Author: JT Smith
Category:
- Linux
Author: JT Smith
Category:
Author: JT Smith
Category:
Author: JT Smith
Author: JT Smith
As you probably know, Linux runs pretty well on low-end hardware. You can get it running pretty well on a 500- to 1000-megabyte hard drive. On the other end of the spectrum, you can run it on enormous drives with gigabytes upon gigabytes of storage running at 10,000 RPM. In this article I review the Quantum Atlas 10K II Ultra160 SCSI hard drive, one of Quantum’s latest entries into the SCSI server/workstation hard drive market.
The drive
The 73.4 gigabyte model of the Atlas 10K II from Quantum stores quite a large amount of information. I came late into the game with computers, and my first machine only had a 40 megabyte Seagate hard drive. Move ahead to the present, and we have hard drive manufacturers playing leapfrog more than AMD and Intel do. Recently many new drives have been announced, including the enormous (both in size and (presumably) cost) Seagate Barracuda 180 gigabyte model. Still, that particular drive has only been announced, and drive in the area of 73 gigabytes are at the top of the line now.
When considering a drive, there are many factors, including drive speed and effeciency, and the physical dimensions of the drive, the temperature at which it runs, and the amount of noise it produces. Now, if you are looking for a quiet drive that runs cool and fits in a low profile case, you probably don’t want this drive. First, it’s a 10,000 RPM hard drive, which means that besides being horrifyingly fast, it is also quite loud when being used heavily. The noise isn’t too bad, and is something that can be drowned out fairly easily by music, a TV, or anything else that makes a more pleasant sound than the high-pitched chattering of a hard drive.
As for the temperature, the drive does run hot to the touch, but with proper ventilation the heat should not be an issue, although this is not a drive you want to stack tightly together into drive bays.
As for the physical dimensions of the drive, the drive is a 1.64-inch high half-height drive, as opposed to the usual 1-inch tall drives. This means that it will take the space of two normal drives. However, with a drive of this size, increased size is to be expected, to account for the size of the 10 7.3-gigabyte platters that make up the 73.4 gigabytes. This might make it difficult to fit into smaller cases, but I wouldn’t recommend using a drive like this in a smaller case than a full tower, because midtowers and smaller might not provide enough ventilation for the drive.
Specifications
Atlas 10K II 73.4 GB | |
Size | 73.4 Gigabytes |
Rotation Speed | 10,000 RPMS |
Avg Seek Time | 5.2 ms |
Internal Data Rate | 280 to 478 Mb/sec |
Sustained Throughput | 24 to 40 (MB/sec) |
Interface | Ultra160 160MB/sec |
Onboard Buffer | 8 megabytes |
Installation
As with most hard drives, installation was fairly simple. Most SCSI hard drives come set for SCSI ID 0, and this one was no different. However, because it was the only SCSI drive in the system, no change was needed. Had it not been the only SCSI hard drive, the change would only need to have been made with a jumper, as with most hard drives, SCSI or IDE. Mounting the drive was simple, and it was mounted away from any other drives to allow maximum air flow around the drive.
Performance
In order to test the performance of this drive, I used two Linux-based benchmarks. The first, Bonnie++ 1.00f, which does quite a few operations on the drive. To save space and time, I will simply provide a link to the Bonnie++ readme.html. Bonnie++ does several I/O tests in order to determine the performance of hard drives. For my other, and less extensive test, I used the utility hdparm, most likely included with your Linux distribution, to test the cached and uncached (but still buffered by the drives onboard buffer) throughput of the drive. In order to test a device with the software disk cache enabled, use the command ‘hdparm -t devicename.’ To test the uncached (but buffered) throughput of the drive, use hdparm -T devicename. Where bonnie++ simulates several real-time situations, hdparm simply attempts to figure out the drive’s maximum throughput.
