Loading ...
Sorry, an error occurred while loading the content.

RE: [nslu2-general] transfer rate should be?

Expand Messages
  • Michael Wagner
    ... It is beginning to look like there are several, possibly unrelated, problems. When driven by different versions of Windows you can get radically different
    Message 1 of 14 , Feb 6, 2005
    • 0 Attachment
      At 2/6/05 06:51 PM, Inge Bjørnvall Arnesen wrote:
      >It is a slug and a slow one at that, and the reason is still unclear.

      It is beginning to look like there are several, possibly unrelated, problems.

      When driven by different versions of Windows you can get radically
      different write performance, from 150KB/s to 1.8MB/s (i.e. more than an
      order of magnitude difference).

      One theory was that it had to do with TCP window size, and bad or faulty
      logic to negotiate the window size.

      I was unable to test this further because I had to take the disk back into
      active service.

      Michael

      http://home.cogeco.ca/~michaelwagner/personal-page.htm
      "All I wanna do is have a little fun before I die" Sheryl Crow


      [Non-text portions of this message have been removed]
    • donna foley
      This is interesting. a 2nd PC backup today went faster. the first PC a P3 600mhz backup to NSLU2 went at about 12 Mbit/s with Network at 12% utilization on
      Message 2 of 14 , Feb 6, 2005
      • 0 Attachment
        This is interesting. a 2nd PC backup today went faster.
        the first PC a P3 600mhz backup to NSLU2 went at about 12 Mbit/s with
        Network at 12% utilization on average.
        the 2nd PC, a P4 1.8ghz backup to NSLU2 went at about 24 Mbit/s with
        Network at 25% utilization. The USB drive access light is running at
        about a 33% duty cycle so the drive is not maxed out.

        Is 25 Mbit/s about the best one can reasonably expect to get on a 100
        baseT?

        It doesnt make sense the first PC was so much slower. I noticed the
        ethernet card is in the last slot which I heard can slow it down cuz
        of interupts. I might try moving NIC and see if that makes a difference.

        Still, even at 25 Mb/s is a long time to wait for 44Gbyte to
        backup..about 3.5 - 4 hours. I think a direct connect USB drive takes
        about 1/3 as long and a backup to a IDE drive is even less.

        Are there ways to easily tweek the LAN to get significant transfer
        rate performance? Or is 25% utilization the best I shold expect on a
        100mb/s LAN?

        It is nice having the NAS on all the time and not worry about having
        to make sure someones PC acting as a server is powered on. I do wish
        is was faster and its disk format was NFTS, but this might work out
        afterall.

        thanks for any tips

        --- In nslu2-general@yahoogroups.com, Michael Wagner <michael@t...> wrote:
        > At 2/6/05 06:51 PM, Inge Bjørnvall Arnesen wrote:
        > >It is a slug and a slow one at that, and the reason is still unclear.
        >
        > It is beginning to look like there are several, possibly unrelated,
        problems.
        >
        > When driven by different versions of Windows you can get radically
        > different write performance, from 150KB/s to 1.8MB/s (i.e. more than an
        > order of magnitude difference).
        >
        > One theory was that it had to do with TCP window size, and bad or
        faulty
        > logic to negotiate the window size.
        >
        > I was unable to test this further because I had to take the disk
        back into
        > active service.
        >
        > Michael
        >
        > http://home.cogeco.ca/~michaelwagner/personal-page.htm
        > "All I wanna do is have a little fun before I die" Sheryl Crow
        >
        >
        > [Non-text portions of this message have been removed]
      • Gary Gutman
        Before I had settled on my enclosure (AMS Venus DS3), I had done some extesive research on realistic through put expectations. I found out that the cypress
        Message 3 of 14 , Feb 8, 2005
        • 0 Attachment
          Before I had settled on my enclosure (AMS Venus DS3), I had done
          some extesive research on realistic through put expectations. I
          found out that the cypress chipset was one of the highest performing
          chipest out there. I was unable to find any enclosure that perfomed
          much higher that about 25Mb/s when attached directly to a PC and
          using HDTach as the benchmark. Same exact drives would produce
          numbers that were almost doubled when attached directly to the IDE
          channel. Most reviewers attributed the loss of performance to the
          additional overhead of the USB chipsets inloved.

