Drive cost would remain the same of course.

i just have a concern related to drive failures - im planning on using the enclosures "built in" RAID features. my concerns lie with compatibility functions between the computer and the enclosure. e.g. if i want to take the hdd out of the enclosure and put it into a computer - would that be possible? im hoping so in a RAID1 situation.

anyway if all goes to plan i will hopefully be getting this within the next month or so (fingers crossed).

If it's done by it's own drivers (I checked Hotway for drivers, and didn't find anything), you may not be able to transfer it, as it's likely the software implementation is different than that of OS X.

But if you create it via OS X's software RAID functions, it would be able to be placed internally in the system.

 

[DoFoT9]

In it's most basic sense, it will work. Keep in mind, daisy chaining them will give you access to more drives, but the overall throughput won't exceed what FW800 can do, so additional members won't improve things for you.

im not bothered by that at all the FW800 drives are at the front and hit about 50MB/s. so i assume this new addition will hit those speeds when copying from methods that dont require other drives on the chain. im perfectly fine with that, its only a movie storage/backup drive etc.

You'd be better off IMO, of going with an eSATA card, and enclosures with eSATA (you can do a RAID 1 via Disk Utility). It should also be cheaper for the card + enclosure/s than what you'd spend on the linked solution (say ~$100USD or so). Drive cost would remain the same of course.

but there is no eSata on the iMac (where i have the daisy chain). $100USD is about the same price as the enclosure that i linked to before as well (excluding postage).

If it's done by it's own drivers (I checked Hotway for drivers, and didn't find anything), you may not be able to transfer it, as it's likely the software implementation is different than that of OS X.

But if you create it via OS X's software RAID functions, it would be able to be placed internally in the system.

ok so if i do go with case (seems quite likely at this point in time), i will do a software RAID just to be safe i take it that the slowdowns of software RAID arent THAT significant (a few % maybe?, and a bit extra CPU usage?).

 

[nanofrog]

but there is no eSata on the iMac (where i have the daisy chain). $100USD is about the same price as the enclosure that i linked to before as well (excluding postage).

I was talking about using an eSATA card, but now I recall you have an iMac, which still makes it impossible. So nevermind.

If you're wanting it for the Hackintosh, it will.

 

[DoFoT9]

I was talking about using an eSATA card, but now I recall you have an iMac, which still makes it impossible. So nevermind.

If you're wanting it for the Hackintosh, it will.

yea i could always put it in the hackintosh - but then i will need to keep it powered on 24/7. power bills have just gone up 30% and will go up another 50% this year (power went private).

using an external enclosure will use SO much less power, my hack eats about 200W @ idle (it has a few drives in it etc). 200W vs ~10W of enclosure? i know what one i will pick

 

[nanofrog]

yea i could always put it in the hackintosh - but then i will need to keep it powered on 24/7. power bills have just gone up 30% and will go up another 50% this year (power went private).

using an external enclosure will use SO much less power, my hack eats about 200W @ idle (it has a few drives in it etc). 200W vs ~10W of enclosure? i know what one i will pick

What about using the Sleep Function?

That should be able to address your power concerns, as both the Hackintosh and iMac consume some power in Sleep mode. Active, there could well be a substantial difference (especially if you've a fair number of add-ons).

 

[DoFoT9]

What about using the Sleep Function?

That should be able to address your power concerns, as both the Hackintosh and iMac consume some power in Sleep mode. Active, there could well be a substantial difference (especially if you've a fair number of add-ons).

hackintosh doesnt have sleep functionality unfortunately. never got around to fixing it.

its only running 10.5.4, id love to upgrade it to SL which i may do sooner or later.

but even still, it would still require an extra computer to be turned on haha.

nano you make things so hard to decide!

 

[nanofrog]

hackintosh doesnt have sleep functionality unfortunately. never got around to fixing it.

its only running 10.5.4, id love to upgrade it to SL which i may do sooner or later.

but even still, it would still require an extra computer to be turned on haha.

nano you make things so hard to decide!

You could get an enclosure/s (you can use singles if you have to) that have FW, USB and eSATA. Run them on the iMac, and if you ever get the sleep function fixed on the Hackintosh, you can get a cheap eSATA card, and move them to that machine (you can always pass data over 1G Ethernet between the systems if you want).

Just to make things that much harder...

 

[DoFoT9]

You could get an enclosure/s (you can use singles if you have to) that have FW, USB and eSATA. Run them on the iMac, and if you ever get the sleep function fixed on the Hackintosh, you can get a cheap eSATA card, and move them to that machine (you can always pass data over 1G Ethernet between the systems if you want).

