火烧云

现在如果说没有听说过云计算、云服务，那简直就是无知的代名词了。但凡是在线服务从PAAS、IAAS到SAAS，都必须把自己贴上云计算的标签才能卖个好价钱，包括云杀毒、云存储、云DC，以后说不定还会有云游戏、云视频、云IM、云会议中心、云客服；还有什么企业云、私有云、有颜色的蓝云、说不定还有乌云、白云、红云、卷积云、火烧云！

这份狂热的劲头不禁让人想起来这些年的流行玩意，从早些时候的RDBMS、4GL、CASE、OODB、MIS、XML、DCOM、Linux、Corba、UML2.0、J2EE、Struts、Spring、SIP、ERP、CRM、Convergence、SOAP、IPv6、Streaming、Jabber、.NET、GPS、GIS、LBS、P2P、Chord到RoR、RSS、Eclipse、Blog、Search、Web2.0、Atom、AJAX、RIA、WebService、iREST、Phone、MID、Android，每一次都掀起了轩然大波，搅得人心惶惶，年老的不知道自己是不是死期将至，年轻的不知道自己方向何在。当大多数下定决心洗心革面重新做人的时候，却不料风云再起，又出现了新玩意儿，看你怎么办！

实则多数新玩意儿都是新瓶装旧酒，汤变药不变。就拿云计算来说，早先的MPI、OpenMP、Globus难道就完全不好使了，MySQL、PostgreSQL、Oracle不搞云计算就要下课？Memcached就失效了？多机多进程+多线程就不如MapReduce？其实核心都是一样的，无非是要解决高性能、高可用性、可伸缩性而已。拿BigTable来说，本质是一个Key-Value Store，和N年前开发的BDB没有什么差别；单Master结构也绝非先进；使用三级索引就是数据库partition的一种简化翻版。Share-nothing体系结构更是老早以前就被提出...

等火烧云变成大白菜还能吃，要是有天烟消云散，那就要饿肚子了！与其腾云驾雾、随波逐流，不如静下心来安心钻研技术，说不定哪天也搞出个Latey、Linuz！

外电：Cloud computing and cloud services have been celebrated as the next big thing and also slammed as the next over-hyped re-tread of an application area that already exists. Everyone from IBM to Microsoft has been viewed as a potential winner, and though telcos have been seen as well-positioned to take advantage of the trend, it has not been clear how that might happen. Now, Telephony has a report looking a little deeper at the benefits of cloud services for telcos, including increasing network traffic, and the less obvious revenue boon that could witness telcos gaining revenue both from end users and cloud service firms as both parties pay a fee for quality of service delivery.

1. Qcon火烧云：http://gojko.net/2009/03/11/qcon-london-2009-database-projects-to-watch-closely/
2. Cisco：http://forge.mysql.com/wiki/MySQL_Internals_Optimizer/Storage_engine_Interface
3. 下个火烧云：http://highscalability.com/are-cloud-based-memory-architectures-next-big-thing
4. 大白菜：http://forge.mysql.com/wiki/MySQL_Internals_Optimizer/Storage_engine_Interface