          Check out the following reviews of my enclosure, which consitently
          received high marks.
          http://www.extremeoverclocking.com/reviews/storage/Venus_DS3_Enclosur
          e_5.html

          http://www.barrys-rigs-n-
          reviews.com/reviews/2004/hardware/venus/venusds3_2.htm


          So the numbers you are seeing are more or less on par with what's
          expected from external enclosures.

          Regarsd,
          mrgary

          --- In nslu2-general@yahoogroups.com, "donna foley"
          <dontbfooled2000@y...> wrote:
          >
          >
          > This is interesting. a 2nd PC backup today went faster.
          > the first PC a P3 600mhz backup to NSLU2 went at about 12 Mbit/s
          with
          > Network at 12% utilization on average.
          > the 2nd PC, a P4 1.8ghz backup to NSLU2 went at about 24 Mbit/s
          with
          > Network at 25% utilization. The USB drive access light is running
          at
          > about a 33% duty cycle so the drive is not maxed out.
          >
          > Is 25 Mbit/s about the best one can reasonably expect to get on a
          100
          > baseT?
          >
          > It doesnt make sense the first PC was so much slower. I noticed
          the
          > ethernet card is in the last slot which I heard can slow it down
          cuz
          > of interupts. I might try moving NIC and see if that makes a
          difference.
          >
          > Still, even at 25 Mb/s is a long time to wait for 44Gbyte to
          > backup..about 3.5 - 4 hours. I think a direct connect USB drive
          takes
          > about 1/3 as long and a backup to a IDE drive is even less.
          >
          > Are there ways to easily tweek the LAN to get significant transfer
          > rate performance? Or is 25% utilization the best I shold expect
          on a
          > 100mb/s LAN?
          >
          > It is nice having the NAS on all the time and not worry about
          having
          > to make sure someones PC acting as a server is powered on. I do
          wish
          > is was faster and its disk format was NFTS, but this might work out
          > afterall.
          >
          > thanks for any tips
          >
          > --- In nslu2-general@yahoogroups.com, Michael Wagner
          <michael@t...> wrote:
          > > At 2/6/05 06:51 PM, Inge Bjørnvall Arnesen wrote:
          > > >It is a slug and a slow one at that, and the reason is still
          unclear.
          > >
          > > It is beginning to look like there are several, possibly
          unrelated,
          > problems.
          > >
          > > When driven by different versions of Windows you can get
          radically
          > > different write performance, from 150KB/s to 1.8MB/s (i.e. more
          than an
          > > order of magnitude difference).
          > >
          > > One theory was that it had to do with TCP window size, and bad or
          > faulty
          > > logic to negotiate the window size.
          > >
          > > I was unable to test this further because I had to take the disk
          > back into
          > > active service.
          > >
          > > Michael
          > >
          > > http://home.cogeco.ca/~michaelwagner/personal-page.htm
          > > "All I wanna do is have a little fun before I die" Sheryl Crow
          > >
          > >
          > > [Non-text portions of this message have been removed]
        • donna foley
          thats real good input. HOwever the numbers Im reporting are bits. the Testing is reporting Bytes so my transfer via the NSLU2 is about 1/8 that reported in the
          Message 4 of 14 , Feb 8, 2005
          • 0 Attachment
            thats real good input. HOwever the numbers Im reporting are bits.
            the Testing is reporting Bytes so my transfer via the NSLU2 is about
            1/8 that reported in the tests. My drive is much much fasster when
            doing direct on the PC USB, its when operating via the NSLU2 that
            seems to slow it down significantly.

