What Web 2.0 Means for Hardware and the Datacenter 125
Tom's Hardware has a quick look at the changes being seen in the datacenter as more and more companies embrace a Web 2.0-style approach to hardware. So far, with Google leading the way, most companies have opted for a commodity server setup. HP and IBM however are betting that an even better setup exists and are striking out to find it. "IBM's Web 2.0 approach involves turning servers sideways and water cooling the rack so you can do away with air conditioning entirely. HP offers petabytes of storage at a fraction of the usual cost. Both say that when you have applications that expect hardware to fail, it's worth choosing systems that make it easier and cheaper to deal with those failures."
Re:No link (Score:4, Informative)
Re:No link (Score:3, Informative)
Sorry, printy links require membership.
Re:RTFA... (Score:5, Informative)
Re:New Buzz Hardware... (Score:4, Informative)
The poster equates Web 2.0 to "java, shockwave, and ads" and gets modded as insightful? Riiiiight.
Even if you were only focused on the technical aspects of Web 2.0, you would realize that these so-called Web 2.0 [blogspot.com] sites used AJAX and neither java nor shockwave. An even more relevant description of web 2.0 would include such terms as collective intelligence [blogspot.com], user generated content [transitionchoices.com], or the long tail [blogspot.com].
Re:WTF ? The Web 2.0 approach to hardware? (Score:4, Informative)
Not to be pedantic, but it depends on what model you are using. According to the OSI model, the Application Layer is 6 layers above the Physical Layer. And according to the TCP/IP model, the Application Layer sits 4 layers above hardware.
Network models with thousands of layers?! Not only is that crazytalk, it's way too precise to be practical.
Re:Get on board or get left behind (Score:3, Informative)
Re:Web 2.0 (Score:5, Informative)
When drives or the fans in the disk enclosures fail, the PolyServe software tells you which one has failed and where - and gives you the part number for ordering a replacement. Add a new blade or replace one that's failed and you don't need to install software manually. When the system detects the new blade, it configures it automatically. That involves imaging it with the Linux OS, the PolyServe storage software and any apps you have chosen to run on the ExDS; booting the new blade; and adding it to the cluster. This is all done automatically. Automatically scaling the system down when you don't need as much performance as you do during heavy server-load periods or marking data that doesn't need to be accessed as often, also keeps costs down. [emphasis mine]
I know, not what you meant, but a funny coincidence.
IBM is offering a more optimized rack, with shared and optimized power supplies, different arrangement for the fans, a heat exchanger in every rack for your building's air conditioner, (which Tom's interprets as water cooling) and a couple other things.
HP has a weird clustering software/hardware hybrid with large amounts/density of RAID 6 storage (for a flickr-style site, for example) together with a cluster of blades that can all access all the storage and can be added/removed at will. Interestingly they point at scaling down the system when load is low, to keep the costs down. I wonder if they put servers on stand-by automatically or something. They are also looking at not spinning all the disks all the time, but they're not there yet. I guess having some disks acting as a write cache could allow you to at least spin down the parity disks of the LRU sections or some such. You could even cache the read side if you're willing to put up with the spinup delay on a cache miss.
Supposedly this is Web 2.0 because you want a google-style cluster with lots of generic hardware where any one computer can go down and the whole thing keeps going. IBM wants to lower the maintenance costs, HP didn't show them the server side, but pushed their storage technology.
On the sideways thing.. (Score:5, Informative)
As to the air cooling aspect, I think the discussion is tilted toward the extremes of bad datacenter design to sound better, but water-cooling is more efficient to pump the distance even with clear path for the air to go. Not saying this is specific to any particular vendor (the difficulty of sticking the converse of a radiator on the back of a rack seems like it would be low), but I think IBM is fishing for ways to take advantage of two-column racks in a remotely meaningful way. In this case, the ratio of usable surface area on the water pipes to unusable plumbing in the design is higher since they can be wider.
Unfortunately there's one single definition (Score:5, Informative)
Now I'm not usually one to make a big fuss over using a word wrong, but this one is actually a trademark. Deciding to use it in any other way, is a bit like deciding to call my Audigy 4 sound card a GeForce or an Audi. It just isn't one.
And the extent to which both tech "pundits" and PHBs use it wrong, while (at least the latter) proclaiming their undying love and commitment to it, just leaves the impression that they use it as yet another buzzword. You don't proclaim your commitment to a technology, unless you actually understand what it is, how it can help you, and preferably how it compares to other technologies to the same end. Just going with a buzzword because it's popular, and ending up pledging your company to the camp of such a buzzword, is as silly (and often has the same effects) as making it your strategy to use scramjets in bicycles. Just because everyone seems to love scramjets lately, and you wouldn't want your mountain bike company to be left behind.
To get back to the actual definition of that trademark, it's not even about technology as such. It's about people. It's not techno-fetishism, as in liking cool new technologies for their sake, it's techno-utopianism: the mis-guided belief that you only need to give more internet tools to a billion monkeys, to get a utopia like nothing imagined before. Although said monkeys never created anything worth reading with a keyboard, if it's keyboards connected to the Internet, now that's how you hit a gold mine.
O'Reilly's idea is sorta along the lines of:
- forget about publishing content (e.g., hiring expensive tech writers and marketers for your site), it's all about participation, baby. Let users write your content. Hust put in some wikis and forums, and a thousand bored monkeys will do the work faster, cheaper and more accurate. (People will just flock to offer you some free, quality work, just because they like donating to a corporation, I guess. And if instead you discover comments about how much your company sucks, the CEO's sexual orientation, and his mom's weight, well, I guess it must be true, 'cause collaborative efforts can't _possibly_ be wrong.)
- forget about setting up your own redundant servers or dealing with Akamai, use BitTorrent. (Ask a lot of people how they felt about Blizzard's going almost exclusively through BitTorrent at launch. Nowadays their own servers serve a lot more of the content, if not enough other users are stuffing your pipe. I wonder why.)
- forget selling media on the Internet, teh future is Napster letting people pirate it, like happened way back then. (No, literally, the "mp3.com --> Napster" line is part of his own page explaining Web 2.0. I guess good thing noone told Steve Jobs that.)
- forget content management systems, use wikis. (I wonder in which alternate reality the piss-poor search engines of wikis can be compared to the capabilities of those systems.)
- for that matter, forget about structuring information in any way, like through directories and portals, just let the users tag it. (I'm _sure_ that the tags "humor, theft, oldnews, !news, digg" will so help me find the story about a manager stealing the server from earlier. Never mind that search engines were already dumping searching for tags, in favour of full text search, even at the time when he came up with that idea.)
Etc.
Basically, if you have the patience to sift through his ramblings, and don't give up at the "well, Google started up as a web database" intro, the meat begins at "Harnessing Collective Intelligence". That's what's it about. It's not as much about what technology you use on the web, it's about connecting a billion clueless monkeys, and believing that the result is something a billion times more intelligent and informed. Anything that helps connect those monkeys is good, anything else is irrelevant. Even whether you us
Re:WTF ? The Web 2.0 approach to hardware? (Score:3, Informative)