01: A brief overview or rather a cheatsheet of MongoDB’s index types and commands. ★

02: I didn’t know the replicating data form Couchbase Lite to Couchbase requires an extra tool, the Sync Gateway. ★

03: A very nice read about how to transform some of the most populat sequential clustering algorithms, k-means, single-linkage, correlation, scale for large amounts of data using a map-reduce massively parallel computation model. ★

04: An intro to using Spark Streaming with some HBase and data visualization. ★

05: Benchmarking Amazon EBS options, spinning vs SSD vs Provisioned IOPS SSD, using Redis. No surprises here. ★

06: Researchers from MIT and the Israel Institute of Technology have proved that for a large-class of non-blocking parallel algorithms, lock-free vs wait-free perform are equal. ★

Lock-free algorithms guarantee that some concurrent operation will make progress. Wait-free algorithms guarantee that all threads make progress.

07: Facebook organized a summit to discuss their storage engines and then look at the challenges they are facing across small & big data, but also hardware. ★

Facebook’s storage is based on: Tao and Memcached. Tao operates at a rate of billions of queries per second. The Memcached caching layer has a critical impact on the service availability.

The problems Facebook would like to address at both small data and big data layers are quite challenging. A couple of examples:

1. how to deal with geographically distributed caches
2. how to deal with huge amounts of logging which is quite difficult to store in their entirety for analysis
3. Facebook’s data warehouse must be partitioned globally and this has important implications on the type of queries that can be executed

Original title and link: NoSQL databases, Hadoop, Big Data: Pinned tabs Nov.5th (NoSQL database©myNoSQL)