Just to make things that much harder...

fine mr. smarty pants

how about this? i have three (one dual enclosure doesnt have a working PSU) of these products already (one single enclosure and two dual enclosure). they are brilliant! ive been trying to find a new PSU for the broken one but i never have!

anyway, maybe that one then?

file sharing will mostly be through the network, e.g. movie streaming to PS3, other iMac in house, MBP etcetc.

at $330Aus + $12Aus postage, its a tiny bit pricey.

just searched the original site that i posted up before.

hotway 4 drive RAID enclosure 1xusb + 1xesata @$185Aus

hotway 2 drive RAID enclosure 1xusb + 1xesata @95Aus

how can eSata be added onto the imac? (if possible)

hmmm

 

[nanofrog]

fine mr. smarty pants

how about this? i have three (one dual enclosure doesnt have a working PSU) of these products already (one single enclosure and two dual enclosure). they are brilliant! ive been trying to find a new PSU for the broken one but i never have!

The Oxford 924 chip (PM chip) in it makes me nervous, as they've caused issues before with eSATA cards.

You could try one you have that works out, and see what happens. I don't know if they use the Oxford 924, but it's likely in the duals (no PM chip in the single).

hotway 4 drive RAID enclosure 1xusb + 1xesata @$185Aus

hotway 2 drive RAID enclosure 1xusb + 1xesata @95Aus

They should work, but I don't know what chip is in it.

how can eSata be added onto the imac? (if possible)

hmmm

The only way to do it, is find a bridge device, such as what you linked. There's no way to get eSATA to the iMac directly. It has to go through either USB or FW.

 

[DoFoT9]

The Oxford 924 chip (PM chip) in it makes me nervous, as they've caused issues before with eSATA cards.

You could try one you have that works out, and see what happens. I don't know if they use the Oxford 924, but it's likely in the duals (no PM chip in the single).

none of the ones i have are eSata so cant exactly test.

what sort of issues are they? minor or major?

They should work, but I don't know what chip is in it.

hmm ok. probably a cheaper chip no doubt.

The only way to do it, is find a bridge device, such as what you linked. There's no way to get eSATA to the iMac directly. It has to go through either USB or FW.

through FW doesnt bother me at all. speed isnt really an issue, i guess i should just go with a FW800 drive.

could you take a look at the other two devices that i found? the dual bay eSata, and the quad bay eSata. i would really like the quad for future expandability! but its eSata and USB2.0 only eSata ->fw800 would be fine by me, provided that it works. ill try find some!

EDIT: found a 4-bay one with eSata, fw400, fw800, usb etc for $319Aus - it uses the OXUFS936QSE chipset. any idea what that chip is? low end? link

 

[nanofrog]

what sort of issues are they? minor or major?

Instability IIRC, so it's major.

hmm ok. probably a cheaper chip no doubt.

Oxford is one of, if not the cheapest part available. They were bought out by PLX in Dec. 2008/Jan. 2009 (924 is an original Oxford design, and it's one of the oldest available - it may be the oldest PM chip they sell now). The 936 is newer, and I've not seen any problems listed with it.

EDIT: found a 4-bay one with eSata, fw400, fw800, usb etc for $319Aus - it uses the OXUFS936QSE chipset. any idea what that chip is? low end? link

It looks like it will work, and I've not seen or heard of any problems with the Oxford 936 chip, so it appears it's the best fit for your needs here.

 

[DoFoT9]

Instability IIRC, so it's major.

blast ok looks like i wont go with the Taurus then. even though i really like the look of them.

Oxford is one of, if not the cheapest part available. They were bought out by PLX in Dec. 2008/Jan. 2009 (924 is an original Oxford design, and it's one of the oldest available - it may be the oldest PM chip they sell now). The 936 is newer, and I've not seen any problems listed with it.

hmm. ok then good to know.

It looks like it will work, and I've not seen or heard of any problems with the Oxford 936 chip, so it appears it's the best fit for your needs here.

awsome! we finally have an answer of sorts!!! the 4 bay will have lots of expandability for me. im probably going to purchase 2x2TBs or 2x1.5TBs to start off with in RAID1 mirrored mode. then when i save up a bit more hopefully the 2TBs will have come down again.

its not possible to expand RAID1s is it? you have to copy off, format, then copy the data back on? easy enough to do!.

cheers nano