Ximeta NetDisk – a new interface for external hard disk

The growing competition, as well as the growing popularity of hard disks attracting more and more companies to this market sector, compels manufacturers to look for new ways of attracting more buyers. One of such ways is equipping external hard disks with new interfaces.

While traditionally external hard disks are equipped with the USB or FireWire interface or both simultaneously, these days there have come up attempts to equip hard disks with network interfaces, including wireless. One of the manufacturers who followed this path became Ximeta established in 2002.

In this article, we'll get acquainted with one of such devices produced by this company, namely, the external hard disk Ximeta NetDisk (NDU10-120). The hard disk capacity is 120 GB (also produced are hard disks of 80, 160, and 250 GB capacity).

The device is manufactured in a combined housing (with the top and bottom made of metal, the sides - of plastic), 215x122x34 mm in dimensions, quite small for devices using hard disks of 3.5" format. On the top panel, there are two indicators: one for power-on, and one for HDD activity monitoring.

On the reverse wall, there is a main electrical breaker, a connector for plugging in the external PSU, a USB 2.0 connector, and a RJ-45 network interface.

The package bundle includes an external PSU, a USB2.0 cable, an Ethernet cable (straight, CAT5), brief instructions software use and installation, and a CD with the software and user's manual. The specimen we had was equipped with a power cable for American plug socket, so we had to look for the right adapter to plug in (not in the bundle).

The device is not to be dismantled by the user, but to get a closer idea of it we removed the upper lid.

Inside, we found a Western Digital WD1200LB hard disk. A strange situation has shaped up with disks of this series (letter L in the designation): the disks are already widely available for sale, but the manufacturer's web site is still proudly silent about their existence. According to third-party web sited, disks of this series are analogs to the WD***JB disks but with fluid bearings.

The USB2.0 interface is implemented with the USB2.0-IDE bridge Prolific PL-2507.

At the same time, we should dwell on the network interface. It uses the proprietary and patented Network Direct Attached Storage (NDAS) technology developed by Ximeta itself. The hardware implementation of the technology is effected by Ximeta NDAS2011 controller.

Using standard network cables, the NetDisk can be plugged in directly to the 10/100base-TX of a switchboard or hub. If any Ximeta software installed on a computer in the local network, that computer will automatically detect and recognize the NetDisk as its own device once it has been connected to the network. In so doing, the NetDisk does without MAC and IP-address. The NetDisk's network driver introduces it to the operating system as a SCSI hard disk. NetDisk identification on the network is done using the ID entered on the software installation.

To effect data exchange between the driver and the NetDisk via the local network, the simplified LPX (Lean Packet eXchange) protocol is used. LPX packages are not routed (an unauthorized, perhaps malicious, user can't access the NetDisk if the latter is connected to the local network having outlet to the Internet). In the local network, LPX packages are not detained by IP- or MAC-filters.

Ximeta's software allows merging two NetDisk's. Of two NetDisk's, one logical disk of larger size can be made up, or they can be made «parallel» the way it is done in RAID1 – in this case, each NetDisk will be a replica of the other.

Tests

The tests of the storage device can be conventionally divided into the two stages: first – determine the speed characteristics of the device upon connection through each interface of different file systems, then – find out the operational specifics for the new storage device, including the possibility of simultaneous operation from various computers.

When determining the speed characteristics, the major focus was given to the file read/write speed in real applications. For our tests, we used the following software:

• WinBench 99 v. 2.0;
• PCMark2004;
• FCTest v. 0.5.3.

The tests were done on a logical disk partitioned for the whole drive space, powered by ÎÑ Windows XP Professional SP1, on a computer of the following configuration (at the second stage, referred to as the "base computer"):

• processor - Intel Pentium 4 3.06 GHz (533 MHz)
• motherboard: MSI 875P Neo (MS-6758)
• memory: 1 GB (2õ512 MB Kingston PC2700 DDR SDRAM
• video: NVIDIA GeForce4 Ti 4800 SE
• system HDD: IC35L090AVV207-0 (80 GB)

The computer uses the most recent version of NetDisk Software - v.2.42.555.

To build a network, we used 3COM Super Stack 3 Switch 42501 (3C17302).

WinBench 99 v. 2.0

This test can be regarded as classic for testing hard disks. First, we bring in the graphs of linear read for the two interfaces:

The drive demonstrates quite a good speed, but there is some operational instability via both interfaces.

Then, the Ximeta NetDisk performance:

During operation via Ethernet, the performance practically does not depend on the file system used. The small advantage due to the use of FAT-32 was gained mainly due to the sub-test FrontPage 98. The situation is entirely different in using the USB 2.0.

The good performance in using the NTFS is accompanied by very low results in using the FAT-32.

