eSATA External Storage Drive Reviewed

Tom's Hardware has a practical look at an eSATA drive offering from Taiwanese storage firm Thecus. From the article: "Thecus' N2050 is one of the first external twin-drive RAID boxes that uses eSATA. As expected, its performance was far better than what USB 2.0 offers. The end result is impressive. The date transfer rate of 30 MB/s that USB 2.0 offers does indeed pale in comparison to 100 MB/s for eSATA, while the WD1500 drives are capable of delivering even better performance in RAID 0. It is also good to see that Thecus did not throw the USB 2.0 interface away, because it is a nice backup interface whenyou want to use the device with other computers via USB 2.0."

Comments

eSATA drawbacks

[Ice Wewe]

eSATA seems to be the new buzz word in computing these days. The only problem is, that you're forced to use USB 2.0 or Ethernet to access the drive because very few motherboards are equipped with eSATA connectors. eSATA connectors have the same pins at the regular SATA connectors, but have different male and female connectors. Thus, although it may seem convient, if you really want to get full performence out of the drive, you'll have to buy one of the few motherboards that sports an eSATA port.

Also, mentioned in the article... the SATA bus boasts a wonderful 3Gb/s (or 300MB/s). This however, is not the bottle neck when it comes to performence. As the article mentions, the top SATA drives on the market today only get about 85MB/s read/write to the disk. So although you may get 300MB/s from the disk cache, and the controller, you'll never really get 300MB/s. Still, it's miles from Ultra-ATA.

Re:eSATA drawbacks

[Jeff DeMaagd]

Surprisingly, the eSATA connector doesn't need to be on the "motherboard". There are brackets that adapt the internal connectors to the card slot openings in the back. There are plenty of SATA and SATA RAID adapters with back port connectors for this very purpose so when you max out the internal bays, you can go external.

Re:eSATA drawbacks

[Kjella]

Or you can use the included eSATA card (yes, Thecus is aware most don't have that port). Furthermore, USB2/Firewire IS a bottleneck both in sustained and burst speed. And for video editing, notice that while the top speed is quite limited by the Thecus unit, the minimum sustained speed is very very close to an internal drive. I was considering getting one earlier but found a most unimpressive hard disk compatibility list and forum posts about incompatible drives. That made me go with a LaCie with disks already installed instead. Maybe next time...

Easy TiVo expansion

[pjcreath]

The upcoming (high def) TiVo Series 3 has an e-SATA port. Adding more recording space will be a whole lot easier than it was before, especially for the less adventurous folks.

D-cinema needs this, for starters.

[adamgeek]

do we really need anything that goes faster?

As someone who works in the realm of digital cinema, I can state unequivocally that "we" (those of us working within the context of d-cinema) definitely do. In fact, I was in LA last week to preview a prototype uncompressedHD recorder that used e-sata based storage expansion. But back to the raw numbers-- even if you're just using e-sata as an archiving interface (which it would be very convenient for), you'll need to move your footage to the devices in as close to realtime as possible. A 2k frame (2048x1556) will run 12MB (approx).. x24 (frames per sec).. 288MB/second bandwidth needed for uncompressed 2k. So yes, we do need faster than USB2.

Does the average user need this? No. But, if the hardware of a blazingly fast interface could manage to be affordable (and thus suitable to become ubiquitous, unlike previous expensive interfaces.. fiber, etc), it would work just as well for the person backing up photos of their cats as the person moving uncompressed scans of 35mm motion picture film.

you're missing the point

[RelliK]

Also, mentioned in the article... the SATA bus boasts a wonderful 3Gb/s (or 300MB/s). This however, is not the bottle neck when it comes to performence. As the article mentions, the top SATA drives on the market today only get about 85MB/s read/write to the disk. So although you may get 300MB/s from the disk cache, and the controller, you'll never really get 300MB/s. Still, it's miles from Ultra-ATA.

It is true that a single drive cannot saturate the SATA channel. (85MB/s is actually a _very_ generous estimate. Typical performance is closer to 50-60MB/s). So, SATA certainly doesn't need more bandwidth in the near future. However, for eSATA, the extra bandwidth is _very_ useful. It would allow manufacturers to produce RAIDs with eSATA ports instead of SCSI or FC. (Right now, you still need a SCSI or FC card if you want to get any sort of performance). This would effectively commoditise the low-end RAID market, which is a very welcome development.

Re:whole article ruined on page 6 for me:

[HardCase]

Gotta read it literally. No mention of redundancy. Of course, the missing (but, I guess, implied) extra sentence telling us that "only" half of our data would be gone wouldn't have killed them.

Yet another review stating an incorrect fact

[this great guy]

Every single review of storage technologies I have read in the past 2 years state at least one (IMHO very important) incorrect fact. This is no exception with this one:

Page 2: "The fastest 3.5" SATA drives do not exceed 85 MB/s. A data transfer rate of 300 MB/s between a PC and a SATA drive cannot thus be matched by the speed of a SATA drive." Yes it can. When data is exchanged to/from the disk's cache, data throughput of 250+ MB/s can be achieved for a fraction of a second. Even if it's only for a fraction of a second, it is still important (else manufacturers would not even put cache memory on disks).

