Cloud Computing Races On, But NoSQL Finds Bumps In The Road
Someone whose career involves computing and software development would have to be lost in a cave in recent months to miss the almost religious fervor surrounding cloud computing and NoSQL databases. Cloud computing continues to gather momentum, but the NoSQL bandwagon has run into bumps in the road. Highly-publicized outages of sites using a NoSQL solution have raised the specter of stability problems. That's not atypical for new software and some big name NoSQL products are not an exception.The cloud has become like the proverbial snowball rolling downhill, aided no doubt by almost daily media coverage.
Three recent news items give us a pretty good indication of the state of cloud computing.
1. The Federal government has issued guidelines for data center consolidation that will have federal agencies moving applications to the cloud.
2. Compuware recently monitored web transactions for 3,000 small, medium and large businesses. It reported that 27% of those transactions had at least one component that involved a server hosted by Amazon EC2.
3. Amazon Web Services (AWS) S3 doubled the number of objects it stored between 2009 and 2010. The AWS Elastic Cloud Cloud (EC2) service has experienced steady growth. To better serve EC2 customers in Europe, Amazon recently purchased a site in Ireland to house a new data center. The 240,000 square foot building in Dublin was formerly a warehouse for the Tesco supermarket chain. Netcraft reports 1/3 of the web servers connected to the Internet in Ireland are operated by Amazon.
The economics of cloud computing (pay-as-you-go, expense vs. capital expenditure) are undoubtedly one reason for the migration to the cloud. Another reason is the cloud is the rage and it appeals to technology fashionistas who want to avoid any appearance of being a Luddite.
People in the early stages of their career are often the ones who are most interested in a change of the guard. Just as they want the latest smart phone, they also want the latest programming language or database. But they are not universally scornful of old school technology. For example, being a fan of the 1970 Dodge Charger, a premier muscle car, is fashionable. That's why a Charger featured in the movie "Fast and Furious" was valued at $200,000+, more than 50 times its original price. Others command 10-15 times their original price.
Unlike a vintage car, software doesn't have problems with rust, corrosion or metal fatigue, but somehow software can acquire an "old and tired" label. So cloud computing, like Internet computing and Windows computing in the '90s, rides a wave generated by eager adopters looking for the excitement of change. But when the fashion trend fades, unmatched capacity and capability will remain a core argument for cloud adoption. Like grid computing, the cloud makes it possible to do work that requires plenty of MIPS, plenty of data, or both. For example, Yahoo researchers were able to sort a petabyte of data in 16.25 hours using 3658 processors.
If your organization needs to do analytics with very large data sets (Big Data), you've undoubtedly looked at the need for a massively scalable architecture. Grids and cloud computing give you the hardware infrastructure you need, but then there's the question of software.
More companies are jumping into the Big Data arena with specialized data stores and solutions that leverage Apache Hadoop Map/Reduce. But don't assume that Big Data means SQL is no longer in the picture. Platforms such as Vertica, Sybase IQ, Oracle Exadata, Teradata and EMC Greenplum have a strong presence in this space and all of them support SQL queries over massive amounts of data. The Teradata data warehouse for eBay is one of the world's largest, measured in petabytes.
Next: NoSQL Speed Bumps