5 Atypical reflections on cloud computing

The original Intel platform Test private Cloud architect, now an IT company Cloud product leader, Sina Weibo, "cloud computing Private Cloud research" blogger Eric Li from his own practical experience, summed up the following about cloud computing 5 of atypical thinking.

1. Do not put eggs in one basket: SmugMug was the "Survivor" of Amazon's cloud failure last month, and it succeeded in averting disaster, first because it did not put all the business in a availability Zone (AZ), The second is that it does not put all the business on the public Cloud (reference:http://don.blogs.smugmug.com/2011/04/24/how-smugmug-survived-the-amazonpocalypse/). If you choose the public cloud platform, remember that its fault day is also your day of the suffering.

2. Cloud computing companies will also die: Banks will run, planes often crash, and cloud-computing vendors offering centralized services have the potential to exit, collapse, or die out, and last week slicehost, the IaaS service provider, withdrew from the cloud-computing market as an example (http:// Searchcloudcomputing.techtarget.com/news/2240035506/what-happens-if-your-cloud-goes-away), This allows customers who put the company's business on the table to be caught Off-guard, so arranging a retreat, or choosing a mix of reasonable mixes, may be a wise choice before cloud computing is fully mature.

3. Early worms are eaten by birds: in September 2008, an American Starup company called Bear (Bluebear) entered the realm of virtualization, offering a very dazzling management console solution across virtualized platforms that support VMware ESX, Citrix XenServer and Microsoft Hyper-V. It is based on the flex technology to the enterprise environment, the virtual machine, network topology, resource utilization and operation buttons are very friendly to the user. From today's point of view, this is a cloud computing company, but its gorgeous Kodiak even have no 1.0 version, the last version of the company is Kodiak 0.2, because of the lack of follow-up funds, in the second quarter of 2010 it did not think that the successor, The last news about Kodiak is about handing his source code to open source community management: http://virtualization.info/en/news/2010/05/with-bluebear-out-of- Business-kodiak-is-now-open-source.html (you can see it in the following address git information: http://bluebear.org/projects/kodiak/repository), Now countless new companies are doing the same thing under the banner of cloud computing.

Another example is Elastra, the cloud-computing pioneer, which once gained Amazon's favor and investment, and ended up closing the business last week. And eucalyptus, the pioneer of cloud computing, was first abandoned in 2010 by NASA as "not really open source," and watched as it defected to OpenStack. Then again the tragedy repeats, by Ubuntu to nearly "finish throwing" way to betray (Eucalyptus originally as UEC Ubuntu Enterprise Cloud core components), was OpenStack robbed business, had to let people sigh boxing fear, yellow yesterday. Cloud computing is an emerging sector, under the mask of bubbles, the technology of rapid innovation and the pursuit of a better cloud computing experience, but also to more and more pioneering companies put pressure, inevitably become a former wave dead on the beach.

4. China's cloud computing ≠ the world's cloud computing: maybe you see Salesforce in a raging foreign country, maybe you see googleapp all over the world, perhaps you have seen Amazon popular, perhaps you see Office365 has been deployed to the United States federal government ... Maybe a lot of ... But all of this you will find in China is very different, China's cloud computing is a "feature", the reason, the complex, there are domestic customers on cloud computing attitude is heavier, there are top international cloud manufacturers to the domestic market, there is not enough attention to the expensive cloud investment fear, There is also the country's well-known second wall of the powerful blockade ability ... All this, decided in China's cloud computing field, the owner of a different landscape.

5.HPC High Performance computing ≠ cloud computing: HPC (High configured Computing) High-performance computing is the superposition of operational performance, while cloud computing more emphasis on the rational allocation of resources and more convenient delivery model. So the idea of confusing HPC with cloud computing is wrong, or that HPC is the main development model of cloud applications may be wrong, look at Amazon's architecture, in addition to a distributed shadow, there is not much parallel computing design, the nodes do not need complete message communication, no MPI, Full of all kinds of OS for general-purpose programs rather than proprietary computing, and virtualization is at its core, with no slightest intention of moving toward HPC evolution. Of course, it's not that HPC cannot be combined with cloud computing, and in some applications there are examples of rapid deployment and flexible scaling of HPC through virtualization, but at the expense of HPC performance, which degrades performance (configured degradation). So the proprietary characteristics of HPC high-performance Computing and the unique architectural design approach determine that it and cloud computing may be the occasional intersection but "different way to return" two areas.

(Responsible editor: admin)