|
|||||||||||||||||||
|
|
|
|||||||||||||||||
|
High-performance computing is not restricted to supercomputers put up by research labs and universities. Imagine a large organization with more than hundred thousand employees with offices spread all over the world. The IT infrastructure required at the data center to handle just its back-office operations would be mind boggling. That's where high-end servers come to the rescue. While a high-end server from a vendor would use similar components as the vendor's entry-level counterparts, it's the underlying architecture that would make the difference. For instance, they could all use the same type of Ultra360 SCSI hard drives and RAM, the architecture that allows these components to communicate with each other and the processors would be totally different.
All high-end servers have to be highly customizable and scalable in terms of the number of processors and RAM that they can handle. They should be able to scale up without resulting in any performance loss. Moreover, as they would all be used in mission-critical environments, you can't bring them down for small changes such as adding, removing or reconfiguring any of the components. All changes have to be done on live servers. Due to this, the architecture has to be able to allow this without affecting any of the applications that are running on the server.
All the raw processing power aside, how do large organizations use these servers? What applications do they run? The answer comes in the form of compute- intensive applications such as business intelligence, data warehousing, ERP, e-commerce, telecom, banking and
OLTP.
In the pages to follow, we've covered some of the big ones from prominent players in the market-how their architectures work, the benefits they offer and what use they are put to. Most players have their own distinctive architectures for their offerings, which provide certain types of benefits.
By Anil Chopra, Ankit Kawatra, Geetaj Channana and
Sushil Oswal