- From: Melvin Carvalho <melvincarvalho@gmail.com>
- Date: Thu, 13 Feb 2014 01:12:27 +0100
- To: cr <_@whats-your.name>
- Cc: public-webize@w3.org
- Message-ID: <CAKaEYhJiZmfCL5rO8d+hFsNMNkAyECs4gz5eLhpMdqNyUZWNCA@mail.gmail.com>
On 12 February 2014 19:09, cr <_@whats-your.name> wrote: > Melvin, i am wondering if you've defined "Web Scale" somewhere. > > see that term crop up regularly in marketing-materials in > big-data/analytics land > > tends to mean "can you store/query 100 billion rows/records? > "can 50% of reddit's userbase login to your server-farm > without falling over? > > "is your mainframe big enough" for your walled-garden to fit a sizable > chunk of web-users.. > > this is really different from how i've seen you use the term web-scale, > more relating to localized regions of the web fan out to the rest in a > cohesive way? > I dont have a specific definition of "web scale" other than the slightly boring "scales like the web". I think the ideal is that when effort is put into interoperability the eco system can be useful to a small number of participants and also a large number, without changing the fundamental architecture too much. > > http://www.w3.org/DesignIssues/Fractal.html is what you're talking about, > i think > > Thanks for the pointer. This is a great article, perhaps we should term "scale free" more often too. I dont know about zipf's law, but I do see the web operating in clusters that can be be potentially ljoined together. Webizing is one way to do this.
Received on Thursday, 13 February 2014 00:12:56 UTC