2003: The year that was

Continued from page: 1

Tuesday, December 02, 2003

Wannabe Stuff of the year

A technology doesn’t always become popular over night. It depends upon factors such as how hot are the prevalent technologies, how much does it cost to implement the new one and vendor acceptance. Several technologies that were introduced either this year or in the past two to three years, fall in this category. They didn’t create much impact in the user space this year, but have strong future potential. 

Gigabit Ethernet
Gbe (Gigabit Ethernet) has been talked about for years now but, unfortunately, it hasn’t managed to conquer the networking space. This year saw some developments that could help the technology grow in the near future. For one, we finally saw product launches in the market this year at far more affordable prices than earlier. Second, we saw it slowly beginning to enter the desktop space with many high-end motherboards coming with integrated Gbe ports. Plus, most servers are equipped with Gbe ports. 

While these will help the technology grow to some extent, the major hurdle lies elsewhere—the structured cabling space. Gbe won’t become widely accepted unless the existing cabling infrastructure can support it. The laying of cables has to be done much more carefully than for the existing 10/100 Mbps networks. That’s because at such high speeds, mis-synchronization can occur very easily, which can actually reduce the performance by as much as 50%, meaning 500 Mb/s.

So, the technology will start with being implemented at the network backbone level and then move from there. The ratification of the Cat 6 standard for Gbe last year will help it to some extent, but the real growth will occur as the cost of equipment comes down. Plus, further developments to this technology like the 10 Gbe are also happening. Who knows, it might even become a better alternative to fiber. 

Serial ATA
SATA is touted to be the successor to the existing PATA (Parallel ATA) interface hard drives. By the end of 2002, people started hearing about SATA, with drives scheduled to arrive by mid-2003. Thankfully, the drives did arrive as per schedule, promising a lot to the user. They sported thinner, more manageable cables, promised faster data rates and better reliability.

They were targeted at the power user in the desktop space and for low-end non mission-critical servers in the enterprise space. So far they’ve received full support from hardware manufacturers with most new motherboards shipping with SATA controllers. 

PATA drives seem to have reached their end of life as far as performance goes, but SATA has a roadmap that will increase their performance many fold. In fact, the next version, namely SATA II, is slated for release next year, and promises transfer rates of up to 300 MBps as opposed to the existing 150 MBps. If that happens, SATA may give some serious competition to SCSI.

With so many benefits and so much support, why didn’t they proliferate this year? Two reasons: cost and performance.

Currently, SATA drives are more expensive than their PATA counterparts. Plus, they don’t offer significant performance gain. 

Desktop RAID
RAID has always been in the r\ealm of servers because it helps achieve a mix of redundancy and performance. This year, RAID became a reality at the desktop using ordinary IDE drives. While IDE-based RAID was possible earlier as well with a number of vendors offering IDE-based RAID cards, this year saw it being integrated with motherboards, such as the Intel 875PBZ. It started with motherboards that had Serial ATA support, which could also be configured to work in RAID level 0 or 1. Later it also moved into the Parallel ATA domain. 

RAID 0 improves performance through data striping and RAID 1 achieves redundancy. In both cases of RAID, you need two hard drives. RAID 0 achieves far better throughputs than when using a single hard drive. 

Till now the hard drive was the biggest bottleneck to building a workstation-class machine. You could either go for a single 7200 rpm IDE hard drive or buy the more expensive SCSI drives. RAID finds an in between path that is not too expensive and yet gives great performance. It can benefit in all throughput intensive applications, be they audio ripping and MP3 encoding, video capture and editing, DVD encoding, running CAD/CAM applications or when using programs like Photoshop and Premiere.

Since the technology entered the desktop market this year, it would take some time to become popular. Hopefully, it should pick up next year. 

Grid computing
This is an offshoot of distributed computing wherein a compute intensive workload is distributed across multiple systems for a speedy delivery of results. Distributed computing has been used in high-end computing environments, such as engineering and bio-medical research. Grid computing brings this concept to the enterprise. Simply put, it harnesses the unused potential of all systems in your organization. Typically, a large part of an organization’s PCs would be used for running productivity apps such as word processing, browsing and e-mail. These apps leave a lot of the PC’s power unutilized. With grid computing, organizations can use this spare power to perform complex tasks. Today, big names such as IBM, HP and Sun have announced grid-computing initiatives. There’s the Global Grid Forum, which has developed a standard for grid computing called the Open Grid Services Architecture. So, though there was a lot of noise around grid computing this year, it was mostly in the vendor space. Not too much happened in the user space, except for niche areas such as life sciences, geo sciences and defense research. The good news, however, is that with all this development, we’re likely to see a lot of solutions being built around grid computing.

Anil Chopra, Anindya Roy, Anoop Mangla, Geetaj Chanana, Sanjay Majumder

Page(s)   1  2