PCMark2004

Of the comprehensive PCMark2004 benchmark, we used the part for testing hard disks:

We are bringing in the complete test results, although we are interested in only the File Copying sub-test (no one is likely to use an external hard disk for installing the operating system and software). As it should be, the operation speed is much higher via USB 2.0 than via Ethernet. Results in this test do not depend on the type of the file system used.

FCTest v. 0.5.3

The full name of the test is File Copy Test. The test was developed by specialists of F-Center company and until the recent times it could be freely downloaded from the company's web site. The test allows measuring the speed of writing, reading and copying files. When testing, the number and size of files is specified in a special "pattern". The "pattern" is created manually or automatically on the base of any folder existing on the disk. Tests of the Ximeta NetDisk storage device were done with the following "patterns":

• ISO - 3 files of 600 MB size each;
• MP3 - 500 files of 3 MB size each;
• Programs - 8504 files of 1.5 GB total size.

The "Programs" pattern was created automatically on the base of the "Program Files" folder and contains a great number of short files.

The storage device has demonstrated a good speed of operation via Ethernet which is determined by the interface speed (in theory, it is less than 12 MB/s) and practically does not depend on the file system used.

During operation via the USB2.0, the performance was strongly affected by the type of the file system used as well as the file size. The file size has the most effect on the performance when using the FAT-32 file system. Even during writing large files of the MP3 pattern, the performance drops significantly, and with the emergence of large number of small files in the Programs pattern the performance turns unacceptably low. At that, the read speed in all the patterns remains high enough.

Repeated measurement of the writing speed (in using USB2.0 - FAT32) done at a computer of different configuration (Athlon XP 2000+, nVIDIA nForce 2 SPP chipset) has brought the same results. Besides, it was noticed that deletion of small files takes as much time as writing.

Operation specifics

First, during operation via the USB2.0 the Windows XP Professional SP1 recognizes the new storage device as an additional disk once it is connected to the computer. No additional drivers are required. The operation speed is quite enough for comfortable editing files (verified with Word, Excel, Photoshop).

Using standard procedures, access to it can be granted to any network user.

To check the simultaneous operation, we plugged in the second computer (hereinafter - Client) to the switchboard. The computer was also powered by Windows XP Professional SP1 and had the following configuration:

• processor - Intel Pentium 4 3.0 GHz (800 MHz)
• motherboard: Elitegroup 865G-M
• memory: 512 MB (2õ256 MB Kingston PC3200 DDR SDRAM
• video: integrated Intel Graphics Controller
• system HDD: ST340014A (40 GB)

First, the Ximeta NetDisk was plugged into the Base computer via the USB2.0 and full access was granted to it, with no additional software installed at the Client. Let's look at the processor usage on the Base computer (according to the Windows Task Manager) when writing to the NetDisk from the Client and during simultaneous writing:

a) – writing by the Client (FAT-32)
b) – writing by the Client and the Base computer (FAT-32)
c) – writing by the Client (NTFS)
d) – writing by the Client and the Base computer (NTFS)

For writing, a folder similar to "Program Files" was used. As can be seen from the graphs, the processor usage is rather high. Let's see what will change during operation via Ethernet using the NDAS technology.

Ximeta ships two types of drivers. The driver of the first type allows reading the NetDisk from any computer where the driver is installed with the ID entered. In so doing, only one user is allowed to write anything on the disk at every moment. This right can be granted from user to user upon agreement with the current owner who may agree to deny. Besides, you have to know the password additionally to the ID.

The second-type driver allows reading from the NetDisk and writing anything to all who know the ID/password without any procedures for granting user rights.

The second-type driver of version 3.10.1216 was installed first on the Base computer, and then on the Client. During writing to the NetDisk from the Client and simultaneous writing, we could see the following processor usage at the Base computer:

a) – writing by the Client (FAT-32)
b) – writing by the Client and the Base computer (FAT-32)
c) – writing by the Client (NTFS)
d) – writing by the Client and the Base computer (NTFS)

As is seen from the graphs, once the NetDisk on the Client is accessed, the processor usage on the Base computer goes up. This suggests that writing is done via the Base computer and not directly as should be expected from the NDAS description.

Of note is the very high processor usage on the Base computer when using the FAT-32 file system.

In using the NTFS file system, the processor usage is a bit lower than that in using via the USB2.0.

Final Words

Ximeta has presented a new, interesting technology. Of course, it requires polishing, but no novelty can avoid that. The speed of operation via Ethernet is not high, but with transition to Gigabit Ethernet it should increase significantly.

In using the Ximeta NetDisk, use of the NTFS file system is recommended, since in using the FAT-32 there come up problems with writing and deletion of small files.