            --- In nslu2-general@yahoogroups.com, "Gary Gutman" <mrgaryg@y...>
            wrote:
            >
            > Before I had settled on my enclosure (AMS Venus DS3), I had done
            > some extesive research on realistic through put expectations. I
            > found out that the cypress chipset was one of the highest
            performing
            > chipest out there. I was unable to find any enclosure that perfomed
            > much higher that about 25Mb/s when attached directly to a PC and
            > using HDTach as the benchmark. Same exact drives would produce
            > numbers that were almost doubled when attached directly to the IDE
            > channel. Most reviewers attributed the loss of performance to the
            > additional overhead of the USB chipsets inloved.
            >
            > Check out the following reviews of my enclosure, which consitently
            > received high marks.
            >
            http://www.extremeoverclocking.com/reviews/storage/Venus_DS3_Enclosur
            > e_5.html
            >
            > http://www.barrys-rigs-n-
            > reviews.com/reviews/2004/hardware/venus/venusds3_2.htm
            >
            >
            > So the numbers you are seeing are more or less on par with what's
            > expected from external enclosures.
            >
            > Regarsd,
            > mrgary
            >
            > --- In nslu2-general@yahoogroups.com, "donna foley"
            > <dontbfooled2000@y...> wrote:
            > >
            > >
            > > This is interesting. a 2nd PC backup today went faster.
            > > the first PC a P3 600mhz backup to NSLU2 went at about 12 Mbit/s
            > with
            > > Network at 12% utilization on average.
            > > the 2nd PC, a P4 1.8ghz backup to NSLU2 went at about 24 Mbit/s
            > with
            > > Network at 25% utilization. The USB drive access light is
            running
            > at
            > > about a 33% duty cycle so the drive is not maxed out.
            > >
            > > Is 25 Mbit/s about the best one can reasonably expect to get on a
            > 100
            > > baseT?
            > >
            > > It doesnt make sense the first PC was so much slower. I noticed
            > the
            > > ethernet card is in the last slot which I heard can slow it down
            > cuz
            > > of interupts. I might try moving NIC and see if that makes a
            > difference.
            > >
            > > Still, even at 25 Mb/s is a long time to wait for 44Gbyte to
            > > backup..about 3.5 - 4 hours. I think a direct connect USB drive
            > takes
            > > about 1/3 as long and a backup to a IDE drive is even less.
            > >
            > > Are there ways to easily tweek the LAN to get significant transfer
            > > rate performance? Or is 25% utilization the best I shold expect
            > on a
            > > 100mb/s LAN?
            > >
            > > It is nice having the NAS on all the time and not worry about
            > having
            > > to make sure someones PC acting as a server is powered on. I do
            > wish
            > > is was faster and its disk format was NFTS, but this might work
            out
            > > afterall.
            > >
            > > thanks for any tips
            > >
            > > --- In nslu2-general@yahoogroups.com, Michael Wagner
            > <michael@t...> wrote:
            > > > At 2/6/05 06:51 PM, Inge Bjørnvall Arnesen wrote:
            > > > >It is a slug and a slow one at that, and the reason is still
            > unclear.
            > > >
            > > > It is beginning to look like there are several, possibly
            > unrelated,
            > > problems.
            > > >
            > > > When driven by different versions of Windows you can get
            > radically
            > > > different write performance, from 150KB/s to 1.8MB/s (i.e. more
            > than an
            > > > order of magnitude difference).
            > > >
            > > > One theory was that it had to do with TCP window size, and bad
            or
            > > faulty
            > > > logic to negotiate the window size.
            > > >
            > > > I was unable to test this further because I had to take the disk
            > > back into
            > > > active service.
            > > >
            > > > Michael
            > > >
            > > > http://home.cogeco.ca/~michaelwagner/personal-page.htm
            > > > "All I wanna do is have a little fun before I die" Sheryl Crow
            > > >
            > > >
            > > > [Non-text portions of this message have been removed]
          • Inge Bjørnvall Arnesen
            What are your enclosure read/write speeds (not network speeds, just disk) when hooked up to the NSLU2? -- Inge ... From: donna foley
            Message 5 of 14 , Feb 8, 2005
            • 0 Attachment
              What are your enclosure read/write speeds (not network speeds, just disk)
              when hooked up to the NSLU2?

              -- Inge

              ----Original Message----
              From: donna foley [mailto:dontbfooled2000@...]
              Sent: 8. februar 2005 17:52
              To: nslu2-general@yahoogroups.com
              Subject: [nslu2-general] Re: transfer rate should be?