System specifications
AMD Athlon 750
128 Megabytes PC133 SDRAM
FIC SD-11 AMD Irongate-based Mainboard
3COM 3C905TX-B 10/100MB NIC
3DFX Voodoo 3 3000 AGP 16MB Video Board
Adaptec 29160 32-BIT PCI SCSI Controller
Western Digital 10gb 7200 RPM WD102BA IDE HDD (boot/OS) (ReiserFS)
Linux Mandrake 7.2 with Kernel 2.2.17
Quantum Atlas 10K II Ultra160 connected to Ultra160 chain (ReiserFS)
Results |
|
WD102BA IDE | 145.45 MB/sec |
Atlas 10K II | 148.84 MB/sec |
Results |
|
WD102BA IDE | 4.44 MB/sec |
Atlas 10K II | 40.00 MB/sec |
While the cached speed of the drives is the same, which is to be expected since it really doesn’t depend all that much on the drive, we see a real difference in the raw throughput of the Atlas 10K II. It can output data at a maximum 40 MB/second uncached, which is pretty amazing. This is due largely to its 8-megabyte buffer and its low seek time.
Results |
|||
Output |
|||
Drive | Per-Character | Block | Rewrite |
Atlas 10K II | 6333 K/sec, 96% CPU | 36865 K/sec, 59% CPU |
13065 K/sec, 20% CPU |
WD102BA IDE | 3172 K/sec, 49% CPU | 5774 K/sec, 20% CPU |
2401 K/sec, 22% CPU |
Input |
|||
Drive | Per-Character | Block | Random |
Atlas 10K II | 6441 K/sec, 92% CPU | 35546 K/sec, 21% CPU |
167.4 Seeks/sec, 1% CPU |
WD102BA IDE | 2876 K/sec, 60% CPU | 3783 K/sec, 32% CPU |
76.8 Seeks/sec, 6% CPU |
Create |
|||
Drive | Create | Read | Delete |
Atlas 10K II | 7410 /sec, 96% CPU | none | 8028 /sec, 98% CPU |
WD102BA IDE | 8806 /sec, 93% CPU | none | 12301 /sec, 96% CPU |
Create |
|||
Drive | Create | Read | Delete |
Atlas 10K II | 6637 /sec, 99% CPU | none | 8094 /sec, 99% CPU |
WD102BA IDE | 9056 /sec, 100% CPU | none | 11182 /sec, 98% CPU |
Bonnie++ is an interesting benchmark in that it reports the CPU utilization during the procedures. Now, mind you, these don’t strictly relate to disk usage, as some of the CPU is used by whatever pattern Bonnie++ is using to read/write. This makes Bonnie++ good to use to test where your bottleneck might be on something like database access, which is what Bonnie++ is targeted toward. In this case, we find that in many, but not all, operations the Atlas 10K II manages to outperform our baseline Western Digital IDE drive. However, somewhat surprisingly, the WD102BA does take the Atlas in the tests related to file creation. Why this would happen I’m not completely sure, but it’s interesting nonetheless. It is, however, much faster in input and output than the baseline hard drive used for these tests.
About Quantum
Quantum.com
Founded in 1980, Quantum is a manufacturer of storage devices for personal computers and networks, from IDE and SCSI hard drives to SNAP plug-in network file servers. Quantum is a public company listed under the symbols DSS (DLT and Storage Systems) and HDD (Hard Disk Drives).
Conclusions
The Atlas 10K II is an excellent drive, from a reputable manufacturer. It has all the features needed for those interested in creating a higher-end server or workstation. Speed, size and a reputable manufacturer. However, many of the people I know who work with high-end systems would rather have a few smaller drives than one large drive. This is for various reasons, but if you are looking for one big drive, or several big drives, the Atlas 10K II is definately a drive to consider. The only real downfall is that the lowest price I could find it for was $1,009 on Pricewatch. However, when you are in the market for high-end drives, that’s the sort of price tag you’ll get.
Category:
Author: JT Smith
On Wednesday, the U.S. Copyright Office will hold its final mandatory hearings in Washington, D.C. with
representatives from the library associations and webcasting, content and digital music industries.”
Wired.com has the preview. Even more from Reuters.
Author: JT Smith
The two companies will collaborate to make the content of SBI’s
extensive
databases of protein structural information more readily accessible to
researchers worldwide via the Internet on a subscription basis. Joint
marketing efforts are also planned, and SBI will be able to draw on
IBM’s
extensive research expertise in computational biology.