Re:Yet another review stating an incorrect fact

[this great guy]

It is important in scenarios where a lot of small burst transfers happen almost all the time, like on a busy file server. In this case switching from 150 MB/s sata links to 300 MB/s ones will reduce the PCI/PCI-X/PCI-e bus utilization. Here is a realistic example: each 160 ms, a block of 8 MB of data needs to be sent to a sata disk, that means on average 6.25 blocks are sent per second (1000 / 160).

• If a 150 MB/s sata link is used, a practical 120+ MB/s burst data transfer rate can be achieved, which means the data block would be transferred in about 67 ms (8 / 120 * 1000). Since 6.25 blocks need to be sent per second, a total of 419 ms (67 * 6.25) is spent doing the transfer from RAM, over the PCI/PCI-X/PCI-e bus, over the sata link, and to the disk. The bus utilization is about 41.9 %.
• If a 300 MB/s sata link is used, a practical 240+ MB/s burst data transfer rate can be achieved, which means the data block would be transferred in about 33 ms (8 / 240 * 1000). Since 6.25 blocks need to be sent per second, a total of 206 ms (33 * 6.25) is spent doing the transfer from RAM, over the PCI/PCI-X/PCI-e bus, over the sata link, and to the disk. The bus utilization is about 20.6 %.

In this example, using 300 MB/s sata links reduced the PCI/PCI-X/PCI-e bus utilization from 41.9 % to 20.6 %. Which is important in case other devices need to use it (e.g. network cards in a file server, etc).

Please note that in the case of the slowest bus (PCI 32-bit 33 MHz), this example doesn't make sense because a 150 MB/s sata link (120 MB/s practical data throughput) already fully saturates the bus. The example remains valid for any other case (PCI 66 MHz, or PCI-X, or PCI-e).

Re:Meant for whom?

[Sylver Dragon]

There are areas which need this. I support GIS(Geospatial Information Systems) users. It's not uncommon for one of my users to need access to a 20GB raster image. Then they load a 100GB LIDAR raster on top of that. Which sounds fine, just put it on the SAN, and we're happy. Then they will ask for access to that data while in the field, with their laptop.
Sure, a USB drive works, but when you are talking about 100's of gigs of data, and everytime they pan around the map a chunk of it is going to be pulled off the drive again, speed makes a difference.
That said, this is the first I've really read about eSATA. I somehow expect thought that I will have a few users ask about it soon.

Umm, why eSATA, just use SATA

[Anubis333]

I have a PCMCIA SATA card on my laptop and get 85MB/sec on it. The PCMCIA card is basically an SATA adapter, then i plug in an external SATA HD. It completely blows away firewire; why do we need a new standard? Check out the cards here.

Re:Is it reliable?

[Dolda2000]

Most importantly, however, during normal usage, most idle time to due hard drive latency isn't spent waiting for the hard drive to transfer the data off the platters, but waiting for the head arm to seek to the right cylinder. Seek time ain't going to be reduced just by switching to another bus interface.

Sorry to be replying to my own post, but looking at THG's IOMeter benchmarks, my theory seems to be verifiable.

• File server scenario: eSATA performance is above USB2, but at most points only very slightly.
• Web server scenario: Performance of eSATA and USB2 is virtually identical.
• Database scenario: eSATA is slight better than USB2. Interestingly, though, RAID1 is a lot better than RAID0.
• Workstation scenario: Fluctuates greatly. At some points, eSATA even performs worse than USB2.

No.

[DrYak]

To be able to do IP over a bus, this bus must accept multiple "hosts" (multiple controllers on the same bus).
Usually, if a bus can accept multiple PCs, or even no PC at all, it's a good sign. If a bus is a 1PC + Multiple Peripheral topology, you're sure i'll probably NOT support IP.

Able to do IP :

• FireWire : can do IP. A FireWire bus can support multiple hosts or no hosts at all.
• SCSI : can do IP. The bus it self is just a bunch of numbered SCSI device talking to each other. So IP is possible as long drivers for controller can do both HOST (controller talks to other targets - your normal everyday controller mode) and TARGET mode (controller can accept input from other hosts - like normal SCSI devices like CD-Roms, etc... do). There are some patch on the internet.
• Bluetooth : Are devices are equal (no host/target distinction). IP is possible as long as each device advertises some Bluetooth Network Profile.

Unable to do IP :

• usb : only a single host that manage devices connected to it. (In PC-2-PC usb connection cable, this is circumvented by having each side see a USB device advertised as a "USB Network device". But the two busses aren't interconnected - Also some device like PDA and Photo camera can switch between being an USB peripheral, when connected to a PC. or a USB host, when connected to a printer) That's also why I don't like it as much as FireWire.
• SATA, ATA, etc. : 1 host controlling 1 or more peripherals. Can't do IP per se. Could be circumvented like USB (although nobody is doing "ATA Network devices" anymore sinces the old days of 16bit PCMCIA and CompactFlash network adapters)
• MIDI, i2c, etc. : can have multiple device talking to each other (see MIDIMaze), but bus speed makes them unusable