              > thats real good input. HOwever the numbers Im reporting
              > are bits. the Testing is reporting Bytes so my transfer
              > via the NSLU2 is about 1/8 that reported in the tests.
              > My drive is much much fasster when doing direct on the PC
              > USB, its when operating via the NSLU2 that seems to slow
              > it down significantly.
              >
              > --- In nslu2-general@yahoogroups.com, "Gary Gutman" <mrgaryg@y...>
              > wrote:
              > >
              > > Before I had settled on my enclosure (AMS Venus DS3), I
              > > had done some extesive research on realistic through
              > > put expectations. I found out that the cypress chipset
              > > was one of the highest
              > performing
              > > chipest out there. I was unable to find any enclosure
              > > that perfomed much higher that about 25Mb/s when
              > > attached directly to a PC and using HDTach as the benchmark. Same
              > > exact drives would produce numbers that were almost doubled when
              > > attached directly to the IDE channel. Most reviewers attributed the
              > > loss of performance to the additional overhead of the USB
              > > chipsets inloved.
              > >
              > > Check out the following reviews of my enclosure, which consitently
              > > received high marks.
              > >
              > http://www.extremeoverclocking.com/reviews/storage/Venus_DS3_Enclosur
              > > e_5.html
              > >
              > > http://www.barrys-rigs-n-
              > > reviews.com/reviews/2004/hardware/venus/venusds3_2.htm
              > >
              > >
              > > So the numbers you are seeing are more or less on par
              > > with what's expected from external enclosures.
              > >
              > > Regarsd,
              > > mrgary
              > >
              > > --- In nslu2-general@yahoogroups.com, "donna foley"
              > > <dontbfooled2000@y...> wrote:
              > > >
              > > >
              > > > This is interesting. a 2nd PC backup today went
              > > > faster.
              > > > the first PC a P3 600mhz backup to NSLU2 went at
              > > > about 12 Mbit/s
              > > with
              > > > Network at 12% utilization on average.
              > > > the 2nd PC, a P4 1.8ghz backup to NSLU2 went at about
              > > > 24 Mbit/s
              > > with
              > > > Network at 25% utilization. The USB drive access
              > > > light is
              > running
              > > at
              > > > about a 33% duty cycle so the drive is not maxed out.
              > > >
              > > > Is 25 Mbit/s about the best one can reasonably expect
              > > > to get on a
              > > 100
              > > > baseT?
              > > >
              > > > It doesnt make sense the first PC was so much slower.
              > > > I noticed
              > > the
              > > > ethernet card is in the last slot which I heard can
              > > > slow it down
              > > cuz
              > > > of interupts. I might try moving NIC and see if that makes a
              > > difference.
              > > >
              > > > Still, even at 25 Mb/s is a long time to wait for
              > > > 44Gbyte to backup..about 3.5 - 4 hours. I think a
              > > > direct connect USB drive
              > > takes
              > > > about 1/3 as long and a backup to a IDE drive is even
              > > > less.
              > > >
              > > > Are there ways to easily tweek the LAN to get
              > > > significant transfer rate performance? Or is 25%
              > > > utilization the best I shold expect
              > > on a
              > > > 100mb/s LAN?
              > > >
              > > > It is nice having the NAS on all the time and not
              > > > worry about
              > > having
              > > > to make sure someones PC acting as a server is
              > > > powered on. I do
              > > wish
              > > > is was faster and its disk format was NFTS, but this
              > > > might work
              > out
              > > > afterall.
              > > >
              > > > thanks for any tips
              > > >
              > > > --- In nslu2-general@yahoogroups.com, Michael Wagner
              > > <michael@t...> wrote:
              > > > > At 2/6/05 06:51 PM, Inge Bjørnvall Arnesen wrote:
              > > > > > It is a slug and a slow one at that, and the
              > > > > > reason is still
              > > unclear.
              > > > >
              > > > > It is beginning to look like there are several,
              > > > > possibly
              > > unrelated,
              > > > problems.
              > > > >
              > > > > When driven by different versions of Windows you
              > > > > can get
              > > radically
              > > > > different write performance, from 150KB/s to
              > > > > 1.8MB/s (i.e. more
              > > than an
              > > > > order of magnitude difference).
              > > > >
              > > > > One theory was that it had to do with TCP window
              > > > > size, and bad
              > or
              > > > faulty
              > > > > logic to negotiate the window size.
              > > > >
              > > > > I was unable to test this further because I had to
              > > > > take the disk
              > > > back into
              > > > > active service.
              > > > >
              > > > > Michael
              > > > >
              > > > > http://home.cogeco.ca/~michaelwagner/personal-page.htm
              > > > > "All I wanna do is have a little fun before I die"
              > > > > Sheryl Crow
              > > > >
              > > > >
              > > > > [Non-text portions of this message have been
              > > > > removed]
              >
              >
              >
              >
              >
              >
              > Yahoo! Groups Links
              >
              >
              >
            • donna foley
              I dont know how to measure or read rate from NSLU2 to USB drive....however..... I can see is that the average transfer rate, from PC over LAN to NSLU2, without
              Message 6 of 14 , Feb 8, 2005
              • 0 Attachment
                I dont know how to measure or read rate from NSLU2 to USB
                drive....however.....

