A lire sur: http://www.cio.com/article/751401/The_Mainframe_Isn_t_Dead_and_Neither_Is_the_PC
The mainframe was supposed to go extinct decades ago, but it's abundant in many habitats. Same goes for the PC, which seems to have adapted for survival better than once thought. Both the mainframe and the PC offer evolutionary advantages that newer, more sophisticated species still struggle to match.
Rob Enderle , Fri, April 11, 2014
CIO — One lesson we should take away from the "death of the mainframe" talk is that we often get so excited about what's new and shiny that we forget the ecosystem and experience that surrounds what that new shiny thing is supposed to replace.
We forget that, regardless of our excitement, we really, really don't like to change. We tend to forget that "better" is relative. Finally, we have a nasty habit of getting so excited about a new technology that we ignore the fact that it fundamentally doesn't work for us.
Sun Microsystems, for example, led the vanguard of the firms declaring, "The mainframe is dead." Well, Sun has been gone for about a decade, while the mainframe remains IBM's most successful hardware platform from the standpoint of revenue and profit.
Similar to this Article
Let's talk about the never-ending death of the mainframe. As long as it evolves, it's likely immortal — and this has implications for the impending death of the PC.
Mainframes' Massive I/O Enable Survival of the Fittest
The mainframe remains relevant for two reasons: There's a ton of software that didn't migrate well to any other architecture, and there's the mainframe's massive input/output.
It isn't that other architectures couldn't match a mainframe's IO — I recall a successful effort by Microsoft to replace mainframes with then-Windows NT servers in the late 1990s. However, the CIO at the time candidly said that, had he paid for the solution, he could've bought three mainframes for the same cost.
[ The Mainframe Turns 50: IBM System/360 Launch Was Dawn of Enterprise IT ]
[ More: Mainframer Remembers 'A Fantastic Time to Be an Engineer' ]
[ More: Mainframer Remembers 'A Fantastic Time to Be an Engineer' ]
This was often the problem (or the benefit) of the mainframe: At a given high I/O load, the mainframe was the most cost effective. This made them ideal for database implementations. But the mainframe did need to evolve — and back in the 1980s, when it was declared dead, it very nearly was because it hadn't evolved.
This has changed in the following decades. Today's mainframe runs current-generation platforms, uses cutting-edge hardware architecture and includes power and cooling solutions that are often a generation (or more) ahead of the high-volume market because it's a relatively low-volume product and can more readily accept those enhancements.
Now, the mainframe isn't for everything; newer architectures perform better for tasks that lend themselves to a ground-up solution, such as hosting and rendering. That said,firms such as BMC provide compelling mainframe tools. While the mainframe clearly isn't the center of the digital universe any more, the platform still carries some of the most critical data center tasks of the enterprise.
Aucun commentaire:
Enregistrer un commentaire