1

Topic: AoE (iSCSI) on the contrary - it is implemented?

The piece of iron which is pretending to be SATA the drive is necessary. Thus implementation of the drive - virtual, is on other computer, and is given to a piece of iron on Ethernet, as a variant - on similar iSCSI to the protocol. There is a such?

2

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, IID, you wrote: IID> There is a such? That is, you have a certain piece of iron with port SATA and without port ethernet, and the task to pick up here a network share? At all I do not know, the special task what that, rare. From that the piece of iron will cost fair money. Saw such, stuck in USB and pretending to be a flash card. Very slow, chtenie/zapis did not hold out and to megabyte per second.

3

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, Stanislaw K, you wrote: SK> that is, you have a certain piece of iron with port SATA and without port ethernet, and the task to pick up here a network share? The task in emulation SATA of the drive. Interception of all operations of record-reading and  the image of a disk. From other machine. SK> at all I do not know, the special task what that, rare. From that the piece of iron will cost fair money. The Task simple. All spoil  frequencies SATA of the bus. Program with it not to consult. While such variant was invented: we take SATA-IDE the adapter for copecks, and it is hung up ready STM32 ARM a board on IDE . It will contact the computer on USB. In the theory nearby 50mb-sek it is possible to receive. In real the crude device well if 20 produces.

4

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, IID, you wrote: IID> There is a such? Is like disk regiments AoE http://xgu.ru/wiki/ATA_over_Ethernet

5

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, Glory, you wrote: There is like disk regiments AoE http://xgu.ru/wiki/ATA_over_Ethernet At me exactly reverse task: to give physical SATA the interface not to a disk, and a host. Virtual I should have a disk, instead of its representation (the driver over LAN) in OS. The protocol probably suits. A question in its implementation.

6

Re: AoE (iSCSI) on the contrary - it is implemented?

IID> the piece of iron which is pretending to be SATA the drive Is necessary. Thus implementation of the drive - virtual, is on other computer, and is given to a piece of iron on Ethernet, as a variant - on similar iSCSI to the protocol. IID> there is a such? And it is necessary here it is direct SATA? USB mass storage on idea  to make almost without a soldering iron, on some .

7

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, ononim, you wrote: O> And it is necessary here it is direct SATA? USB mass storage on idea  to make almost without a soldering iron, on some . There are some objections: - USB on  the network sits on the same IO, as. Speed falls monumentally. - it is necessary to squat, convincing OS that this Mass Storage is pleasant to it also it wants to be installed on it. But about  council good. It is possible to look among clones - can eat somewhere  the controler which is able in the client. GPIO on  will be again more bright. And storage has more. And it at me too is As changeover  with STM32 - quite norms a variant.

8

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, IID, you wrote: IID> But about  council good. It is possible to look among clones - can eat somewhere  the controler which is able in the client. IID> GPIO on  will be again more bright. And storage has more. And it at me too is As changeover  with STM32 - quite norms a variant.  it is filled up by adapters SATA on SDCARD.  it is enough to teach to pretend to be SD a card.  it is possible to make of several raid.

9

Re: AoE (iSCSI) on the contrary - it is implemented?

SK>  it is enough to teach to pretend to be SD a card. SK>  it is possible to make of several raid. Then to use SATA-PATA the bridge + a feint ears.

10

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, ononim, you wrote: SK>>  it is enough to teach to pretend to be SD a card. O> then to use SATA-PATA the bridge + a feint ears. SATA-> (sata-ide the bridge)-> PATA-> (NetPieIDE FPGA)-> SPI-> (RaspberriPie)-> ethernet Actually, I also offered it, only without the intermediate conversion in IDE. At once in SPI. SATA-> (sdcard adapter)-> SPI-> (RaspberriPie)-> ethernet p.s. As this NetPieIDE has been started to become the fan for pc-junior in 2014 and in 2017 to mind is not finished. It is thrown in an unknown state. In the theory he should be able PIO-4, but whether completely not clearly is able, yes where it is so much, in the same place a host - pc junior. intel 8088, 4.77 MHz, 64 Kilobyte of storage. SATA-SDCARD the adapter is a ready commercial product, bugs are licked on so many that its sellers sell clients buy and do not complain. NetPieIDE the good, interesting project for fun. For itself. As the occasion to run away from the wife and the mother-in-law of a hobby.