                I can see is that the average transfer rate, from PC over LAN to
                NSLU2, without any other network activity, and it is 25-30
                MegaBits/second. Thats on 2 seperate and brand new PCs (P4 3Ghz. XP
                SP2). Im fairly sure that the disk or enclosure are not near data
                input rate capacity as the Disk access light is only on for about 20-
                30% of the time during a backup. ON both of these the CPU is about
                30% utilized.

                On an older P3 600Mhz XP PC, the transfer rate is only about 12
                Megbits/s and CPU is 100% utilized.

                I think the Wiki referred to early, I think you posted?....states
                about the same transfer rates as Im seeing and ultimately lays the
                poor performance blame on the NSLU2 with hopefully a firmware
                improvement down the road to improve that.

                I guess thats where I am on it now. I have to live with it. Its
                still a better situation then using one of my PCs as a server. But a
                44G backup takes over 4 hours. I should say that Im using Ghost to
                backup disk images and it does do processing and compression but Im
                sure thats not the bottle neck becuase I can backup Disk to Disk or
                ever Disk to USB connected direct to the PC in probably 1/5 the time
                as when using the NFSLU2.

                If you think there is a way to improve transfer rate, Id love to hear
                it.

                thanks!


                --- In nslu2-general@yahoogroups.com, Inge Bjørnvall Arnesen
                <i.b.arnesen@f...> wrote:
                > What are your enclosure read/write speeds (not network speeds, just
                disk)
                > when hooked up to the NSLU2?
                >
                > -- Inge
                >
                > ----Original Message----
                > From: donna foley [mailto:dontbfooled2000@y...]
                > Sent: 8. februar 2005 17:52
                > To: nslu2-general@yahoogroups.com
                > Subject: [nslu2-general] Re: transfer rate should be?
                >
                > > thats real good input. HOwever the numbers Im reporting
                > > are bits. the Testing is reporting Bytes so my transfer
                > > via the NSLU2 is about 1/8 that reported in the tests.
                > > My drive is much much fasster when doing direct on the PC
                > > USB, its when operating via the NSLU2 that seems to slow
                > > it down significantly.
                > >
                > > --- In nslu2-general@yahoogroups.com, "Gary Gutman"
                <mrgaryg@y...>
                > > wrote:
                > > >
                > > > Before I had settled on my enclosure (AMS Venus DS3), I
                > > > had done some extesive research on realistic through
                > > > put expectations. I found out that the cypress chipset
                > > > was one of the highest
                > > performing
                > > > chipest out there. I was unable to find any enclosure
                > > > that perfomed much higher that about 25Mb/s when
                > > > attached directly to a PC and using HDTach as the benchmark.
                Same
                > > > exact drives would produce numbers that were almost doubled
                when
                > > > attached directly to the IDE channel. Most reviewers
                attributed the
                > > > loss of performance to the additional overhead of the USB
                > > > chipsets inloved.
                > > >
                > > > Check out the following reviews of my enclosure, which
                consitently
                > > > received high marks.
                > > >
                > >
                http://www.extremeoverclocking.com/reviews/storage/Venus_DS3_Enclosur
                > > > e_5.html
                > > >
                > > > http://www.barrys-rigs-n-
                > > > reviews.com/reviews/2004/hardware/venus/venusds3_2.htm
                > > >
                > > >
                > > > So the numbers you are seeing are more or less on par
                > > > with what's expected from external enclosures.
                > > >
                > > > Regarsd,
                > > > mrgary
                > > >
                > > > --- In nslu2-general@yahoogroups.com, "donna foley"
                > > > <dontbfooled2000@y...> wrote:
                > > > >
                > > > >
                > > > > This is interesting. a 2nd PC backup today went
                > > > > faster.
                > > > > the first PC a P3 600mhz backup to NSLU2 went at
                > > > > about 12 Mbit/s
                > > > with
                > > > > Network at 12% utilization on average.
                > > > > the 2nd PC, a P4 1.8ghz backup to NSLU2 went at about