“The average cost of developing a new drug today is $500 million, and
the
development cycle can take up to 15 years,” said Dr. Caroline Kovac,
vice
president of IBM Life Sciences. “SBI’s high-quality protein structures
and
computational analyses can significantly speed up the cycle and help
get
new treatments for many of today’s diseases to patients faster.”
Protein Structures Hold Key to Curing Diseases
The successful mapping of the human genome has accelerated the study of
proteins and protein-protein interactions, or proteomics, by enabling
rapid
conversion of gene sequence data into protein structures. The more than
one
million proteins in the human body regulate the structure and function
of
cells, tissues and organs. They also play a key role in triggering and
de-activating genetically-linked diseases such as cancer, AIDS,
Alzheimers,
and diabetes.
SBI, which maintains some of the world’s largest proprietary databases
of
atomic-resolution 3-D protein structures, provides proteomic
information to
pharmaceutical researchers for use in designing molecules that interact
with proteins associated with diseases. Access to this information
facilitates the development of powerful drugs that bind to proteins,
either
activating or blocking their activities.
SBI’s products are among the first “in silico” approaches to drug
discovery, allowing researchers to analyze protein models and predict
interactions between proteins and chemicals using information
technology,
rather than performing time-consuming and costly laboratory
experiments.
IBM’s DB2(1) Universal Database will become SBI’s strategic development
platform. IBM WebSphere(1) will provide the Internet software
infrastructure for accessing protein structures on the Web.
Additionally,
SBI will begin marketing its Variome(3) structural variant database
modules
to pharmaceutical companies in early 2001 with DB2 as the preferred
database.
“We selected IBM for our IT needs because of its complementary
technologies and expertise in managing volume and complexity – both
characteristics of proteomic data,” said Dr. Edward T. Maggio,
chairman,
president and chief executive officer of SBI. “Our partnership with
IBM
will allow us to expand the content of our protein databases, while
providing our customers with more comprehensive tools for analysis of
proteins.”
The minority equity investment in SBI is the latest initiative by IBM’s
Life Sciences business unit, which was formed in August to form
partnerships and develop IT solutions for biotechnology, genomic,
pharmaceutical, e-health, and other life sciences industries.
About Structural Bioinformatics, Inc.
Structural Bioinformatics is a world leader in computational proteomics
—
the large-scale generation and use of protein structure and protein
structural information. The company has developed advanced technologies
to
generate highly refined three-dimensional structural models of proteins
from primary genetic information and commercializes these technologies
through its structural database products and through drug discovery
collaborations with leading pharmaceutical companies. SBI has offices
and
research facilities in San Diego, CA; Cambridge, MA; and Horsholm,
Denmark.
More information about SBI can be found through the website,
www.strubix.com.
About IBM Life Sciences
IBM’s Life Sciences business unit brings together the company’s
high-performance computing, e-business and data and storage management
resources, and research expertise to offer new solutions for the
fast-growing life sciences market. IBM is the world’s largest
information
technology company, with 80 years of leadership in helping businesses
innovate. The fastest way to get more information about IBM Life
Sciences
is through the website,
www-4.ibm.com/software/webservers/lifesciences/.
(1) Indicates trademark or registered trademark of IBM Corporation. (2)
Indicates trademark or registered trademark of Linus Torvalds. (3)
Indicates trademark or registered trademark of respective companies.
Author: JT Smith
Category:
Author: JT Smith
Mikotel manufactures network appliances based on the Linux open source,
open architecture operating system.
Author: JT Smith
Author: JT Smith
Which caused me to wonder if maybe I’m turning more profit from Linux than is, say, Red Hat Software,
which, when it posts earnings next month, is expected to lose 2 cents per share, according to the
analysts …
I ended up with no evidence that any of the Linux companies that are publicly traded is turning a
nickel’s profit. The non-public companies I’ve talked with are facing hard times as well. One or two are
just about out of money, with not much coming in. There may be some others — companies that resell
books and T-shirts and distributions at discount — who are doing okay.” (VA Linux, mentioned in the article, owns NewsForge.)