11

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, IID, you wrote: IID> the Task simple. All spoil  frequencies SATA of the bus. Program with it not to consult. IID> while such variant was invented: we take SATA-IDE the adapter for copecks, and it is hung up ready STM32 ARM a board on IDE . It will contact the computer on USB. In the theory nearby 50mb-sek it is possible to receive. In real the crude device well if 20 produces. And what for STM32 is able to work with 16-bit parallel port on frequency in 100 + MHz? Even F7 hardly digests 90 MHz SDRAM Briefly forget. To you it is necessary FPGA with  transceivers, and rather fresh - type Artix-7 which are able in SATA 6 Gbps.

12

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, Stanislaw K, you wrote: SK> SATA-> (sata-ide the bridge)-> PATA-> (NetPieIDE FPGA)-> SPI-> (RaspberriPie)-> ethernet SK> Actually, I also offered it, only without the intermediate conversion in IDE. At once in SPI. SK> SATA-> (sdcard adapter)-> SPI-> (RaspberriPie)-> ethernet Perverts - they the same people. Only they perverts. Take Artix-7 and do the normal decision, instead of pile up against each other crutches. If not hunting to solder BGA most these  are on ready units already with  and  and  (for loading). Simply add  to extension port!

13

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, koandrew, you wrote: K> Take Artix-7 and do the normal decision, instead of pile up against each other crutches. If not hunting to solder BGA most these  are on ready units already with  and  and  (for loading). Simply add  to extension port! ! And at it Atrix-7 already is ready ethernet? TCP/IP? AoE? Or it is necessary most to write  and implementation AoE? Whether overkill it?

14

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, koandrew, you wrote: K> And what for STM32 is able to work with 16-bit parallel port on frequency in 100 + MHz? Even F7 hardly digests 90 MHz SDRAM With 16 bit enough 25 (ATA-100) or 33 (ATA-133). But! Both on front and on recession. Or is 4 SPI to half of clock frequency (168/2 for F4). I.e. if  the decision with TF adapters which also itself  access - is possible between SATA and STM to build an exchange on 80 on everyone SPI, i.e. 40 Mb-sek in spot-check, at a successful deal. The problem begins to exchange as them with a computer Since without  is accessible only FullSpeed, on 12. And with  no more 30-40, . K> Briefly forget. To you it is necessary FPGA with  transceivers, and rather fresh - type Artix-7 which are able in SATA 6 Gbps. Gigabits budgetary  in megabits, without FPGA and PHY. But also with megabits uneasy to consult that Mahlo that I  (or what there?) I do not know... Source codes of implementations SATA/Ethernet where to take?

15

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, Stanislaw K, you wrote: SK> Pogodika! And at it Atrix-7 already is ready ethernet? TCP/IP? AoE? All is, if you are ready to pay the big grandmas for IP SK> or it is necessary most to write  and implementation AoE? If at you as at me, tens kilodollars do not burn a hole in a purse it is necessary to saw partially most. However, the Xilinx gives something and for so - for example 10/100 Ethernet MAC (1000 currents for the grandma), or a wrapper over  transceivers with implementation of physical (signal) level. That in general it is already a lot of. Generally the interesting project - whether it will be necessary to think SK> overkill it? Why overkill? A normal engineering solution of a problem. Actually  also are intended for the decision of tasks for which there are no ready decisions in the form of type chips "thrust and earned".

16

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, IID, you wrote: IID> the Problem begins to exchange as them with a computer Since without  is accessible only FullSpeed, on 12. And with  no more 30-40, . Yes here generally the processor - not the most successful decision, since the task stream that is why the sequence  here a current hinders. Such tasks ideally lay down on COTTON VELVET. IID> gigabits budgetary  in megabits, without FPGA and PHY. But also with megabits uneasy to consult IID> that Mahlo that I  (or what there?) I do not know... Source codes of implementations SATA/Ethernet where to take? Well 10/100 Ethernet MAC the Xilinx gives implementation for how also implementation of channel level SATA. And further - or you pay  for the ready decision in the form of IP, or you take in teeth I will bake and  I heard that it not too difficult - and is even something in  so to me it does not see  the task.

17

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, IID, you wrote: IID> the piece of iron which is pretending to be SATA the drive Is necessary. Thus implementation of the drive - virtual, is on other computer, and is given to a piece of iron on Ethernet, as a variant - on similar iSCSI to the protocol. IID> there is a such? And what hinders to launch the virtual machine and to pretend to be the virtual screw? To eat better https://openpowerfoundation.org/blogs/n … for-power/

18

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, kov_serg, you wrote: _> And what hinders to launch the virtual machine and to pretend to be the virtual screw? Necessity of start on real iron hinders.  too are widely used, and to them questions are not present. Just it would be desirable to receive a feature , roughly speaking. _> to eat better https://openpowerfoundation.org/blogs/n … for-power/ the Abrupt piece! But how she solves my task? Virtualization of disks (with possibility of simultaneous access to them from other computer).