                > > > > 24 Mbit/s
                > > > with
                > > > > Network at 25% utilization. The USB drive access
                > > > > light is
                > > running
                > > > at
                > > > > about a 33% duty cycle so the drive is not maxed out.
                > > > >
                > > > > Is 25 Mbit/s about the best one can reasonably expect
                > > > > to get on a
                > > > 100
                > > > > baseT?
                > > > >
                > > > > It doesnt make sense the first PC was so much slower.
                > > > > I noticed
                > > > the
                > > > > ethernet card is in the last slot which I heard can
                > > > > slow it down
                > > > cuz
                > > > > of interupts. I might try moving NIC and see if that makes a
                > > > difference.
                > > > >
                > > > > Still, even at 25 Mb/s is a long time to wait for
                > > > > 44Gbyte to backup..about 3.5 - 4 hours. I think a
                > > > > direct connect USB drive
                > > > takes
                > > > > about 1/3 as long and a backup to a IDE drive is even
                > > > > less.
                > > > >
                > > > > Are there ways to easily tweek the LAN to get
                > > > > significant transfer rate performance? Or is 25%
                > > > > utilization the best I shold expect
                > > > on a
                > > > > 100mb/s LAN?
                > > > >
                > > > > It is nice having the NAS on all the time and not
                > > > > worry about
                > > > having
                > > > > to make sure someones PC acting as a server is
                > > > > powered on. I do
                > > > wish
                > > > > is was faster and its disk format was NFTS, but this
                > > > > might work
                > > out
                > > > > afterall.
                > > > >
                > > > > thanks for any tips
                > > > >
                > > > > --- In nslu2-general@yahoogroups.com, Michael Wagner
                > > > <michael@t...> wrote:
                > > > > > At 2/6/05 06:51 PM, Inge Bjørnvall Arnesen wrote:
                > > > > > > It is a slug and a slow one at that, and the
                > > > > > > reason is still
                > > > unclear.
                > > > > >
                > > > > > It is beginning to look like there are several,
                > > > > > possibly
                > > > unrelated,
                > > > > problems.
                > > > > >
                > > > > > When driven by different versions of Windows you
                > > > > > can get
                > > > radically
                > > > > > different write performance, from 150KB/s to
                > > > > > 1.8MB/s (i.e. more
                > > > than an
                > > > > > order of magnitude difference).
                > > > > >
                > > > > > One theory was that it had to do with TCP window
                > > > > > size, and bad
                > > or
                > > > > faulty
                > > > > > logic to negotiate the window size.
                > > > > >
                > > > > > I was unable to test this further because I had to
                > > > > > take the disk
                > > > > back into
                > > > > > active service.
                > > > > >
                > > > > > Michael
                > > > > >
                > > > > > http://home.cogeco.ca/~michaelwagner/personal-page.htm
                > > > > > "All I wanna do is have a little fun before I die"
                > > > > > Sheryl Crow
                > > > > >
                > > > > >
                > > > > > [Non-text portions of this message have been
                > > > > > removed]
                > >
                > >
                > >
                > >
                > >
                > >
                > > Yahoo! Groups Links
                > >
                > >
                > >
              • Inge Bjørnvall Arnesen
                ... For how to do a rough speed measurement of your disks on the slug, see the Disk speed section of http://www.nslu2-linux.org/wiki/Info/Performance In
                Message 7 of 14 , Feb 8, 2005
                • 0 Attachment
                  > I dont know how to measure or read rate from NSLU2 to USB
                  > drive....however.....

                  For how to do a rough speed measurement of your disks on the slug, see the
                  "Disk speed" section of http://www.nslu2-linux.org/wiki/Info/Performance

                  In general: If your Samba transfer numbers differ significantly from the
                  numbers found on the wiki, then you could do worse than trying to isolate
                  which of the chains of the link that are to blame.

