What are the best recommended research topics on databases according to edge technologies and recent research trends?

• Acording to the DBWorld mailing list trends the hottest topic of research today is Distributed Databases  (tera-scale and peta-scale): http://www.cs.wisc.edu/dbworld/
• Also see VLDB conference program: http://www.vldb2010.org/accept.htm and CIDR: http://www.cidrdb.org/cidr2011/p...
• Check out Google Research  papers on Distributed Systems : http://research.google.com/pubs/... and Information Retrieval : http://research.google.com/pubs/... and What are the most interesting Google Research papers?
• See Microsoft (company)'s  recent publications in the field: http://research.microsoft.com/en... and the list of major conferences: http://academic.research.microso...

• I highly recommend the research blog by Daniel Abadi for some emerging topics and trends: http://dbmsmusings.blogspot.com/ and his publications: http://cs-www.cs.yale.edu/homes/... also see  Jeff Hammerbacher's answer to What are the best, insightful blogs about data, including how businesses are using data?
• Commercial databases and data stores such as Amazon Dynamo http://en.wikipedia.org/wiki/Dyn...), Google Bigtable  (http://en.wikipedia.org/wiki/Big...) and  Google Percolator(http://research.google.com/pubs/...),  kdb+ by KX systems(http://kx.com/Products/kdb+.php), C-store by Vertica (company)  (http://en.wikipedia.org/wiki/Mic...), Times Ten(http://www.oracle.com/timesten/i...) and open source implementations such as Redis , Cassandra (database), HBase, MongoDB, Riak, MonetDB, Scalaris, H-store(http://hstore.cs.brown.edu/ ) are worth exploring in depth. Also see http://en.wikipedia.org/wiki/Dis...
• The recent Data explosion (http://en.wikipedia.org/wiki/Big...) led to a major revision of relational model applicability in certain domains where its restrictions are overly limiting (see http://en.wikipedia.org/wiki/NoSQL). For a good overview of emerging non-relational databases and key-value stores see  Varley.  No Relation: The Mixed Blessings of Non-Relational Databases:  http://ianvarley.com/UT/MR/Varle... ,  Abadi's thesis, Query Execution in Column-Oriented Database Systems: http://cs-www.cs.yale.edu/homes/... and Meijer & Bierman, A co-Relational Model of Data for Large Shared Data Banks: http://queue.acm.org/detail.cfm?...
• Graph Databases  (http://scholar.google.com/schola...) are extremely important in the modern Online Social Networks and many other domains, this is a topic of an active research (e.g. see Neo4j, HyperGraphDB, InfiniteGraph) , http://www.graph-database.org/ , http://nosql-database.org/
• See Microsoft Trinity,  a graph database over distributed memory cloud: http://research.microsoft.com/en... and Google Pregel,  a system for large-scale graph processing: http://portal.acm.org/citation.c...
• Specialized Database Systems and Data Warehousing in Bioinformatics  could be a good topic for applied research e.g see Atlas: http://www.biomedcentral.com/147... and Bowtie ecosystem: http://bowtie-bio.sourceforge.ne...
• Check out some interesting work done by Luis Gravano on structured search and Information Extraction from the "hidden-web": http://www.cs.columbia.edu/~grav...
• Since the costs of Random-Access Memory are continually decreasing the topic of Main-memory Databases  will probably gain more and more attention (see http://en.wikipedia.org/wiki/In-...). Check out Memcached, Hazelcast, Membase , MemSQL, FastDB, SciDB and RAMCloud: http://fiz.stanford.edu:8081/display/ramcloud/Home
• Druid: A Distributed, In-Memory OLAP Store: http://metamarketsgroup.com/blog...
• Google Snappy, http://code.google.com/p/snappy/,  a compression/decompression library used in BigTable.
• Heroku Doozer: http://xph.us/2011/04/13/introdu... and http://blog.golang.org/2011/04/g...
• LevelDB - a fast and lightweight key/value database library: http://code.google.com/p/leveldb/
• Yet Another SQL-to-MapReduce Translator: http://www.cse.ohio-state.edu/hp...
• The proliferation of 10 Gigabit Ethernet, Infiniband/RDMA and other High Performance Computing  technologies into mainstream may require rethinking of some basic assumptions in database design (See When will 10 gigabit Ethernet overtake 1 gigabit Ethernet in deployment? and It's time for low latency: http://www.matt-welsh.blogspot.c... )
• RethinkDB  (http://www.rethinkdb.com/blog/) are doing some interesting work on Solid-State Drives based databases, you may want to check it out
• I also think a few integrated specialized products are either missing, or over-engineered, unscalable, and/or expensive: 1) a tightly integrated db + analytics engine, e.g. for EEG or financial time series  2) a tightly integrated messaging framework + db  optimized for really fast ETL
• As a general piece of advice I'd try to avoid over-specialized topics in favor of building a database system for a certain real-world domain (e.g see How do scientists share data and code?). Also see Patterson, "How to Have a Bad Career in Academia": http://www.cs.berkeley.edu/~patt...
• Related: What is the best literature on the design of database platforms? Why?