19

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, koandrew, you wrote: whether SK>> overkill it? K> why overkill? A normal engineering solution of a problem. Actually  also are intended for the decision of tasks for which there are no ready decisions in the form of type chips "thrust and earned". Here, in my opinion, the question costs so - if it is individual need 1-3 copies I would not become  and did by hinged mounting on the adapters marked above. And if it is a least series 10 + that will do already intelligently the high-grade device.

20

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, Stanislaw K, you wrote: SK> Here, in my opinion, the question costs so - if it is individual need 1-3 copies I would not become  and did by hinged mounting on the adapters marked above. And if it is a least series 10 + that will do already intelligently the high-grade device. A problem that these decisions are absolutely nonequivalent. While the decision on COTTON VELVET allows to reach speeds, comparable with speed of a network (as slowest link), in your decision it is so much time delays and potential  that I even find it difficult to tell straight off that there will be the slowest link. At least, one problem sees that  is SPI Master, that is the date transmission direction will be reverse "" to a control flow that forces it permanently to interrogate port (or to hang on interruption), plus the fact of presence of OS in it also very negatively affects latency (to the decision on OS COTTON VELVET it is not necessary absolutely). That is the offered system of crutches at least on orders is worse than the normal decision.

21

Re: AoE (iSCSI) on the contrary - it is implemented?

K> the Problem that these decisions are absolutely nonequivalent. While the decision on COTTON VELVET allows to reach speeds, comparable with speed of a network (as slowest link), in your decision it is so much time delays and potential  that I even find it difficult to tell straight off that there will be the slowest link. At least, one problem sees that  is SPI Master, that is the date transmission direction will be reverse "" to a control flow that forces it permanently to interrogate port (or to hang on interruption), plus the fact of presence of OS in it also very negatively affects latency (to the decision on OS COTTON VELVET it is not necessary absolutely). That is the offered system of crutches at least on orders is worse than the normal decision. COTTON VELVET  SATA-high-grade  the client could even be sold, probably... The Startup, ,  on yeast?

22

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, ononim, you wrote: O> COTTON VELVET  SATA-high-grade  the client could even be sold, probably... O> the Startup, ,  on yeast? I too would be inscribed on some pieces.

23

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, koandrew, you wrote: SK>> Here, in my opinion, the question costs so - if it is individual need 1-3 copies I would not become  and did by hinged mounting on the adapters marked above. And if it is a least series 10 + that will do already intelligently the high-grade device. K> a problem that these decisions are absolutely nonequivalent. While the decision on COTTON VELVET allows to reach speeds, comparable with speed of a network (as slowest link), in your decision it is so much time delays and potential  that I even find it difficult to tell straight off that there will be the slowest link. Certainly, making of the decision, it is necessary to weigh all pluses and minuses of each variant. K> at least, one problem sees that  is SPI Master, that is the date transmission direction will be reverse "" to a control flow that forces it permanently to interrogate port (or to hang on interruption), Most likely you are right. But, well and let hangs. That does not hinder to write SPI slave using not SPI and GPIo the interface. I doubt that turns out more slowly SPI. K> plus the fact of presence of OS in it also very negatively affects latency (to the decision on OS COTTON VELVET it is not necessary absolutely). Here from "that side" in a network who is not known, it is not known as, will be host/server AoE. How  concerns that that host will be occupied (than that important) and another does not answer ten packets? At  there is a gigabyte , for caching of network operations. K> that is the offered system of crutches at least on orders is worse than the normal decision. Yes.

24

Re: AoE (iSCSI) on the contrary - it is implemented?

SK> Here from "that side" in a network who is not known, it is not known as, will be host/server AoE. How  concerns that that host will be occupied (than that important) and another does not answer ten packets? At  there is a gigabyte , for caching of network operations. What means as? As you will program and will be. Upon it is necessary to do  on . (I dealt AoE, for fun)

25

Re: AoE (iSCSI) on the contrary - it is implemented?

Hello, IID, you wrote: IID> the piece of iron which is pretending to be SATA the drive Is necessary. Thus implementation of the drive - virtual, is on other computer, and is given to a piece of iron on Ethernet, as a variant - on similar iSCSI to the protocol. And you it is critical SATA to pretend to be? Simply "people" that arrive easier, load a system kernel on a network, PXE, mount network to a full-sphere than the kernel (samba ssh cfs netfs ftp https) and all is able. And MS-DOS was able, both win nt, and  with .