                  > I think the Wiki referred to early, I think you
                  > posted?....states
                  > about the same transfer rates as Im seeing and ultimately
                  > lays the
                  > poor performance blame on the NSLU2 with hopefully a
                  > firmware
                  > improvement down the road to improve that.

                  Don't bet on that. For future investment in your NSLU2, expect performance
                  to remaint pretty much like it is today.

                  > If you think there is a way to improve transfer rate, Id
                  > love to hear it.

                  Well, it would involve looking at the datasheets and see if the appalingly
                  slow USB read/write speeds are in tune with them. If they are - forget it -
                  just bad HW design. If they are not, or there are no such performance
                  claims, hooking up a USB bus analyser and looking at the driver could be a
                  way to go (for those capable and with the right equipment).

                  best,

                  -- Inge
                • Michael Wagner
                  ... As would we all ... :-) Michael http://home.cogeco.ca/~michaelwagner/personal-page.htm All I wanna do is have a little fun before I die Sheryl Crow
                  Message 8 of 14 , Feb 8, 2005
                  • 0 Attachment
                    At 2/8/05 01:03 PM, donna foley wrote:
                    >If you think there is a way to improve transfer rate, Id love to hear it.

                    As would we all ... :-)

                    Michael

                    http://home.cogeco.ca/~michaelwagner/personal-page.htm
                    "All I wanna do is have a little fun before I die" Sheryl Crow


                    [Non-text portions of this message have been removed]
                  • donna foley
                    OK..thanks. Im running the factory OS. I suspect your referring to one the other Linux loads below? I looked at the Buffalo NAS 200gb unit to see performance
                    Message 9 of 14 , Feb 8, 2005
                    • 0 Attachment
                      OK..thanks. Im running the factory OS. I suspect your referring to
                      one the other Linux loads below?

                      I looked at the Buffalo NAS 200gb unit to see performance comparison
                      and found some info....I think it was PC Magazine..sorry dont have
                      url...(I did a search on Buffalo NAS). But write speed was about
                      5.5MegBytes/s or about 44Megabits/s. I would think thats unit would
                      represent what is typical for commerical purpose built NAS w/ matched
                      harddrive. Thats not to horribly far off what Im seeing which can be
                      a sustained high as much as about 35 Megbit/s. So its not like Nslu2
                      is way off much. It could also be something else in my system
                      slowing things down.


                      --- In nslu2-general@yahoogroups.com, Inge Bjørnvall Arnesen
                      <i.b.arnesen@f...> wrote:
                      > > I dont know how to measure or read rate from NSLU2 to USB
                      > > drive....however.....
                      >
                      > For how to do a rough speed measurement of your disks on the slug,
                      see the
                      > "Disk speed" section of http://www.nslu2-
                      linux.org/wiki/Info/Performance
                      >
                      > In general: If your Samba transfer numbers differ significantly
                      from the
                      > numbers found on the wiki, then you could do worse than trying to
                      isolate
                      > which of the chains of the link that are to blame.
                      >
                      > > I think the Wiki referred to early, I think you
                      > > posted?....states
                      > > about the same transfer rates as Im seeing and ultimately
                      > > lays the
                      > > poor performance blame on the NSLU2 with hopefully a
                      > > firmware
                      > > improvement down the road to improve that.
                      >
                      > Don't bet on that. For future investment in your NSLU2, expect
                      performance
                      > to remaint pretty much like it is today.
                      >
                      > > If you think there is a way to improve transfer rate, Id
                      > > love to hear it.
                      >
                      > Well, it would involve looking at the datasheets and see if the
                      appalingly
                      > slow USB read/write speeds are in tune with them. If they are -
                      forget it -
                      > just bad HW design. If they are not, or there are no such
                      performance
                      > claims, hooking up a USB bus analyser and looking at the driver
                      could be a
                      > way to go (for those capable and with the right equipment).
                      >
                      > best,
                      >
                      > -- Inge
                    • Michael Wagner
                      ... Long ago, in a previous lifetime it seems, I was a performance analyst. It was before the time of 100BT, but the basic ideas are the same. About the best
                      Message 10 of 14 , Feb 8, 2005
                      • 0 Attachment
                        At 2/8/05 07:01 PM, you wrote:
                        >Buffalo NAS ... write speed was about 5.5MegBytes/s or about 44Megabits/s.

                        Long ago, in a previous lifetime it seems, I was a performance analyst.

                        It was before the time of 100BT, but the basic ideas are the same.

                        About the best performance you can expect out of shared media (old style
                        co-ax ethernet or ethernet with hubs rather than switches) is about half
                        the theoretical bandwidth, because collisions (and collision avoidance
                        techniques) eat up a lot and each node has a certain latency when it gets
                        the go-ahead again.

                        So 100BT (the "normal" ethernet these days) should deliver about 50Mb/s,
                        more or less.

                        If the Buffalo does 44, that's not bad.

                        A switch rather than a hub should theoretically improve the situation, but
                        it depends on a lot of factors whether it in fact does or not.

                        My tests of the Linksys (hub, not switch) yielded a maximum speed of
                        1.8MB/s or 14.4Mb/s (so nowhere near the speed where the network is the
                        bottleneck).

                        If and when time permits, I'll hook up my slug again and try to test it
                        more rigorously.

                        Don't hold your breath, though :-( Life has been hectic.

                        Michael

                        > I would think thats unit would
                        >represent what is typical for commerical purpose built NAS w/ matched
                        >harddrive. Thats not to horribly far off what Im seeing which can be
                        >a sustained high as much as about 35 Megbit/s. So its not like Nslu2
                        >is way off much. It could also be something else in my system
                        >slowing things down.
                        >
                        >
                        >--- In nslu2-general@yahoogroups.com, Inge Bjørnvall Arnesen
                        ><i.b.arnesen@f...> wrote:
                        > > > I dont know how to measure or read rate from NSLU2 to USB
                        > > > drive....however.....
                        > >
                        > > For how to do a rough speed measurement of your disks on the slug,
                        >see the
                        > > "Disk speed" section of http://www.nslu2-
                        >linux.org/wiki/Info/Performance
                        > >
                        > > In general: If your Samba transfer numbers differ significantly
                        >from the
                        > > numbers found on the wiki, then you could do worse than trying to
                        >isolate
                        > > which of the chains of the link that are to blame.
                        > >
                        > > > I think the Wiki referred to early, I think you
                        > > > posted?....states
                        > > > about the same transfer rates as Im seeing and ultimately
                        > > > lays the
                        > > > poor performance blame on the NSLU2 with hopefully a
                        > > > firmware
                        > > > improvement down the road to improve that.
                        > >
                        > > Don't bet on that. For future investment in your NSLU2, expect
                        >performance
                        > > to remaint pretty much like it is today.
                        > >
                        > > > If you think there is a way to improve transfer rate, Id
                        > > > love to hear it.
                        > >
                        > > Well, it would involve looking at the datasheets and see if the
                        >appalingly
                        > > slow USB read/write speeds are in tune with them. If they are -
                        >forget it -
                        > > just bad HW design. If they are not, or there are no such
                        >performance
                        > > claims, hooking up a USB bus analyser and looking at the driver
                        >could be a
                        > > way to go (for those capable and with the right equipment).
                        > >
                        > > best,
                        > >
                        > > -- Inge
                        >
                        >
                        >
                        >
                        >
                        >
                        >Yahoo! Groups Links
                        >
                        >
                        >
                        >

                        http://home.cogeco.ca/~michaelwagner/personal-page.htm
                        "All I wanna do is have a little fun before I die" Sheryl Crow


                        [Non-text portions of this message have been removed]
                      • Michael Wagner
                        ... I should have said, out of shared media running IP I believe certain combinations of protocols and circumstances can do better than 50%, at least in the
                        Message 11 of 14 , Feb 8, 2005
                        • 0 Attachment
                          At 2/8/05 08:48 PM, I wrote:
                          >About the best performance you can expect out of shared media (old style
                          >co-ax ethernet or ethernet with hubs rather than switches)

                          I should have said, "out of shared media running IP"

                          I believe certain combinations of protocols and circumstances can do better
                          than 50%, at least in the short term.

                          Michael

                          http://home.cogeco.ca/~michaelwagner/personal-page.htm
                          "All I wanna do is have a little fun before I die" Sheryl Crow


                          [Non-text portions of this message have been removed]
                        Your message has been successfully submitted and would be delivered to recipients shortly.